1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
|
/**
* Copyright (C) 2018-present MongoDB, Inc.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the Server Side Public License, version 1,
* as published by MongoDB, Inc.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* Server Side Public License for more details.
*
* You should have received a copy of the Server Side Public License
* along with this program. If not, see
* <http://www.mongodb.com/licensing/server-side-public-license>.
*
* As a special exception, the copyright holders give permission to link the
* code of portions of this program with the OpenSSL library under certain
* conditions as described in each individual source file and distribute
* linked combinations including the program with the OpenSSL library. You
* must comply with the Server Side Public License in all respects for
* all of the code used other than as permitted herein. If you modify file(s)
* with this exception, you may extend this exception to your version of the
* file(s), but you are not obligated to do so. If you do not wish to do so,
* delete this exception statement from your version. If you delete this
* exception statement from all source files in the program, then also delete
* it in the license file.
*/
#define MONGO_LOG_DEFAULT_COMPONENT ::mongo::logger::LogComponent::kReplication
#define LOG_FOR_ELECTION(level) \
MONGO_LOG_COMPONENT(level, ::mongo::logger::LogComponent::kReplicationElection)
#define LOG_FOR_HEARTBEATS(level) \
MONGO_LOG_COMPONENT(level, ::mongo::logger::LogComponent::kReplicationHeartbeats)
#include "mongo/platform/basic.h"
#include "mongo/db/repl/topology_coordinator.h"
#include "mongo/db/repl/topology_coordinator_gen.h"
#include <limits>
#include <string>
#include "mongo/bson/simple_bsonobj_comparator.h"
#include "mongo/db/audit.h"
#include "mongo/db/catalog/commit_quorum_options.h"
#include "mongo/db/client.h"
#include "mongo/db/mongod_options.h"
#include "mongo/db/operation_context.h"
#include "mongo/db/repl/heartbeat_response_action.h"
#include "mongo/db/repl/isself.h"
#include "mongo/db/repl/member_data.h"
#include "mongo/db/repl/rslog.h"
#include "mongo/rpc/metadata/oplog_query_metadata.h"
#include "mongo/rpc/metadata/repl_set_metadata.h"
#include "mongo/util/assert_util.h"
#include "mongo/util/fail_point.h"
#include "mongo/util/hex.h"
#include "mongo/util/log.h"
#include "mongo/util/net/socket_utils.h"
#include "mongo/util/scopeguard.h"
#include "mongo/util/str.h"
namespace mongo {
namespace repl {
MONGO_FAIL_POINT_DEFINE(forceSyncSourceCandidate);
MONGO_FAIL_POINT_DEFINE(voteNoInElection);
MONGO_FAIL_POINT_DEFINE(voteYesInDryRunButNoInRealElection);
// If this fail point is enabled, TopologyCoordinator::shouldChangeSyncSource() will ignore
// the option TopologyCoordinator::Options::maxSyncSourceLagSecs. The sync source will not be
// re-evaluated if it lags behind another node by more than 'maxSyncSourceLagSecs' seconds.
MONGO_FAIL_POINT_DEFINE(disableMaxSyncSourceLagSecs);
constexpr Milliseconds TopologyCoordinator::PingStats::UninitializedPingTime;
std::string TopologyCoordinator::roleToString(TopologyCoordinator::Role role) {
switch (role) {
case TopologyCoordinator::Role::kLeader:
return "leader";
case TopologyCoordinator::Role::kFollower:
return "follower";
case TopologyCoordinator::Role::kCandidate:
return "candidate";
}
MONGO_UNREACHABLE;
}
TopologyCoordinator::~TopologyCoordinator() {}
std::ostream& operator<<(std::ostream& os, TopologyCoordinator::Role role) {
return os << TopologyCoordinator::roleToString(role);
}
std::ostream& operator<<(std::ostream& os,
TopologyCoordinator::PrepareFreezeResponseResult result) {
switch (result) {
case TopologyCoordinator::PrepareFreezeResponseResult::kNoAction:
return os << "no action";
case TopologyCoordinator::PrepareFreezeResponseResult::kSingleNodeSelfElect:
return os << "single node self elect";
}
MONGO_UNREACHABLE;
}
namespace {
template <typename T>
int indexOfIterator(const std::vector<T>& vec, typename std::vector<T>::const_iterator& it) {
return static_cast<int>(it - vec.begin());
}
/**
* Returns true if the only up heartbeats are auth errors.
*/
bool _hasOnlyAuthErrorUpHeartbeats(const std::vector<MemberData>& hbdata, const int selfIndex) {
bool foundAuthError = false;
for (std::vector<MemberData>::const_iterator it = hbdata.begin(); it != hbdata.end(); ++it) {
if (indexOfIterator(hbdata, it) == selfIndex) {
continue;
}
if (it->up()) {
return false;
}
if (it->hasAuthIssue()) {
foundAuthError = true;
}
}
return foundAuthError;
}
void appendOpTime(BSONObjBuilder* bob, const char* elemName, const OpTime& opTime) {
opTime.append(bob, elemName);
}
void appendIP(BSONObjBuilder* bob, const char* elemName, const HostAndPort& hostAndPort) {
auto ip = hostbyname(hostAndPort.host().c_str());
if (ip == "") {
bob->appendNull("ip");
} else {
bob->append("ip", ip);
}
}
} // namespace
void TopologyCoordinator::PingStats::start(Date_t now) {
_lastHeartbeatStartDate = now;
_numFailuresSinceLastStart = 0;
_state = HeartbeatState::TRYING;
}
void TopologyCoordinator::PingStats::hit(Milliseconds millis) {
_state = HeartbeatState::SUCCEEDED;
++hitCount;
averagePingTimeMs = averagePingTimeMs == UninitializedPingTime
? millis
: Milliseconds((averagePingTimeMs * 4 + millis) / 5);
}
void TopologyCoordinator::PingStats::miss() {
++_numFailuresSinceLastStart;
// Transition to 'FAILED' state if this was our last retry.
if (_numFailuresSinceLastStart > kMaxHeartbeatRetries) {
_state = PingStats::HeartbeatState::FAILED;
}
}
TopologyCoordinator::TopologyCoordinator(Options options)
: _role(Role::kFollower),
_term(OpTime::kUninitializedTerm),
_currentPrimaryIndex(-1),
_forceSyncSourceIndex(-1),
_options(std::move(options)),
_selfIndex(-1),
_maintenanceModeCalls(0),
_followerMode(MemberState::RS_STARTUP2) {
invariant(getMemberState() == MemberState::RS_STARTUP);
// Need an entry for self in the memberHearbeatData.
_memberData.emplace_back();
_memberData.back().setIsSelf(true);
}
TopologyCoordinator::Role TopologyCoordinator::getRole() const {
return _role;
}
void TopologyCoordinator::setForceSyncSourceIndex(int index) {
invariant(_forceSyncSourceIndex < _rsConfig.getNumMembers());
_forceSyncSourceIndex = index;
}
HostAndPort TopologyCoordinator::getSyncSourceAddress() const {
return _syncSource;
}
HostAndPort TopologyCoordinator::chooseNewSyncSource(Date_t now,
const OpTime& lastOpTimeFetched,
ChainingPreference chainingPreference) {
// If we are not a member of the current replica set configuration, no sync source is valid.
if (_selfIndex == -1) {
LOG(1) << "Cannot sync from any members because we are not in the replica set config";
return HostAndPort();
}
if (auto sfp = forceSyncSourceCandidate.scoped(); MONGO_unlikely(sfp.isActive())) {
const auto& data = sfp.getData();
const auto hostAndPortElem = data["hostAndPort"];
if (!hostAndPortElem) {
severe() << "'forceSyncSoureCandidate' parameter set with invalid host and port: "
<< data;
fassertFailed(50835);
}
const auto hostAndPort = HostAndPort(hostAndPortElem.checkAndGetStringData());
const int syncSourceIndex = _rsConfig.findMemberIndexByHostAndPort(hostAndPort);
if (syncSourceIndex < 0) {
log() << "'forceSyncSourceCandidate' failed due to host and port not in "
"replica set config: "
<< hostAndPort.toString();
fassertFailed(50836);
}
if (_memberIsBlacklisted(_rsConfig.getMemberAt(syncSourceIndex), now)) {
log() << "Cannot select a sync source because forced candidate is blacklisted: "
<< hostAndPort.toString();
_syncSource = HostAndPort();
return _syncSource;
}
_syncSource = _rsConfig.getMemberAt(syncSourceIndex).getHostAndPort();
log() << "choosing sync source candidate due to 'forceSyncSourceCandidate' parameter: "
<< _syncSource;
std::string msg(str::stream() << "syncing from: " << _syncSource.toString()
<< " by 'forceSyncSourceCandidate' parameter");
setMyHeartbeatMessage(now, msg);
return _syncSource;
}
// if we have a target we've requested to sync from, use it
if (_forceSyncSourceIndex != -1) {
invariant(_forceSyncSourceIndex < _rsConfig.getNumMembers());
_syncSource = _rsConfig.getMemberAt(_forceSyncSourceIndex).getHostAndPort();
_forceSyncSourceIndex = -1;
log() << "choosing sync source candidate by request: " << _syncSource;
std::string msg(str::stream()
<< "syncing from: " << _syncSource.toString() << " by request");
setMyHeartbeatMessage(now, msg);
return _syncSource;
}
// wait for 2N pings (not counting ourselves) before choosing a sync target
int needMorePings = (_memberData.size() - 1) * 2 - _getTotalPings();
if (needMorePings > 0) {
static Occasionally sampler;
if (sampler.tick()) {
log() << "waiting for " << needMorePings << " pings from other members before syncing";
}
_syncSource = HostAndPort();
return _syncSource;
}
// If we are only allowed to sync from the primary, set that
if (chainingPreference == ChainingPreference::kUseConfiguration &&
!_rsConfig.isChainingAllowed()) {
if (_currentPrimaryIndex == -1) {
LOG(1) << "Cannot select a sync source because chaining is"
" not allowed and primary is unknown/down";
_syncSource = HostAndPort();
return _syncSource;
} else if (_memberIsBlacklisted(*_currentPrimaryMember(), now)) {
LOG(1) << "Cannot select a sync source because chaining is not allowed and primary "
"member is blacklisted: "
<< _currentPrimaryMember()->getHostAndPort();
_syncSource = HostAndPort();
return _syncSource;
} else if (_currentPrimaryIndex == _selfIndex) {
LOG(1)
<< "Cannot select a sync source because chaining is not allowed and we are primary";
_syncSource = HostAndPort();
return _syncSource;
} else {
_syncSource = _currentPrimaryMember()->getHostAndPort();
log() << "chaining not allowed, choosing primary as sync source candidate: "
<< _syncSource;
std::string msg(str::stream() << "syncing from primary: " << _syncSource.toString());
setMyHeartbeatMessage(now, msg);
return _syncSource;
}
}
// find the member with the lowest ping time that is ahead of me
// choose a time that will exclude no candidates by default, in case we don't see a primary
OpTime oldestSyncOpTime;
// Find primary's oplog time. Reject sync candidates that are more than
// _options.maxSyncSourceLagSecs seconds behind.
if (_currentPrimaryIndex != -1) {
OpTime primaryOpTime = _memberData.at(_currentPrimaryIndex).getHeartbeatAppliedOpTime();
// Check if primaryOpTime is still close to 0 because we haven't received
// our first heartbeat from a new primary yet.
unsigned int maxLag =
static_cast<unsigned int>(durationCount<Seconds>(_options.maxSyncSourceLagSecs));
if (primaryOpTime.getSecs() >= maxLag) {
oldestSyncOpTime =
OpTime(Timestamp(primaryOpTime.getSecs() - maxLag, 0), primaryOpTime.getTerm());
}
}
int closestIndex = -1;
// Make two attempts, with less restrictive rules the second time.
//
// During the first attempt, we ignore those nodes that have a larger slave
// delay, hidden nodes or non-voting, and nodes that are excessively behind.
//
// For the second attempt include those nodes, in case those are the only ones we can reach.
//
// This loop attempts to set 'closestIndex', to select a viable candidate.
for (int attempts = 0; attempts < 2; ++attempts) {
for (std::vector<MemberData>::const_iterator it = _memberData.begin();
it != _memberData.end();
++it) {
const int itIndex = indexOfIterator(_memberData, it);
// Don't consider ourselves.
if (itIndex == _selfIndex) {
continue;
}
const MemberConfig& itMemberConfig(_rsConfig.getMemberAt(itIndex));
// Candidate must be up to be considered.
if (!it->up()) {
LOG(2) << "Cannot select sync source because it is not up: "
<< itMemberConfig.getHostAndPort();
continue;
}
// Candidate must be PRIMARY or SECONDARY state to be considered.
if (!it->getState().readable()) {
LOG(2) << "Cannot select sync source because it is not readable: "
<< itMemberConfig.getHostAndPort();
continue;
}
// On the first attempt, we skip candidates that do not match these criteria.
if (attempts == 0) {
// Candidate must be a voter if we are a voter.
if (_selfConfig().isVoter() && !itMemberConfig.isVoter()) {
LOG(2) << "Cannot select sync source because we are a voter and it is not: "
<< itMemberConfig.getHostAndPort();
continue;
}
// Candidates must not be hidden.
if (itMemberConfig.isHidden()) {
LOG(2) << "Cannot select sync source because it is hidden: "
<< itMemberConfig.getHostAndPort();
continue;
}
// Candidates cannot be excessively behind.
if (it->getHeartbeatAppliedOpTime() < oldestSyncOpTime) {
LOG(2) << "Cannot select sync source because it is too far behind."
<< "Latest optime of sync candidate " << itMemberConfig.getHostAndPort()
<< ": " << it->getHeartbeatAppliedOpTime()
<< ", oldest acceptable optime: " << oldestSyncOpTime;
continue;
}
// Candidate must not have a configured delay larger than ours.
if (_selfConfig().getSlaveDelay() < itMemberConfig.getSlaveDelay()) {
LOG(2) << "Cannot select sync source with larger slaveDelay than ours: "
<< itMemberConfig.getHostAndPort();
continue;
}
}
// Candidate must build indexes if we build indexes, to be considered.
if (_selfConfig().shouldBuildIndexes()) {
if (!itMemberConfig.shouldBuildIndexes()) {
LOG(2) << "Cannot select sync source with shouldBuildIndex differences: "
<< itMemberConfig.getHostAndPort();
continue;
}
}
// Only select a candidate that is ahead of me.
if (it->getHeartbeatAppliedOpTime() <= lastOpTimeFetched) {
LOG(1) << "Cannot select this sync source. Sync source must be ahead of me. "
<< "Sync candidate: " << itMemberConfig.getHostAndPort()
<< ", my last fetched oplog optime: " << lastOpTimeFetched.toBSON()
<< ", latest oplog optime of sync candidate: "
<< it->getHeartbeatAppliedOpTime().toBSON();
continue;
}
// Candidate cannot be more latent than anything we've already considered.
if ((closestIndex != -1) &&
(_getPing(itMemberConfig.getHostAndPort()) >
_getPing(_rsConfig.getMemberAt(closestIndex).getHostAndPort()))) {
LOG(2) << "Cannot select sync source with higher latency than the best candidate: "
<< itMemberConfig.getHostAndPort();
continue;
}
// Candidate cannot be blacklisted.
if (_memberIsBlacklisted(itMemberConfig, now)) {
LOG(1) << "Cannot select sync source which is blacklisted: "
<< itMemberConfig.getHostAndPort();
continue;
}
// This candidate has passed all tests; set 'closestIndex'
closestIndex = itIndex;
}
if (closestIndex != -1)
break; // no need for second attempt
}
if (closestIndex == -1) {
// Did not find any members to sync from
std::string msg("could not find member to sync from");
// Only log when we had a valid sync source before
if (!_syncSource.empty()) {
log() << msg << rsLog;
}
setMyHeartbeatMessage(now, msg);
_syncSource = HostAndPort();
return _syncSource;
}
_syncSource = _rsConfig.getMemberAt(closestIndex).getHostAndPort();
log() << "sync source candidate: " << _syncSource;
std::string msg(str::stream() << "syncing from: " << _syncSource.toString(), 0);
setMyHeartbeatMessage(now, msg);
return _syncSource;
}
bool TopologyCoordinator::_memberIsBlacklisted(const MemberConfig& memberConfig, Date_t now) const {
std::map<HostAndPort, Date_t>::const_iterator blacklisted =
_syncSourceBlacklist.find(memberConfig.getHostAndPort());
if (blacklisted != _syncSourceBlacklist.end()) {
if (blacklisted->second > now) {
return true;
}
}
return false;
}
void TopologyCoordinator::blacklistSyncSource(const HostAndPort& host, Date_t until) {
LOG(2) << "blacklisting " << host << " until " << until.toString();
_syncSourceBlacklist[host] = until;
}
void TopologyCoordinator::unblacklistSyncSource(const HostAndPort& host, Date_t now) {
std::map<HostAndPort, Date_t>::iterator hostItr = _syncSourceBlacklist.find(host);
if (hostItr != _syncSourceBlacklist.end() && now >= hostItr->second) {
LOG(2) << "unblacklisting " << host;
_syncSourceBlacklist.erase(hostItr);
}
}
void TopologyCoordinator::clearSyncSourceBlacklist() {
_syncSourceBlacklist.clear();
}
void TopologyCoordinator::prepareSyncFromResponse(const HostAndPort& target,
BSONObjBuilder* response,
Status* result) {
response->append("syncFromRequested", target.toString());
if (_selfIndex == -1) {
*result = Status(ErrorCodes::NotSecondary, "Removed and uninitialized nodes do not sync");
return;
}
const MemberConfig& selfConfig = _selfConfig();
if (selfConfig.isArbiter()) {
*result = Status(ErrorCodes::NotSecondary, "arbiters don't sync");
return;
}
if (_selfIndex == _currentPrimaryIndex) {
*result = Status(ErrorCodes::NotSecondary, "primaries don't sync");
return;
}
ReplSetConfig::MemberIterator targetConfig = _rsConfig.membersEnd();
int targetIndex = 0;
for (ReplSetConfig::MemberIterator it = _rsConfig.membersBegin(); it != _rsConfig.membersEnd();
++it) {
if (it->getHostAndPort() == target) {
targetConfig = it;
break;
}
++targetIndex;
}
if (targetConfig == _rsConfig.membersEnd()) {
*result = Status(ErrorCodes::NodeNotFound,
str::stream() << "Could not find member \"" << target.toString()
<< "\" in replica set");
return;
}
if (targetIndex == _selfIndex) {
*result = Status(ErrorCodes::InvalidOptions, "I cannot sync from myself");
return;
}
if (targetConfig->isArbiter()) {
*result = Status(ErrorCodes::InvalidOptions,
str::stream() << "Cannot sync from \"" << target.toString()
<< "\" because it is an arbiter");
return;
}
if (!targetConfig->shouldBuildIndexes() && selfConfig.shouldBuildIndexes()) {
*result = Status(ErrorCodes::InvalidOptions,
str::stream() << "Cannot sync from \"" << target.toString()
<< "\" because it does not build indexes");
return;
}
if (selfConfig.isVoter() && !targetConfig->isVoter()) {
*result = Status(ErrorCodes::InvalidOptions,
str::stream() << "Cannot sync from \"" << target.toString()
<< "\" because it is not a voter");
return;
}
const MemberData& hbdata = _memberData.at(targetIndex);
if (hbdata.hasAuthIssue()) {
*result =
Status(ErrorCodes::Unauthorized,
str::stream() << "not authorized to communicate with " << target.toString());
return;
}
if (hbdata.getHealth() == 0) {
*result =
Status(ErrorCodes::HostUnreachable,
str::stream() << "I cannot reach the requested member: " << target.toString());
return;
}
const OpTime lastOpApplied = getMyLastAppliedOpTime();
if (hbdata.getHeartbeatAppliedOpTime().getSecs() + 10 < lastOpApplied.getSecs()) {
warning() << "attempting to sync from " << target << ", but its latest opTime is "
<< hbdata.getHeartbeatAppliedOpTime().getSecs() << " and ours is "
<< lastOpApplied.getSecs() << " so this may not work";
response->append("warning",
str::stream() << "requested member \"" << target.toString()
<< "\" is more than 10 seconds behind us");
// not returning bad Status, just warning
}
HostAndPort prevSyncSource = getSyncSourceAddress();
if (!prevSyncSource.empty()) {
response->append("prevSyncTarget", prevSyncSource.toString());
}
setForceSyncSourceIndex(targetIndex);
*result = Status::OK();
}
// produce a reply to a heartbeat
Status TopologyCoordinator::prepareHeartbeatResponseV1(Date_t now,
const ReplSetHeartbeatArgsV1& args,
const std::string& ourSetName,
ReplSetHeartbeatResponse* response) {
// Verify that replica set names match
const std::string rshb = args.getSetName();
if (ourSetName != rshb) {
log() << "replSet set names do not match, ours: " << ourSetName
<< "; remote node's: " << rshb;
return Status(ErrorCodes::InconsistentReplicaSetNames,
str::stream() << "Our set name of " << ourSetName << " does not match name "
<< rshb << " reported by remote node");
}
const MemberState myState = getMemberState();
if (_selfIndex == -1) {
if (myState.removed()) {
return Status(ErrorCodes::InvalidReplicaSetConfig,
"Our replica set configuration is invalid or does not include us");
}
} else {
if (args.getSenderId() == _selfConfig().getId().getData()) {
return Status(ErrorCodes::BadValue,
str::stream() << "Received heartbeat from member with the same "
"member ID as ourself: "
<< args.getSenderId());
}
}
response->setSetName(ourSetName);
response->setState(myState.s);
if (myState.primary()) {
response->setElectionTime(_electionTime);
}
const OpTimeAndWallTime lastOpApplied = getMyLastAppliedOpTimeAndWallTime();
const OpTimeAndWallTime lastOpDurable = getMyLastDurableOpTimeAndWallTime();
response->setAppliedOpTimeAndWallTime(lastOpApplied);
response->setDurableOpTimeAndWallTime(lastOpDurable);
if (_currentPrimaryIndex != -1) {
response->setPrimaryId(_rsConfig.getMemberAt(_currentPrimaryIndex).getId().getData());
}
response->setTerm(_term);
if (!_syncSource.empty()) {
response->setSyncingTo(_syncSource);
}
if (!_rsConfig.isInitialized()) {
response->setConfigVersion(-2);
return Status::OK();
}
const long long v = _rsConfig.getConfigVersion();
response->setConfigVersion(v);
// Deliver new config if caller's version is older than ours
if (v > args.getConfigVersion()) {
response->setConfig(_rsConfig);
}
// Resolve the caller's id in our Member list
int from = -1;
if (v == args.getConfigVersion() && args.getSenderId() != -1) {
from = _getMemberIndex(args.getSenderId());
}
if (from == -1) {
return Status::OK();
}
invariant(from != _selfIndex);
auto& fromNodeData = _memberData.at(from);
// note that we got a heartbeat from this node
fromNodeData.setLastHeartbeatRecv(now);
// Update liveness for sending node.
fromNodeData.updateLiveness(now);
return Status::OK();
}
int TopologyCoordinator::_getMemberIndex(int id) const {
int index = 0;
for (ReplSetConfig::MemberIterator it = _rsConfig.membersBegin(); it != _rsConfig.membersEnd();
++it, ++index) {
if (it->getId() == MemberId(id)) {
return index;
}
}
return -1;
}
std::pair<ReplSetHeartbeatArgsV1, Milliseconds> TopologyCoordinator::prepareHeartbeatRequestV1(
Date_t now, const std::string& ourSetName, const HostAndPort& target) {
PingStats& hbStats = _pings[target];
Milliseconds alreadyElapsed(now.asInt64() - hbStats.getLastHeartbeatStartDate().asInt64());
if ((!_rsConfig.isInitialized()) || !hbStats.trying() ||
(alreadyElapsed >= _rsConfig.getHeartbeatTimeoutPeriodMillis())) {
// This is either the first request ever for "target", or the heartbeat timeout has
// passed, so we're starting a "new" heartbeat.
hbStats.start(now);
alreadyElapsed = Milliseconds(0);
}
ReplSetHeartbeatArgsV1 hbArgs;
if (_rsConfig.isInitialized()) {
hbArgs.setSetName(_rsConfig.getReplSetName());
hbArgs.setConfigVersion(_rsConfig.getConfigVersion());
if (_selfIndex >= 0) {
const MemberConfig& me = _selfConfig();
hbArgs.setSenderId(me.getId().getData());
hbArgs.setSenderHost(me.getHostAndPort());
}
hbArgs.setTerm(_term);
} else {
hbArgs.setSetName(ourSetName);
// Config version -2 is for uninitialized config.
hbArgs.setConfigVersion(-2);
hbArgs.setTerm(OpTime::kInitialTerm);
}
hbArgs.setHeartbeatVersion(1);
const Milliseconds timeoutPeriod(
_rsConfig.isInitialized() ? _rsConfig.getHeartbeatTimeoutPeriodMillis()
: Milliseconds{ReplSetConfig::kDefaultHeartbeatTimeoutPeriod});
const Milliseconds timeout(timeoutPeriod - alreadyElapsed);
return std::make_pair(hbArgs, timeout);
}
HeartbeatResponseAction TopologyCoordinator::processHeartbeatResponse(
Date_t now,
Milliseconds networkRoundTripTime,
const HostAndPort& target,
const StatusWith<ReplSetHeartbeatResponse>& hbResponse) {
const MemberState originalState = getMemberState();
PingStats& hbStats = _pings[target];
invariant(hbStats.getLastHeartbeatStartDate() != Date_t());
const bool isUnauthorized = (hbResponse.getStatus().code() == ErrorCodes::Unauthorized) ||
(hbResponse.getStatus().code() == ErrorCodes::AuthenticationFailed);
if (hbResponse.isOK() || isUnauthorized) {
hbStats.hit(networkRoundTripTime);
} else {
hbStats.miss();
}
// If a node is not PRIMARY and has no sync source, we increase the heartbeat rate in order
// to help it find a sync source more quickly, which helps ensure the PRIMARY will continue to
// see the majority of the cluster.
//
// Arbiters also decrease their heartbeat interval to at most half the election timeout period.
Milliseconds heartbeatInterval = _rsConfig.getHeartbeatInterval();
if (getMemberState().arbiter()) {
heartbeatInterval =
std::min(_rsConfig.getElectionTimeoutPeriod() / 2, _rsConfig.getHeartbeatInterval());
} else if (getSyncSourceAddress().empty() && !_iAmPrimary()) {
heartbeatInterval = std::min(_rsConfig.getElectionTimeoutPeriod() / 2,
_rsConfig.getHeartbeatInterval() / 4);
}
const Milliseconds alreadyElapsed = now - hbStats.getLastHeartbeatStartDate();
Date_t nextHeartbeatStartDate;
// Determine the next heartbeat start time. If a heartbeat has not succeeded or failed, and we
// have not used up the timeout period, we should retry.
if (hbStats.trying() && (alreadyElapsed < _rsConfig.getHeartbeatTimeoutPeriod())) {
// There are still retries left, let's use one.
nextHeartbeatStartDate = now;
} else {
nextHeartbeatStartDate = now + heartbeatInterval;
}
if (hbStats.failed()) {
LOG_FOR_HEARTBEATS(0) << "Heartbeat to " << target << " failed after "
<< kMaxHeartbeatRetries
<< " retries, response status: " << hbResponse.getStatus();
}
if (hbResponse.isOK() && hbResponse.getValue().hasConfig()) {
const long long currentConfigVersion =
_rsConfig.isInitialized() ? _rsConfig.getConfigVersion() : -2;
const ReplSetConfig& newConfig = hbResponse.getValue().getConfig();
if (newConfig.getConfigVersion() > currentConfigVersion) {
HeartbeatResponseAction nextAction = HeartbeatResponseAction::makeReconfigAction();
nextAction.setNextHeartbeatStartDate(nextHeartbeatStartDate);
return nextAction;
} else {
// Could be we got the newer version before we got the response, or the
// target erroneously sent us one, even through it isn't newer.
if (newConfig.getConfigVersion() < currentConfigVersion) {
LOG(1) << "Config version from heartbeat was older than ours.";
} else {
LOG(2) << "Config from heartbeat response was same as ours.";
}
if (logger::globalLogDomain()->shouldLog(MongoLogDefaultComponent_component,
::mongo::LogstreamBuilder::severityCast(2))) {
LogstreamBuilder lsb = log();
if (_rsConfig.isInitialized()) {
lsb << "Current config: " << _rsConfig.toBSON() << "; ";
}
lsb << "Config in heartbeat: " << newConfig.toBSON();
}
}
}
// Check if the heartbeat target is in our config. If it isn't, there's nothing left to do,
// so return early.
if (!_rsConfig.isInitialized()) {
HeartbeatResponseAction nextAction = HeartbeatResponseAction::makeNoAction();
nextAction.setNextHeartbeatStartDate(nextHeartbeatStartDate);
return nextAction;
}
// If we're not in the config, we don't need to respond to heartbeats.
if (_selfIndex == -1) {
LOG(1) << "Could not find ourself in current config so ignoring heartbeat from " << target
<< " -- current config: " << _rsConfig.toBSON();
HeartbeatResponseAction nextAction = HeartbeatResponseAction::makeNoAction();
nextAction.setNextHeartbeatStartDate(nextHeartbeatStartDate);
return nextAction;
}
const int memberIndex = _rsConfig.findMemberIndexByHostAndPort(target);
if (memberIndex == -1) {
LOG(1) << "Could not find " << target
<< " in current config so ignoring --"
" current config: "
<< _rsConfig.toBSON();
HeartbeatResponseAction nextAction = HeartbeatResponseAction::makeNoAction();
nextAction.setNextHeartbeatStartDate(nextHeartbeatStartDate);
return nextAction;
}
invariant(memberIndex != _selfIndex);
MemberData& hbData = _memberData.at(memberIndex);
const MemberConfig member = _rsConfig.getMemberAt(memberIndex);
bool advancedOpTime = false;
if (!hbResponse.isOK()) {
if (isUnauthorized) {
hbData.setAuthIssue(now);
}
// If the heartbeat has failed i.e. used up all retries, then we mark the target node as
// down.
else if (hbStats.failed() || (alreadyElapsed >= _rsConfig.getHeartbeatTimeoutPeriod())) {
hbData.setDownValues(now, hbResponse.getStatus().reason());
} else {
LOG(3) << "Bad heartbeat response from " << target
<< "; trying again; Retries left: " << (hbStats.retriesLeft()) << "; "
<< alreadyElapsed << " have already elapsed";
}
} else {
ReplSetHeartbeatResponse hbr = std::move(hbResponse.getValue());
LOG(3) << "setUpValues: heartbeat response good for member _id:" << member.getId();
advancedOpTime = hbData.setUpValues(now, std::move(hbr));
}
HeartbeatResponseAction nextAction;
nextAction = _updatePrimaryFromHBDataV1(memberIndex, originalState, now);
nextAction.setNextHeartbeatStartDate(nextHeartbeatStartDate);
nextAction.setAdvancedOpTime(advancedOpTime);
return nextAction;
}
bool TopologyCoordinator::haveNumNodesReachedOpTime(const OpTime& targetOpTime,
int numNodes,
bool durablyWritten) {
// Replication progress that is for some reason ahead of us should not allow us to
// satisfy a write concern if we aren't caught up ourselves.
OpTime myOpTime = durablyWritten ? getMyLastDurableOpTime() : getMyLastAppliedOpTime();
if (myOpTime < targetOpTime) {
return false;
}
// Invariants that we only wait for an OpTime in the term that this node is currently writing
// to. In other words, we do not support waiting for an OpTime written by a previous primary
// because comparing members' lastApplied/lastDurable alone is not sufficient to tell if the
// OpTime has been replicated.
invariant(targetOpTime.getTerm() == getMyLastAppliedOpTime().getTerm());
for (auto&& memberData : _memberData) {
const auto isArbiter = _rsConfig.getMemberAt(memberData.getConfigIndex()).isArbiter();
// We do not count arbiters towards the write concern.
if (isArbiter) {
continue;
}
const OpTime& memberOpTime =
durablyWritten ? memberData.getLastDurableOpTime() : memberData.getLastAppliedOpTime();
// In addition to checking if a member has a greater/equal timestamp field we also need to
// make sure that the memberOpTime is in the same term as the OpTime we wait for. If a
// member's OpTime has a higher term, it indicates that this node will be stepping down. And
// thus we do not know if the target OpTime in our previous term has been replicated to the
// member because the memberOpTime in a higher term could correspond to an operation in a
// divergent branch of history regardless of its timestamp.
if (memberOpTime.getTerm() == targetOpTime.getTerm() &&
memberOpTime.getTimestamp() >= targetOpTime.getTimestamp()) {
--numNodes;
}
if (numNodes <= 0) {
return true;
}
}
return false;
}
bool TopologyCoordinator::haveTaggedNodesReachedOpTime(const OpTime& opTime,
const ReplSetTagPattern& tagPattern,
bool durablyWritten) {
ReplSetTagMatch matcher(tagPattern);
// Invariants that we only wait for an OpTime in the term that this node is currently writing
// to. In other words, we do not support waiting for an OpTime written by a previous primary
// because comparing members' lastApplied/lastDurable alone is not sufficient to tell if the
// OpTime has been replicated.
invariant(opTime.getTerm() == getMyLastAppliedOpTime().getTerm());
for (auto&& memberData : _memberData) {
const OpTime& memberOpTime =
durablyWritten ? memberData.getLastDurableOpTime() : memberData.getLastAppliedOpTime();
// In addition to checking if a member has a greater/equal timestamp field we also need to
// make sure that the memberOpTime is in the same term as the OpTime we wait for. If a
// member's OpTime has a higher term, it indicates that this node will be stepping down. And
// thus we do not know if the target OpTime in our previous term has been replicated to the
// member because the memberOpTime in a higher term could correspond to an operation in a
// divergent branch of history regardless of its timestamp.
if (memberOpTime.getTerm() == opTime.getTerm() &&
memberOpTime.getTimestamp() >= opTime.getTimestamp()) {
// This node has reached the desired optime, now we need to check if it is a part
// of the tagPattern.
int memberIndex = memberData.getConfigIndex();
invariant(memberIndex >= 0);
const MemberConfig& memberConfig = _rsConfig.getMemberAt(memberIndex);
for (MemberConfig::TagIterator it = memberConfig.tagsBegin();
it != memberConfig.tagsEnd();
++it) {
if (matcher.update(*it)) {
return true;
}
}
}
}
return false;
}
HeartbeatResponseAction TopologyCoordinator::checkMemberTimeouts(Date_t now) {
bool stepdown = false;
for (int memberIndex = 0; memberIndex < static_cast<int>(_memberData.size()); memberIndex++) {
auto& memberData = _memberData[memberIndex];
if (!memberData.isSelf() && !memberData.lastUpdateStale() &&
now - memberData.getLastUpdate() >= _rsConfig.getElectionTimeoutPeriod()) {
memberData.markLastUpdateStale();
if (_iAmPrimary()) {
stepdown = stepdown || setMemberAsDown(now, memberIndex);
}
}
}
if (stepdown) {
log() << "can't see a majority of the set, relinquishing primary";
return HeartbeatResponseAction::makeStepDownSelfAction(_selfIndex);
}
return HeartbeatResponseAction::makeNoAction();
}
std::vector<HostAndPort> TopologyCoordinator::getHostsWrittenTo(const OpTime& op,
bool durablyWritten) {
std::vector<HostAndPort> hosts;
for (const auto& memberData : _memberData) {
if (durablyWritten) {
if (memberData.getLastDurableOpTime() < op) {
continue;
}
} else if (memberData.getLastAppliedOpTime() < op) {
continue;
}
hosts.push_back(memberData.getHostAndPort());
}
return hosts;
}
bool TopologyCoordinator::setMemberAsDown(Date_t now, const int memberIndex) {
invariant(memberIndex != _selfIndex);
invariant(memberIndex != -1);
invariant(_currentPrimaryIndex == _selfIndex);
MemberData& hbData = _memberData.at(memberIndex);
hbData.setDownValues(now, "no response within election timeout period");
if (CannotSeeMajority &
_getMyUnelectableReason(now, StartElectionReasonEnum::kElectionTimeout)) {
return true;
}
return false;
}
std::pair<MemberId, Date_t> TopologyCoordinator::getStalestLiveMember() const {
Date_t earliestDate = Date_t::max();
MemberId earliestMemberId;
for (const auto& memberData : _memberData) {
if (memberData.isSelf()) {
continue;
}
if (memberData.lastUpdateStale()) {
// Already stale.
continue;
}
LOG(3) << "memberData lastupdate is: " << memberData.getLastUpdate();
if (earliestDate > memberData.getLastUpdate()) {
earliestDate = memberData.getLastUpdate();
earliestMemberId = memberData.getMemberId();
}
}
LOG(3) << "stalest member " << earliestMemberId << " date: " << earliestDate;
return std::make_pair(earliestMemberId, earliestDate);
}
void TopologyCoordinator::resetAllMemberTimeouts(Date_t now) {
for (auto&& memberData : _memberData)
memberData.updateLiveness(now);
}
void TopologyCoordinator::resetMemberTimeouts(Date_t now,
const stdx::unordered_set<HostAndPort>& member_set) {
for (auto&& memberData : _memberData) {
if (member_set.count(memberData.getHostAndPort()))
memberData.updateLiveness(now);
}
}
OpTime TopologyCoordinator::getMyLastAppliedOpTime() const {
return _selfMemberData().getLastAppliedOpTime();
}
OpTimeAndWallTime TopologyCoordinator::getMyLastAppliedOpTimeAndWallTime() const {
return {_selfMemberData().getLastAppliedOpTime(), _selfMemberData().getLastAppliedWallTime()};
}
void TopologyCoordinator::setMyLastAppliedOpTimeAndWallTime(OpTimeAndWallTime opTimeAndWallTime,
Date_t now,
bool isRollbackAllowed) {
auto opTime = opTimeAndWallTime.opTime;
auto& myMemberData = _selfMemberData();
auto myLastAppliedOpTime = myMemberData.getLastAppliedOpTime();
if (!(isRollbackAllowed || opTime == myLastAppliedOpTime)) {
invariant(opTime > myLastAppliedOpTime);
// In pv1, oplog entries are ordered by non-decreasing term and strictly increasing
// timestamp. So, in pv1, its not possible for us to get opTime with higher term and
// timestamp lesser than or equal to our current lastAppliedOptime.
invariant(opTime.getTerm() == OpTime::kUninitializedTerm ||
myLastAppliedOpTime.getTerm() == OpTime::kUninitializedTerm ||
opTime.getTimestamp() > myLastAppliedOpTime.getTimestamp());
}
myMemberData.setLastAppliedOpTimeAndWallTime(opTimeAndWallTime, now);
}
OpTime TopologyCoordinator::getMyLastDurableOpTime() const {
return _selfMemberData().getLastDurableOpTime();
}
OpTimeAndWallTime TopologyCoordinator::getMyLastDurableOpTimeAndWallTime() const {
return {_selfMemberData().getLastDurableOpTime(), _selfMemberData().getLastDurableWallTime()};
}
void TopologyCoordinator::setMyLastDurableOpTimeAndWallTime(OpTimeAndWallTime opTimeAndWallTime,
Date_t now,
bool isRollbackAllowed) {
auto opTime = opTimeAndWallTime.opTime;
auto& myMemberData = _selfMemberData();
invariant(isRollbackAllowed || opTime >= myMemberData.getLastDurableOpTime());
myMemberData.setLastDurableOpTimeAndWallTime(opTimeAndWallTime, now);
}
StatusWith<bool> TopologyCoordinator::setLastOptime(const UpdatePositionArgs::UpdateInfo& args,
Date_t now,
long long* configVersion) {
if (_selfIndex == -1) {
// Ignore updates when we're in state REMOVED.
return Status(ErrorCodes::NotMasterOrSecondary,
"Received replSetUpdatePosition command but we are in state REMOVED");
}
invariant(_rsConfig.isInitialized()); // Can only use setLastOptime in replSet mode.
MemberId memberId;
try {
memberId = MemberId(args.memberId);
} catch (const DBException& ex) {
return ex.toStatus();
}
if (memberId == _rsConfig.getMemberAt(_selfIndex).getId()) {
// Do not let remote nodes tell us what our optime is.
return false;
}
LOG(2) << "received notification that node with memberID " << memberId
<< " in config with version " << args.cfgver
<< " has reached optime: " << args.appliedOpTime
<< " and is durable through: " << args.durableOpTime;
if (args.cfgver != _rsConfig.getConfigVersion()) {
std::string errmsg = str::stream()
<< "Received replSetUpdatePosition for node with memberId " << memberId
<< " whose config version of " << args.cfgver << " doesn't match our config version of "
<< _rsConfig.getConfigVersion();
LOG(1) << errmsg;
*configVersion = _rsConfig.getConfigVersion();
return Status(ErrorCodes::InvalidReplicaSetConfig, errmsg);
}
auto* memberData = _findMemberDataByMemberId(memberId.getData());
if (!memberData) {
invariant(!_rsConfig.findMemberByID(memberId.getData()));
std::string errmsg = str::stream()
<< "Received replSetUpdatePosition for node with memberId " << memberId
<< " which doesn't exist in our config";
LOG(1) << errmsg;
return Status(ErrorCodes::NodeNotFound, errmsg);
}
invariant(memberId == memberData->getMemberId());
LOG(3) << "Node with memberID " << memberId << " currently has optime "
<< memberData->getLastAppliedOpTime() << " durable through "
<< memberData->getLastDurableOpTime() << "; updating to optime " << args.appliedOpTime
<< " and durable through " << args.durableOpTime;
bool advancedOpTime = memberData->advanceLastAppliedOpTimeAndWallTime(
{args.appliedOpTime, args.appliedWallTime}, now);
advancedOpTime = memberData->advanceLastDurableOpTimeAndWallTime(
{args.durableOpTime, args.durableWallTime}, now) ||
advancedOpTime;
return advancedOpTime;
}
MemberData* TopologyCoordinator::_findMemberDataByMemberId(const int memberId) {
const int memberIndex = _getMemberIndex(memberId);
if (memberIndex >= 0)
return &_memberData[memberIndex];
return nullptr;
}
HeartbeatResponseAction TopologyCoordinator::_updatePrimaryFromHBDataV1(
int updatedConfigIndex, const MemberState& originalState, Date_t now) {
//
// Updates the local notion of which remote node, if any is primary.
// Start the priority takeover process if we are eligible.
//
invariant(updatedConfigIndex != _selfIndex);
// If we are missing from the config, do not participate in primary maintenance or election.
if (_selfIndex == -1) {
return HeartbeatResponseAction::makeNoAction();
}
// If we are the primary, there must be no other primary, otherwise its higher term would
// have already made us step down.
if (_currentPrimaryIndex == _selfIndex) {
return HeartbeatResponseAction::makeNoAction();
}
// Scan the member list's heartbeat data for who is primary, and update _currentPrimaryIndex.
int primaryIndex = -1;
for (size_t i = 0; i < _memberData.size(); i++) {
const MemberData& member = _memberData.at(i);
if (member.getState().primary() && member.up()) {
if (primaryIndex == -1 || _memberData.at(primaryIndex).getTerm() < member.getTerm()) {
primaryIndex = i;
}
}
}
_currentPrimaryIndex = primaryIndex;
if (_currentPrimaryIndex == -1) {
return HeartbeatResponseAction::makeNoAction();
}
// Clear last heartbeat message on ourselves.
setMyHeartbeatMessage(now, "");
// Takeover when the replset is stable.
//
// Take over the primary only if the remote primary is in the latest term I know.
// This is done only when we get a heartbeat response from the primary.
// Otherwise, there must be an outstanding election, which may succeed or not, but
// the remote primary will become aware of that election eventually and step down.
if (_memberData.at(primaryIndex).getTerm() == _term && updatedConfigIndex == primaryIndex) {
// Don't schedule catchup takeover if catchup takeover or primary catchup is disabled.
bool catchupTakeoverDisabled =
ReplSetConfig::kCatchUpDisabled == _rsConfig.getCatchUpTimeoutPeriod() ||
ReplSetConfig::kCatchUpTakeoverDisabled == _rsConfig.getCatchUpTakeoverDelay();
bool scheduleCatchupTakeover = false;
bool schedulePriorityTakeover = false;
if (!catchupTakeoverDisabled &&
(_memberData.at(primaryIndex).getLastAppliedOpTime() <
_memberData.at(_selfIndex).getLastAppliedOpTime())) {
LOG_FOR_ELECTION(2) << "I can take over the primary due to fresher data."
<< " Current primary index: " << primaryIndex << " in term "
<< _memberData.at(primaryIndex).getTerm() << "."
<< " Current primary optime: "
<< _memberData.at(primaryIndex).getLastAppliedOpTime()
<< " My optime: "
<< _memberData.at(_selfIndex).getLastAppliedOpTime();
LOG_FOR_ELECTION(4) << _getReplSetStatusString();
scheduleCatchupTakeover = true;
}
if (_rsConfig.getMemberAt(primaryIndex).getPriority() <
_rsConfig.getMemberAt(_selfIndex).getPriority()) {
LOG_FOR_ELECTION(2) << "I can take over the primary due to higher priority."
<< " Current primary index: " << primaryIndex << " in term "
<< _memberData.at(primaryIndex).getTerm();
LOG_FOR_ELECTION(4) << _getReplSetStatusString();
schedulePriorityTakeover = true;
}
// Calculate rank of current node. A rank of 0 indicates that it has the highest priority.
auto currentNodePriority = _rsConfig.getMemberAt(_selfIndex).getPriority();
// Schedule a priority takeover early only if we know that the current node has the highest
// priority in the replica set, has a higher priority than the primary, and is the most
// up to date node.
// Otherwise, prefer to schedule a catchup takeover over a priority takeover
if (scheduleCatchupTakeover && schedulePriorityTakeover &&
_rsConfig.calculatePriorityRank(currentNodePriority) == 0) {
LOG_FOR_ELECTION(2)
<< "I can take over the primary because I have a higher priority, the highest "
<< "priority in the replica set, and fresher data."
<< " Current primary index: " << primaryIndex << " in term "
<< _memberData.at(primaryIndex).getTerm();
return HeartbeatResponseAction::makePriorityTakeoverAction();
}
if (scheduleCatchupTakeover) {
return HeartbeatResponseAction::makeCatchupTakeoverAction();
}
if (schedulePriorityTakeover) {
return HeartbeatResponseAction::makePriorityTakeoverAction();
}
}
return HeartbeatResponseAction::makeNoAction();
}
bool TopologyCoordinator::_aMajoritySeemsToBeUp() const {
int vUp = 0;
for (std::vector<MemberData>::const_iterator it = _memberData.begin(); it != _memberData.end();
++it) {
const int itIndex = indexOfIterator(_memberData, it);
if (itIndex == _selfIndex || it->up()) {
vUp += _rsConfig.getMemberAt(itIndex).getNumVotes();
}
}
return vUp * 2 > _rsConfig.getTotalVotingMembers();
}
int TopologyCoordinator::_findHealthyPrimaryOfEqualOrGreaterPriority(
const int candidateIndex) const {
const double candidatePriority = _rsConfig.getMemberAt(candidateIndex).getPriority();
for (auto it = _memberData.begin(); it != _memberData.end(); ++it) {
if (!it->up() || it->getState() != MemberState::RS_PRIMARY) {
continue;
}
const int itIndex = indexOfIterator(_memberData, it);
const double priority = _rsConfig.getMemberAt(itIndex).getPriority();
if (itIndex != candidateIndex && priority >= candidatePriority) {
return itIndex;
}
}
return -1;
}
bool TopologyCoordinator::_amIFreshEnoughForPriorityTakeover() const {
const OpTime ourLatestKnownOpTime = latestKnownOpTime();
// Rules are:
// - If the terms don't match, we don't call for priority takeover.
// - If our optime and the latest optime happen in different seconds, our optime must be within
// at least priorityTakeoverFreshnessWindowSeconds seconds of the latest optime.
// - If our optime and the latest optime happen in the same second, our optime must be within
// at least 1000 oplog entries of the latest optime (i.e. the increment portion of the timestamp
// must be within 1000). This is to handle the case where a primary had its clock set far into
// the future, took some writes, then had its clock set back. In that case the timestamp
// component of all future oplog entries generated will be the same, until real world time
// passes the timestamp component of the last oplog entry.
const OpTime ourLastOpApplied = getMyLastAppliedOpTime();
if (ourLastOpApplied.getTerm() != ourLatestKnownOpTime.getTerm()) {
return false;
}
if (ourLastOpApplied.getTimestamp().getSecs() !=
ourLatestKnownOpTime.getTimestamp().getSecs()) {
return ourLastOpApplied.getTimestamp().getSecs() +
gPriorityTakeoverFreshnessWindowSeconds >=
ourLatestKnownOpTime.getTimestamp().getSecs();
} else {
return ourLastOpApplied.getTimestamp().getInc() + 1000 >=
ourLatestKnownOpTime.getTimestamp().getInc();
}
}
bool TopologyCoordinator::_amIFreshEnoughForCatchupTakeover() const {
const OpTime ourLatestKnownOpTime = latestKnownOpTime();
// Rules are:
// - We must have the freshest optime of all the up nodes.
// - We must specifically have a fresher optime than the primary (can't be equal).
// - The term of our last applied op must be less than the current term. This ensures that no
// writes have happened since the most recent election and that the primary is still in
// catchup mode.
// There is no point to a catchup takeover if we aren't the freshest node because
// another node would immediately perform another catchup takeover when we become primary.
const OpTime ourLastOpApplied = getMyLastAppliedOpTime();
if (ourLastOpApplied < ourLatestKnownOpTime) {
return false;
}
if (_currentPrimaryIndex == -1) {
return false;
}
// If we aren't ahead of the primary, there is no point to having a catchup takeover.
const OpTime primaryLastOpApplied = _memberData[_currentPrimaryIndex].getLastAppliedOpTime();
if (ourLastOpApplied <= primaryLastOpApplied) {
return false;
}
// If the term of our last applied op is less than the current term, the primary didn't write
// anything and it is still in catchup mode.
return ourLastOpApplied.getTerm() < _term;
}
bool TopologyCoordinator::_iAmPrimary() const {
if (_role == Role::kLeader) {
invariant(_currentPrimaryIndex == _selfIndex);
invariant(_leaderMode != LeaderMode::kNotLeader);
return true;
}
return false;
}
bool TopologyCoordinator::prepareForUnconditionalStepDown() {
if (_leaderMode == LeaderMode::kSteppingDown) {
// Can only be processing one required stepdown at a time.
return false;
}
// Heartbeat and reconfig (via cmd or heartbeat) initiated stepdowns take precedence over
// stepdown command initiated stepdowns, so it's safe to transition from kAttemptingStepDown
// to kSteppingDown.
_setLeaderMode(LeaderMode::kSteppingDown);
return true;
}
StatusWith<TopologyCoordinator::StepDownAttemptAbortFn>
TopologyCoordinator::prepareForStepDownAttempt() {
if (_leaderMode == LeaderMode::kSteppingDown ||
_leaderMode == LeaderMode::kAttemptingStepDown) {
return Status{ErrorCodes::ConflictingOperationInProgress,
"This node is already in the process of stepping down"};
}
if (_leaderMode == LeaderMode::kNotLeader) {
return Status{ErrorCodes::NotMaster, "This node is not a primary."};
}
invariant(_leaderMode == LeaderMode::kMaster || _leaderMode == LeaderMode::kLeaderElect);
const auto previousLeaderMode = _leaderMode;
_setLeaderMode(LeaderMode::kAttemptingStepDown);
return {[this, previousLeaderMode] {
if (_leaderMode == TopologyCoordinator::LeaderMode::kAttemptingStepDown) {
_setLeaderMode(previousLeaderMode);
}
}};
}
void TopologyCoordinator::changeMemberState_forTest(const MemberState& newMemberState,
const Timestamp& electionTime) {
invariant(_selfIndex != -1);
if (newMemberState == getMemberState())
return;
switch (newMemberState.s) {
case MemberState::RS_PRIMARY:
_role = Role::kCandidate;
processWinElection(OID(), electionTime);
invariant(_role == Role::kLeader);
break;
case MemberState::RS_SECONDARY:
case MemberState::RS_ROLLBACK:
case MemberState::RS_RECOVERING:
case MemberState::RS_STARTUP2:
_role = Role::kFollower;
_followerMode = newMemberState.s;
if (_currentPrimaryIndex == _selfIndex) {
_currentPrimaryIndex = -1;
_setLeaderMode(LeaderMode::kNotLeader);
}
break;
case MemberState::RS_STARTUP:
updateConfig(ReplSetConfig(), -1, Date_t());
break;
default:
severe() << "Cannot switch to state " << newMemberState;
MONGO_UNREACHABLE;
}
if (getMemberState() != newMemberState.s) {
severe() << "Expected to enter state " << newMemberState << " but am now in "
<< getMemberState();
MONGO_UNREACHABLE;
}
log() << newMemberState;
}
void TopologyCoordinator::setCurrentPrimary_forTest(int primaryIndex,
const Timestamp& electionTime) {
if (primaryIndex == _selfIndex) {
changeMemberState_forTest(MemberState::RS_PRIMARY);
} else {
if (_iAmPrimary()) {
changeMemberState_forTest(MemberState::RS_SECONDARY);
}
if (primaryIndex != -1) {
ReplSetHeartbeatResponse hbResponse;
hbResponse.setState(MemberState::RS_PRIMARY);
hbResponse.setElectionTime(electionTime);
hbResponse.setAppliedOpTimeAndWallTime(
{_memberData.at(primaryIndex).getHeartbeatAppliedOpTime(), Date_t() + Seconds(1)});
hbResponse.setSyncingTo(HostAndPort());
_memberData.at(primaryIndex)
.setUpValues(_memberData.at(primaryIndex).getLastHeartbeat(),
std::move(hbResponse));
}
_currentPrimaryIndex = primaryIndex;
}
}
const MemberConfig* TopologyCoordinator::_currentPrimaryMember() const {
if (_currentPrimaryIndex == -1)
return nullptr;
return &(_rsConfig.getMemberAt(_currentPrimaryIndex));
}
std::string TopologyCoordinator::_getReplSetStatusString() {
// Construct a ReplSetStatusArgs using default parameters. Missing parameters will not be
// included in the status string.
ReplSetStatusArgs rsStatusArgs{
Date_t::now(), 0U, OpTimeAndWallTime(), BSONObj(), BSONObj(), BSONObj(), boost::none};
BSONObjBuilder builder;
Status result(ErrorCodes::InternalError, "didn't set status in prepareStatusResponse");
prepareStatusResponse(rsStatusArgs, &builder, &result);
if (!result.isOK()) {
return str::stream() << "Could not get replSetGetStatus output: " << result.toString();
}
return str::stream() << "Current replSetGetStatus output: " << builder.done().toString();
}
void TopologyCoordinator::prepareStatusResponse(const ReplSetStatusArgs& rsStatusArgs,
BSONObjBuilder* response,
Status* result) {
// output for each member
std::vector<BSONObj> membersOut;
const MemberState myState = getMemberState();
const Date_t now = rsStatusArgs.now;
const OpTime lastOpApplied = getMyLastAppliedOpTime();
const Date_t lastOpAppliedWall = getMyLastAppliedOpTimeAndWallTime().wallTime;
const OpTime lastOpDurable = getMyLastDurableOpTime();
const Date_t lastOpDurableWall = getMyLastDurableOpTimeAndWallTime().wallTime;
const BSONObj& initialSyncStatus = rsStatusArgs.initialSyncStatus;
const BSONObj& electionCandidateMetrics = rsStatusArgs.electionCandidateMetrics;
const BSONObj& electionParticipantMetrics = rsStatusArgs.electionParticipantMetrics;
const boost::optional<Timestamp>& lastStableRecoveryTimestamp =
rsStatusArgs.lastStableRecoveryTimestamp;
if (_selfIndex == -1) {
// We're REMOVED or have an invalid config
response->append("state", static_cast<int>(myState.s));
response->append("stateStr", myState.toString());
response->append("uptime", static_cast<int>(rsStatusArgs.selfUptime));
appendOpTime(response, "optime", lastOpApplied);
response->appendDate("optimeDate",
Date_t::fromDurationSinceEpoch(Seconds(lastOpApplied.getSecs())));
if (_maintenanceModeCalls) {
response->append("maintenanceMode", _maintenanceModeCalls);
}
response->append("lastHeartbeatMessage", "");
response->append("syncingTo", "");
response->append("syncSourceHost", "");
response->append("syncSourceId", -1);
response->append("infoMessage", _getHbmsg(now));
*result = Status(ErrorCodes::InvalidReplicaSetConfig,
"Our replica set config is invalid or we are not a member of it");
return;
}
for (std::vector<MemberData>::const_iterator it = _memberData.begin(); it != _memberData.end();
++it) {
const int itIndex = indexOfIterator(_memberData, it);
if (itIndex == _selfIndex) {
// add self
BSONObjBuilder bb;
bb.append("_id", _selfConfig().getId().getData());
bb.append("name", _selfConfig().getHostAndPort().toString());
appendIP(&bb, "ip", _selfConfig().getHostAndPort());
bb.append("health", 1.0);
bb.append("state", static_cast<int>(myState.s));
bb.append("stateStr", myState.toString());
bb.append("uptime", static_cast<int>(rsStatusArgs.selfUptime));
if (!_selfConfig().isArbiter()) {
appendOpTime(&bb, "optime", lastOpApplied);
bb.appendDate("optimeDate",
Date_t::fromDurationSinceEpoch(Seconds(lastOpApplied.getSecs())));
}
if (!_syncSource.empty() && !_iAmPrimary()) {
bb.append("syncingTo", _syncSource.toString());
bb.append("syncSourceHost", _syncSource.toString());
const MemberConfig* member = _rsConfig.findMemberByHostAndPort(_syncSource);
bb.append("syncSourceId", member ? member->getId().getData() : -1);
} else {
bb.append("syncingTo", "");
bb.append("syncSourceHost", "");
bb.append("syncSourceId", -1);
}
if (_maintenanceModeCalls) {
bb.append("maintenanceMode", _maintenanceModeCalls);
}
bb.append("infoMessage", _getHbmsg(now));
if (myState.primary()) {
bb.append("electionTime", _electionTime);
bb.appendDate("electionDate",
Date_t::fromDurationSinceEpoch(Seconds(_electionTime.getSecs())));
}
bb.appendIntOrLL("configVersion", _rsConfig.getConfigVersion());
bb.append("self", true);
bb.append("lastHeartbeatMessage", "");
membersOut.push_back(bb.obj());
} else {
// add non-self member
const MemberConfig& itConfig = _rsConfig.getMemberAt(itIndex);
BSONObjBuilder bb;
bb.append("_id", itConfig.getId().getData());
bb.append("name", itConfig.getHostAndPort().toString());
appendIP(&bb, "ip", itConfig.getHostAndPort());
double h = it->getHealth();
bb.append("health", h);
const MemberState state = it->getState();
bb.append("state", static_cast<int>(state.s));
if (h == 0) {
// if we can't connect the state info is from the past
// and could be confusing to show
bb.append("stateStr", "(not reachable/healthy)");
} else {
bb.append("stateStr", it->getState().toString());
}
const int uptime = static_cast<int>((
it->getUpSince() != Date_t() ? durationCount<Seconds>(now - it->getUpSince()) : 0));
bb.append("uptime", uptime);
if (!itConfig.isArbiter()) {
appendOpTime(&bb, "optime", it->getHeartbeatAppliedOpTime());
appendOpTime(&bb, "optimeDurable", it->getHeartbeatDurableOpTime());
bb.appendDate("optimeDate",
Date_t::fromDurationSinceEpoch(
Seconds(it->getHeartbeatAppliedOpTime().getSecs())));
bb.appendDate("optimeDurableDate",
Date_t::fromDurationSinceEpoch(
Seconds(it->getHeartbeatDurableOpTime().getSecs())));
}
bb.appendDate("lastHeartbeat", it->getLastHeartbeat());
bb.appendDate("lastHeartbeatRecv", it->getLastHeartbeatRecv());
Milliseconds ping = _getPing(itConfig.getHostAndPort());
bb.append("pingMs", durationCount<Milliseconds>(ping));
bb.append("lastHeartbeatMessage", it->getLastHeartbeatMsg());
if (it->hasAuthIssue()) {
bb.append("authenticated", false);
}
const HostAndPort& syncSource = it->getSyncSource();
if (!syncSource.empty() && !state.primary()) {
bb.append("syncingTo", syncSource.toString());
bb.append("syncSourceHost", syncSource.toString());
const MemberConfig* member = _rsConfig.findMemberByHostAndPort(syncSource);
bb.append("syncSourceId", member ? member->getId().getData() : -1);
} else {
bb.append("syncingTo", "");
bb.append("syncSourceHost", "");
bb.append("syncSourceId", -1);
}
bb.append("infoMessage", "");
if (state == MemberState::RS_PRIMARY) {
bb.append("electionTime", it->getElectionTime());
bb.appendDate(
"electionDate",
Date_t::fromDurationSinceEpoch(Seconds(it->getElectionTime().getSecs())));
}
bb.appendIntOrLL("configVersion", it->getConfigVersion());
membersOut.push_back(bb.obj());
}
}
// sort members bson
sort(membersOut.begin(), membersOut.end(), SimpleBSONObjComparator::kInstance.makeLessThan());
response->append("set", _rsConfig.isInitialized() ? _rsConfig.getReplSetName() : "");
response->append("date", now);
response->append("myState", myState.s);
response->append("term", _term);
if (rsStatusArgs.tooStale) {
response->append("tooStale", true);
}
// Add sync source info
if (!_syncSource.empty() && !myState.primary() && !myState.removed()) {
response->append("syncingTo", _syncSource.toString());
response->append("syncSourceHost", _syncSource.toString());
const MemberConfig* member = _rsConfig.findMemberByHostAndPort(_syncSource);
response->append("syncSourceId", member ? member->getId().getData() : -1);
} else {
response->append("syncingTo", "");
response->append("syncSourceHost", "");
response->append("syncSourceId", -1);
}
if (_rsConfig.isConfigServer()) {
response->append("configsvr", true);
}
response->append("heartbeatIntervalMillis",
durationCount<Milliseconds>(_rsConfig.getHeartbeatInterval()));
response->append("majorityVoteCount", _rsConfig.getMajorityVoteCount());
response->append("writeMajorityCount", _rsConfig.getWriteMajority());
// New optimes, to hold them all.
BSONObjBuilder optimes;
_lastCommittedOpTimeAndWallTime.opTime.append(&optimes, "lastCommittedOpTime");
optimes.appendDate("lastCommittedWallTime", _lastCommittedOpTimeAndWallTime.wallTime);
if (!rsStatusArgs.readConcernMajorityOpTime.opTime.isNull()) {
rsStatusArgs.readConcernMajorityOpTime.opTime.append(&optimes, "readConcernMajorityOpTime");
optimes.appendDate("readConcernMajorityWallTime",
rsStatusArgs.readConcernMajorityOpTime.wallTime);
}
appendOpTime(&optimes, "appliedOpTime", lastOpApplied);
appendOpTime(&optimes, "durableOpTime", lastOpDurable);
optimes.appendDate("lastAppliedWallTime", lastOpAppliedWall);
optimes.appendDate("lastDurableWallTime", lastOpDurableWall);
response->append("optimes", optimes.obj());
if (lastStableRecoveryTimestamp) {
// Only include this field if the storage engine supports RTT.
response->append("lastStableRecoveryTimestamp", *lastStableRecoveryTimestamp);
}
if (!initialSyncStatus.isEmpty()) {
response->append("initialSyncStatus", initialSyncStatus);
}
if (!electionCandidateMetrics.isEmpty()) {
response->append("electionCandidateMetrics", electionCandidateMetrics);
}
if (!electionParticipantMetrics.isEmpty()) {
response->append("electionParticipantMetrics", electionParticipantMetrics);
}
response->append("members", membersOut);
*result = Status::OK();
}
StatusWith<BSONObj> TopologyCoordinator::prepareReplSetUpdatePositionCommand(
OpTime currentCommittedSnapshotOpTime) const {
BSONObjBuilder cmdBuilder;
invariant(_rsConfig.isInitialized());
// Do not send updates if we have been removed from the config.
if (_selfIndex == -1) {
return Status(ErrorCodes::NodeNotFound,
"This node is not in the current replset configuration.");
}
cmdBuilder.append(UpdatePositionArgs::kCommandFieldName, 1);
// Create an array containing objects each live member connected to us and for ourself.
BSONArrayBuilder arrayBuilder(cmdBuilder.subarrayStart("optimes"));
for (const auto& memberData : _memberData) {
if (memberData.getLastAppliedOpTime().isNull()) {
// Don't include info on members we haven't heard from yet.
continue;
}
// Don't include members we think are down.
if (!memberData.isSelf() && memberData.lastUpdateStale()) {
continue;
}
BSONObjBuilder entry(arrayBuilder.subobjStart());
memberData.getLastDurableOpTime().append(&entry,
UpdatePositionArgs::kDurableOpTimeFieldName);
entry.appendDate(UpdatePositionArgs::kDurableWallTimeFieldName,
memberData.getLastDurableWallTime());
memberData.getLastAppliedOpTime().append(&entry,
UpdatePositionArgs::kAppliedOpTimeFieldName);
entry.appendDate(UpdatePositionArgs::kAppliedWallTimeFieldName,
memberData.getLastAppliedWallTime());
entry.append(UpdatePositionArgs::kMemberIdFieldName, memberData.getMemberId().getData());
entry.append(UpdatePositionArgs::kConfigVersionFieldName, _rsConfig.getConfigVersion());
}
arrayBuilder.done();
// Add metadata to command
prepareReplSetMetadata(currentCommittedSnapshotOpTime)
.writeToMetadata(&cmdBuilder)
.transitional_ignore();
return cmdBuilder.obj();
}
void TopologyCoordinator::fillMemberData(BSONObjBuilder* result) {
BSONArrayBuilder replicationProgress(result->subarrayStart("replicationProgress"));
{
for (const auto& memberData : _memberData) {
BSONObjBuilder entry(replicationProgress.subobjStart());
entry.append("host", memberData.getHostAndPort().toString());
const auto lastDurableOpTime = memberData.getLastDurableOpTime();
entry.append("optime", lastDurableOpTime.toBSON());
const auto lastAppliedOpTime = memberData.getLastAppliedOpTime();
entry.append("lastAppliedOpTime", lastAppliedOpTime.toBSON());
const auto heartbeatAppliedOpTime = memberData.getHeartbeatAppliedOpTime();
entry.append("heartbeatAppliedOpTime", heartbeatAppliedOpTime.toBSON());
const auto heartbeatDurableOpTime = memberData.getHeartbeatDurableOpTime();
entry.append("heartbeatDurableOpTime", heartbeatDurableOpTime.toBSON());
if (_selfIndex >= 0) {
entry.append("memberId", memberData.getMemberId().getData());
}
}
}
}
void TopologyCoordinator::fillIsMasterForReplSet(IsMasterResponse* const response,
const SplitHorizon::Parameters& horizonParams) {
const MemberState myState = getMemberState();
if (!_rsConfig.isInitialized()) {
response->markAsNoConfig();
return;
}
response->setReplSetName(_rsConfig.getReplSetName());
if (myState.removed()) {
response->markAsNoConfig();
return;
}
invariant(!_rsConfig.members().empty());
const auto& self = _rsConfig.members()[_selfIndex];
const auto horizon = self.determineHorizon(horizonParams);
for (const auto& member : _rsConfig.members()) {
if (member.isHidden() || member.getSlaveDelay() > Seconds{0}) {
continue;
}
auto hostView = member.getHostAndPort(horizon);
if (member.isElectable()) {
response->addHost(std::move(hostView));
} else if (member.isArbiter()) {
response->addArbiter(std::move(hostView));
} else {
response->addPassive(std::move(hostView));
}
}
response->setReplSetVersion(_rsConfig.getConfigVersion());
// "ismaster" is false if we are not primary. If we're stepping down, we're waiting for the
// Replication State Transition Lock before we can change to secondary, but we should report
// "ismaster" false to indicate that we can't accept new writes.
response->setIsMaster(myState.primary() && !isSteppingDown());
response->setIsSecondary(myState.secondary());
const MemberConfig* curPrimary = _currentPrimaryMember();
if (curPrimary) {
response->setPrimary(curPrimary->getHostAndPort(horizon));
}
const MemberConfig& selfConfig = _rsConfig.getMemberAt(_selfIndex);
if (selfConfig.isArbiter()) {
response->setIsArbiterOnly(true);
} else if (selfConfig.getPriority() == 0) {
response->setIsPassive(true);
}
if (selfConfig.getSlaveDelay() > Seconds(0)) {
response->setSlaveDelay(selfConfig.getSlaveDelay());
}
if (selfConfig.isHidden()) {
response->setIsHidden(true);
}
if (!selfConfig.shouldBuildIndexes()) {
response->setShouldBuildIndexes(false);
}
const ReplSetTagConfig tagConfig = _rsConfig.getTagConfig();
if (selfConfig.hasTags(tagConfig)) {
for (MemberConfig::TagIterator tag = selfConfig.tagsBegin(); tag != selfConfig.tagsEnd();
++tag) {
std::string tagKey = tagConfig.getTagKey(*tag);
if (tagKey[0] == '$') {
// Filter out internal tags
continue;
}
response->addTag(tagKey, tagConfig.getTagValue(*tag));
}
}
response->setMe(selfConfig.getHostAndPort());
if (_iAmPrimary()) {
response->setElectionId(_electionId);
}
}
StatusWith<TopologyCoordinator::PrepareFreezeResponseResult>
TopologyCoordinator::prepareFreezeResponse(Date_t now, int secs, BSONObjBuilder* response) {
if (_role != TopologyCoordinator::Role::kFollower) {
std::string msg = str::stream()
<< "cannot freeze node when primary or running for election. state: "
<< (_role == TopologyCoordinator::Role::kLeader ? "Primary" : "Running-Election");
log() << msg;
return Status(ErrorCodes::NotSecondary, msg);
}
if (secs == 0) {
_stepDownUntil = now;
log() << "'unfreezing'";
response->append("info", "unfreezing");
return PrepareFreezeResponseResult::kSingleNodeSelfElect;
} else {
if (secs == 1)
response->append("warning", "you really want to freeze for only 1 second?");
_stepDownUntil = std::max(_stepDownUntil, now + Seconds(secs));
log() << "'freezing' for " << secs << " seconds";
}
return PrepareFreezeResponseResult::kNoAction;
}
Timestamp TopologyCoordinator::getElectionTime() const {
return _electionTime;
}
OID TopologyCoordinator::getElectionId() const {
return _electionId;
}
int TopologyCoordinator::getCurrentPrimaryIndex() const {
return _currentPrimaryIndex;
}
Date_t TopologyCoordinator::getStepDownTime() const {
return _stepDownUntil;
}
void TopologyCoordinator::_updateHeartbeatDataForReconfig(const ReplSetConfig& newConfig,
int selfIndex,
Date_t now) {
std::vector<MemberData> oldHeartbeats;
_memberData.swap(oldHeartbeats);
int index = 0;
for (ReplSetConfig::MemberIterator it = newConfig.membersBegin(); it != newConfig.membersEnd();
++it, ++index) {
const MemberConfig& newMemberConfig = *it;
MemberData newHeartbeatData;
for (auto&& oldMemberData : oldHeartbeats) {
if ((oldMemberData.getMemberId() == newMemberConfig.getId() &&
oldMemberData.getHostAndPort() == newMemberConfig.getHostAndPort()) ||
(index == selfIndex && oldMemberData.isSelf())) {
// This member existed in the old config with the same member ID and
// HostAndPort, so copy its heartbeat data over.
newHeartbeatData = oldMemberData;
break;
}
}
newHeartbeatData.setConfigIndex(index);
newHeartbeatData.setIsSelf(index == selfIndex);
newHeartbeatData.setHostAndPort(newMemberConfig.getHostAndPort());
newHeartbeatData.setMemberId(newMemberConfig.getId());
_memberData.push_back(newHeartbeatData);
}
if (selfIndex < 0) {
// It's necessary to have self member data even if self isn't in the configuration.
// We don't need data for the other nodes (which no longer know about us, or soon won't)
_memberData.clear();
// We're not in the config, we can't sync any more.
_syncSource = HostAndPort();
MemberData newHeartbeatData;
for (auto&& oldMemberData : oldHeartbeats) {
if (oldMemberData.isSelf()) {
newHeartbeatData = oldMemberData;
break;
}
}
newHeartbeatData.setConfigIndex(-1);
newHeartbeatData.setIsSelf(true);
_memberData.push_back(newHeartbeatData);
}
}
// This function installs a new config object and recreates MemberData objects
// that reflect the new config.
void TopologyCoordinator::updateConfig(const ReplSetConfig& newConfig, int selfIndex, Date_t now) {
invariant(_role != Role::kCandidate);
invariant(selfIndex < newConfig.getNumMembers());
// Reset term on startup.
if (!_rsConfig.isInitialized()) {
_term = OpTime::kInitialTerm;
LOG(1) << "Updated term in topology coordinator to " << _term << " due to new config";
}
_updateHeartbeatDataForReconfig(newConfig, selfIndex, now);
_rsConfig = newConfig;
_selfIndex = selfIndex;
_forceSyncSourceIndex = -1;
if (_role == Role::kLeader) {
if (_selfIndex == -1) {
log() << "Could not remain primary because no longer a member of the replica set";
} else if (!_selfConfig().isElectable()) {
log() << " Could not remain primary because no longer electable";
} else {
// Don't stepdown if you don't have to.
_currentPrimaryIndex = _selfIndex;
return;
}
_role = Role::kFollower;
_setLeaderMode(LeaderMode::kNotLeader);
}
// By this point we know we are in Role::kFollower
_currentPrimaryIndex = -1; // force secondaries to re-detect who the primary is
if (_isElectableNodeInSingleNodeReplicaSet()) {
// If the new config describes a one-node replica set, we're the one member,
// we're electable, we're not in maintenance mode and we are currently in followerMode
// SECONDARY, we must transition to candidate, in leiu of heartbeats.
_role = Role::kCandidate;
}
}
std::string TopologyCoordinator::_getHbmsg(Date_t now) const {
// ignore messages over 2 minutes old
if ((now - _hbmsgTime) > Seconds{120}) {
return "";
}
return _hbmsg;
}
void TopologyCoordinator::setMyHeartbeatMessage(const Date_t now, const std::string& message) {
_hbmsgTime = now;
_hbmsg = message;
}
const MemberConfig& TopologyCoordinator::_selfConfig() const {
return _rsConfig.getMemberAt(_selfIndex);
}
const MemberData& TopologyCoordinator::_selfMemberData() const {
return _memberData[_selfMemberDataIndex()];
}
MemberData& TopologyCoordinator::_selfMemberData() {
return _memberData[_selfMemberDataIndex()];
}
const int TopologyCoordinator::_selfMemberDataIndex() const {
invariant(!_memberData.empty());
if (_selfIndex >= 0)
return _selfIndex;
// If there is no config or we're not in the config, the first-and-only entry should be for
// self.
return 0;
}
TopologyCoordinator::UnelectableReasonMask TopologyCoordinator::_getUnelectableReason(
int index) const {
invariant(index != _selfIndex);
const MemberConfig& memberConfig = _rsConfig.getMemberAt(index);
const MemberData& hbData = _memberData.at(index);
UnelectableReasonMask result = None;
if (memberConfig.isArbiter()) {
result |= ArbiterIAm;
}
if (memberConfig.getPriority() <= 0) {
result |= NoPriority;
}
if (hbData.getState() != MemberState::RS_SECONDARY) {
result |= NotSecondary;
}
invariant(result || memberConfig.isElectable());
return result;
}
TopologyCoordinator::UnelectableReasonMask TopologyCoordinator::_getMyUnelectableReason(
const Date_t now, StartElectionReasonEnum reason) const {
UnelectableReasonMask result = None;
const OpTime lastApplied = getMyLastAppliedOpTime();
if (lastApplied.isNull()) {
result |= NoData;
}
if (!_aMajoritySeemsToBeUp()) {
result |= CannotSeeMajority;
}
if (_selfIndex == -1) {
result |= NotInitialized;
return result;
}
if (_selfConfig().isArbiter()) {
result |= ArbiterIAm;
}
if (_selfConfig().getPriority() <= 0) {
result |= NoPriority;
}
if (_stepDownUntil > now) {
result |= StepDownPeriodActive;
}
// Cannot be electable unless secondary or already primary
if (!getMemberState().secondary() && !_iAmPrimary()) {
result |= NotSecondary;
}
if (reason == StartElectionReasonEnum::kPriorityTakeover &&
!_amIFreshEnoughForPriorityTakeover()) {
result |= NotCloseEnoughToLatestForPriorityTakeover;
}
if (reason == StartElectionReasonEnum::kCatchupTakeover &&
!_amIFreshEnoughForCatchupTakeover()) {
result |= NotFreshEnoughForCatchupTakeover;
}
return result;
}
std::string TopologyCoordinator::_getUnelectableReasonString(const UnelectableReasonMask ur) const {
invariant(ur);
str::stream ss;
bool hasWrittenToStream = false;
if (ur & NoData) {
ss << "node has no applied oplog entries";
hasWrittenToStream = true;
}
if (ur & CannotSeeMajority) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "I cannot see a majority";
}
if (ur & ArbiterIAm) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "member is an arbiter";
}
if (ur & NoPriority) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "member has zero priority";
}
if (ur & StepDownPeriodActive) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "I am still waiting for stepdown period to end at "
<< dateToISOStringLocal(_stepDownUntil);
}
if (ur & NotSecondary) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "member is not currently a secondary";
}
if (ur & NotCloseEnoughToLatestForPriorityTakeover) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "member is not caught up enough to the most up-to-date member to call for priority "
"takeover - must be within "
<< gPriorityTakeoverFreshnessWindowSeconds << " seconds";
}
if (ur & NotFreshEnoughForCatchupTakeover) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "member is either not the most up-to-date member or not ahead of the primary, and "
"therefore cannot call for catchup takeover";
}
if (ur & NotInitialized) {
if (hasWrittenToStream) {
ss << "; ";
}
hasWrittenToStream = true;
ss << "node is not a member of a valid replica set configuration";
}
if (!hasWrittenToStream) {
severe() << "Invalid UnelectableReasonMask value 0x" << integerToHex(ur);
fassertFailed(26011);
}
ss << " (mask 0x" << integerToHex(ur) << ")";
return ss;
}
Milliseconds TopologyCoordinator::_getPing(const HostAndPort& host) {
return _pings[host].getMillis();
}
void TopologyCoordinator::_setElectionTime(const Timestamp& newElectionTime) {
_electionTime = newElectionTime;
}
int TopologyCoordinator::_getTotalPings() {
PingMap::iterator it = _pings.begin();
PingMap::iterator end = _pings.end();
int totalPings = 0;
while (it != end) {
totalPings += it->second.getCount();
it++;
}
return totalPings;
}
bool TopologyCoordinator::isSteppingDownUnconditionally() const {
return _leaderMode == LeaderMode::kSteppingDown;
}
bool TopologyCoordinator::isSteppingDown() const {
return isSteppingDownUnconditionally() || _leaderMode == LeaderMode::kAttemptingStepDown;
}
void TopologyCoordinator::_setLeaderMode(TopologyCoordinator::LeaderMode newMode) {
// Invariants for valid state transitions.
switch (_leaderMode) {
case LeaderMode::kNotLeader:
invariant(newMode == LeaderMode::kLeaderElect);
break;
case LeaderMode::kLeaderElect:
invariant(newMode == LeaderMode::kNotLeader || // TODO(SERVER-30852): remove this case
newMode == LeaderMode::kMaster ||
newMode == LeaderMode::kAttemptingStepDown ||
newMode == LeaderMode::kSteppingDown);
break;
case LeaderMode::kMaster:
invariant(newMode == LeaderMode::kNotLeader || // TODO(SERVER-30852): remove this case
newMode == LeaderMode::kAttemptingStepDown ||
newMode == LeaderMode::kSteppingDown);
break;
case LeaderMode::kAttemptingStepDown:
invariant(newMode == LeaderMode::kNotLeader || newMode == LeaderMode::kMaster ||
newMode == LeaderMode::kSteppingDown || newMode == LeaderMode::kLeaderElect);
break;
case LeaderMode::kSteppingDown:
invariant(newMode == LeaderMode::kNotLeader);
break;
}
_leaderMode = std::move(newMode);
}
MemberState TopologyCoordinator::getMemberState() const {
if (_selfIndex == -1) {
if (_rsConfig.isInitialized()) {
return MemberState::RS_REMOVED;
}
return MemberState::RS_STARTUP;
}
if (_rsConfig.isConfigServer()) {
if (_options.clusterRole != ClusterRole::ConfigServer && !skipShardingConfigurationChecks) {
return MemberState::RS_REMOVED;
} else {
invariant(_storageEngineSupportsReadCommitted != ReadCommittedSupport::kUnknown);
if (_storageEngineSupportsReadCommitted == ReadCommittedSupport::kNo) {
return MemberState::RS_REMOVED;
}
}
} else {
if (_options.clusterRole == ClusterRole::ConfigServer && !skipShardingConfigurationChecks) {
return MemberState::RS_REMOVED;
}
}
if (_role == Role::kLeader) {
invariant(_currentPrimaryIndex == _selfIndex);
invariant(_leaderMode != LeaderMode::kNotLeader);
return MemberState::RS_PRIMARY;
}
const MemberConfig& myConfig = _selfConfig();
if (myConfig.isArbiter()) {
return MemberState::RS_ARBITER;
}
if (((_maintenanceModeCalls > 0) || (_hasOnlyAuthErrorUpHeartbeats(_memberData, _selfIndex))) &&
(_followerMode == MemberState::RS_SECONDARY)) {
return MemberState::RS_RECOVERING;
}
return _followerMode;
}
std::vector<MemberData> TopologyCoordinator::getMemberData() const {
return _memberData;
}
bool TopologyCoordinator::canAcceptWrites() const {
return _leaderMode == LeaderMode::kMaster;
}
void TopologyCoordinator::setElectionInfo(OID electionId, Timestamp electionOpTime) {
invariant(_role == Role::kLeader);
_electionTime = electionOpTime;
_electionId = electionId;
}
void TopologyCoordinator::processWinElection(OID electionId, Timestamp electionOpTime) {
invariant(_role == Role::kCandidate);
invariant(_leaderMode == LeaderMode::kNotLeader);
_role = Role::kLeader;
_setLeaderMode(LeaderMode::kLeaderElect);
setElectionInfo(electionId, electionOpTime);
_currentPrimaryIndex = _selfIndex;
_syncSource = HostAndPort();
_forceSyncSourceIndex = -1;
// Prevent last committed optime from updating until we finish draining.
_firstOpTimeOfMyTerm =
OpTime(Timestamp(std::numeric_limits<int>::max(), 0), std::numeric_limits<int>::max());
}
void TopologyCoordinator::processLoseElection() {
invariant(_role == Role::kCandidate);
invariant(_leaderMode == LeaderMode::kNotLeader);
const HostAndPort syncSourceAddress = getSyncSourceAddress();
_electionTime = Timestamp(0, 0);
_electionId = OID();
_role = Role::kFollower;
}
bool TopologyCoordinator::tryToStartStepDown(
long long termAtStart, Date_t now, Date_t waitUntil, Date_t stepDownUntil, bool force) {
if (_role != Role::kLeader || _leaderMode == LeaderMode::kSteppingDown ||
_term != termAtStart) {
uasserted(ErrorCodes::PrimarySteppedDown,
"While waiting for secondaries to catch up before stepping down, "
"this node decided to step down for other reasons");
}
invariant(_leaderMode == LeaderMode::kAttemptingStepDown);
if (now >= stepDownUntil) {
uasserted(ErrorCodes::ExceededTimeLimit,
"By the time we were ready to step down, we were already past the "
"time we were supposed to step down until");
}
if (!_canCompleteStepDownAttempt(now, waitUntil, force)) {
// Stepdown attempt failed.
// Check waitUntil after at least one stepdown attempt, so that stepdown could succeed even
// if secondaryCatchUpPeriodSecs == 0.
if (now >= waitUntil) {
uasserted(ErrorCodes::ExceededTimeLimit,
str::stream() << "No electable secondaries caught up as of "
<< dateToISOStringLocal(now)
<< ". Please use the replSetStepDown command with the argument "
<< "{force: true} to force node to step down.");
}
// Stepdown attempt failed, but in a way that can be retried
return false;
}
// Stepdown attempt success!
_stepDownUntil = stepDownUntil;
prepareForUnconditionalStepDown();
return true;
}
bool TopologyCoordinator::_canCompleteStepDownAttempt(Date_t now, Date_t waitUntil, bool force) {
const bool forceNow = force && (now >= waitUntil);
if (forceNow) {
return true;
}
return isSafeToStepDown();
}
bool TopologyCoordinator::_isCaughtUpAndElectable(int memberIndex, OpTime lastApplied) {
if (_getUnelectableReason(memberIndex)) {
return false;
}
return (_memberData.at(memberIndex).getLastAppliedOpTime() >= lastApplied);
}
bool TopologyCoordinator::isSafeToStepDown() {
if (!_rsConfig.isInitialized() || _selfIndex < 0) {
return false;
}
OpTime lastApplied = getMyLastAppliedOpTime();
auto tagStatus = _rsConfig.findCustomWriteMode(ReplSetConfig::kMajorityWriteConcernModeName);
invariant(tagStatus.isOK());
// Check if a majority of nodes have reached the last applied optime.
if (!haveTaggedNodesReachedOpTime(lastApplied, tagStatus.getValue(), false)) {
return false;
}
// Now check that we also have at least one caught up node that is electable.
for (int memberIndex = 0; memberIndex < _rsConfig.getNumMembers(); memberIndex++) {
// ignore your self
if (memberIndex == _selfIndex) {
continue;
}
if (_isCaughtUpAndElectable(memberIndex, lastApplied)) {
return true;
}
}
return false;
}
int TopologyCoordinator::chooseElectionHandoffCandidate() {
OpTime lastApplied = getMyLastAppliedOpTime();
int bestCandidateIndex = -1;
int highestPriority = -1;
for (int memberIndex = 0; memberIndex < _rsConfig.getNumMembers(); memberIndex++) {
// Skip your own member index.
if (memberIndex == _selfIndex) {
continue;
}
// Skip this node if it is not eligible to become primary. This includes nodes with
// priority 0.
if (!_isCaughtUpAndElectable(memberIndex, lastApplied)) {
continue;
}
// Only update best if priority is strictly greater. This guarantees that
// we will pick the member with the lowest index in case of a tie. Note that
// member priority is always a non-negative number.
auto memberPriority = _rsConfig.getMemberAt(memberIndex).getPriority();
if (memberPriority > highestPriority) {
bestCandidateIndex = memberIndex;
highestPriority = memberPriority;
}
}
// This is the most suitable node.
return bestCandidateIndex;
}
void TopologyCoordinator::setFollowerMode(MemberState::MS newMode) {
invariant(_role == Role::kFollower);
switch (newMode) {
case MemberState::RS_RECOVERING:
case MemberState::RS_ROLLBACK:
case MemberState::RS_SECONDARY:
case MemberState::RS_STARTUP2:
_followerMode = newMode;
break;
default:
MONGO_UNREACHABLE;
}
if (_followerMode != MemberState::RS_SECONDARY) {
return;
}
// When a single node replica set transitions to SECONDARY, we must check if we should
// be a candidate here. This is necessary because a single node replica set has no
// heartbeats that would normally change the role to candidate.
if (_isElectableNodeInSingleNodeReplicaSet()) {
_role = Role::kCandidate;
}
}
bool TopologyCoordinator::_isElectableNodeInSingleNodeReplicaSet() const {
return _followerMode == MemberState::RS_SECONDARY && _rsConfig.getNumMembers() == 1 &&
_selfIndex == 0 && _rsConfig.getMemberAt(_selfIndex).isElectable() &&
_maintenanceModeCalls == 0;
}
void TopologyCoordinator::finishUnconditionalStepDown() {
invariant(_leaderMode == LeaderMode::kSteppingDown);
int remotePrimaryIndex = -1;
for (std::vector<MemberData>::const_iterator it = _memberData.begin(); it != _memberData.end();
++it) {
const int itIndex = indexOfIterator(_memberData, it);
if (itIndex == _selfIndex) {
continue;
}
if (it->getState().primary() && it->up()) {
if (remotePrimaryIndex != -1) {
// two other nodes think they are primary (asynchronously polled)
// -- wait for things to settle down.
remotePrimaryIndex = -1;
warning() << "two remote primaries (transiently)";
break;
}
remotePrimaryIndex = itIndex;
}
}
_stepDownSelfAndReplaceWith(remotePrimaryIndex);
}
void TopologyCoordinator::_stepDownSelfAndReplaceWith(int newPrimary) {
invariant(_role == Role::kLeader);
invariant(_selfIndex != -1);
invariant(_selfIndex != newPrimary);
invariant(_selfIndex == _currentPrimaryIndex);
_currentPrimaryIndex = newPrimary;
_role = Role::kFollower;
_setLeaderMode(LeaderMode::kNotLeader);
}
bool TopologyCoordinator::updateLastCommittedOpTimeAndWallTime() {
// If we're not primary or we're stepping down due to learning of a new term then we must not
// advance the commit point. If we are stepping down due to a user request, however, then it
// is safe to advance the commit point, and in fact we must since the stepdown request may be
// waiting for the commit point to advance enough to be able to safely complete the step down.
if (!_iAmPrimary() || _leaderMode == LeaderMode::kSteppingDown) {
return false;
}
// Whether we use the applied or durable OpTime for the commit point is decided here.
const bool useDurableOpTime = _rsConfig.getWriteConcernMajorityShouldJournal();
std::vector<OpTimeAndWallTime> votingNodesOpTimesAndWallTimes;
for (const auto& memberData : _memberData) {
int memberIndex = memberData.getConfigIndex();
invariant(memberIndex >= 0);
const auto& memberConfig = _rsConfig.getMemberAt(memberIndex);
if (memberConfig.isVoter()) {
const OpTimeAndWallTime durableOpTime = {memberData.getLastDurableOpTime(),
memberData.getLastDurableWallTime()};
const OpTimeAndWallTime appliedOpTime = {memberData.getLastAppliedOpTime(),
memberData.getLastAppliedWallTime()};
const OpTimeAndWallTime opTime = useDurableOpTime ? durableOpTime : appliedOpTime;
votingNodesOpTimesAndWallTimes.push_back(opTime);
}
}
invariant(votingNodesOpTimesAndWallTimes.size() > 0);
if (votingNodesOpTimesAndWallTimes.size() <
static_cast<unsigned long>(_rsConfig.getWriteMajority())) {
return false;
}
std::sort(votingNodesOpTimesAndWallTimes.begin(), votingNodesOpTimesAndWallTimes.end());
// need the majority to have this OpTime
OpTimeAndWallTime committedOpTime =
votingNodesOpTimesAndWallTimes[votingNodesOpTimesAndWallTimes.size() -
_rsConfig.getWriteMajority()];
const bool fromSyncSource = false;
return advanceLastCommittedOpTimeAndWallTime(committedOpTime, fromSyncSource);
}
bool TopologyCoordinator::advanceLastCommittedOpTimeAndWallTime(OpTimeAndWallTime committedOpTime,
bool fromSyncSource) {
if (_selfIndex == -1) {
// The config hasn't been installed or we are not in the config. This could happen
// on heartbeats before installing a config.
return false;
}
// This check is performed to ensure primaries do not commit an OpTime from a previous term.
if (_iAmPrimary() && committedOpTime.opTime < _firstOpTimeOfMyTerm) {
LOG(1) << "Ignoring older committed snapshot from before I became primary, optime: "
<< committedOpTime.opTime << ", firstOpTimeOfMyTerm: " << _firstOpTimeOfMyTerm;
return false;
}
// Arbiters don't have data so they always advance their commit point via heartbeats.
if (!_selfConfig().isArbiter() &&
getMyLastAppliedOpTime().getTerm() != committedOpTime.opTime.getTerm()) {
if (fromSyncSource) {
committedOpTime = std::min(committedOpTime, getMyLastAppliedOpTimeAndWallTime());
} else {
LOG(1) << "Ignoring commit point with different term than my lastApplied, since it "
"may "
"not be on the same oplog branch as mine. optime: "
<< committedOpTime
<< ", my last applied: " << getMyLastAppliedOpTimeAndWallTime();
return false;
}
}
if (committedOpTime.opTime == _lastCommittedOpTimeAndWallTime.opTime) {
return false; // Hasn't changed, so ignore it.
}
if (committedOpTime.opTime < _lastCommittedOpTimeAndWallTime.opTime) {
LOG(1) << "Ignoring older committed snapshot optime: " << committedOpTime
<< ", currentCommittedOpTime: " << _lastCommittedOpTimeAndWallTime;
return false;
}
LOG(2) << "Updating _lastCommittedOpTimeAndWallTime to " << committedOpTime;
_lastCommittedOpTimeAndWallTime = committedOpTime;
return true;
}
OpTime TopologyCoordinator::getLastCommittedOpTime() const {
return _lastCommittedOpTimeAndWallTime.opTime;
}
OpTimeAndWallTime TopologyCoordinator::getLastCommittedOpTimeAndWallTime() const {
return _lastCommittedOpTimeAndWallTime;
}
bool TopologyCoordinator::canCompleteTransitionToPrimary(long long termWhenDrainCompleted) const {
if (termWhenDrainCompleted != _term) {
return false;
}
// Allow completing the transition to primary even when in the middle of a stepdown attempt,
// in case the stepdown attempt fails.
if (_leaderMode != LeaderMode::kLeaderElect && _leaderMode != LeaderMode::kAttemptingStepDown) {
return false;
}
return true;
}
Status TopologyCoordinator::completeTransitionToPrimary(const OpTime& firstOpTimeOfTerm) {
if (!canCompleteTransitionToPrimary(firstOpTimeOfTerm.getTerm())) {
return Status(ErrorCodes::PrimarySteppedDown,
"By the time this node was ready to complete its transition to PRIMARY it "
"was no longer eligible to do so");
}
if (_leaderMode == LeaderMode::kLeaderElect) {
_setLeaderMode(LeaderMode::kMaster);
}
_firstOpTimeOfMyTerm = firstOpTimeOfTerm;
return Status::OK();
}
void TopologyCoordinator::adjustMaintenanceCountBy(int inc) {
invariant(_role == Role::kFollower);
_maintenanceModeCalls += inc;
invariant(_maintenanceModeCalls >= 0);
}
int TopologyCoordinator::getMaintenanceCount() const {
return _maintenanceModeCalls;
}
TopologyCoordinator::UpdateTermResult TopologyCoordinator::updateTerm(long long term, Date_t now) {
if (term <= _term) {
return TopologyCoordinator::UpdateTermResult::kAlreadyUpToDate;
}
// Don't run election if we just stood up or learned about a new term.
_electionSleepUntil = now + _rsConfig.getElectionTimeoutPeriod();
// Don't update the term just yet if we are going to step down, as we don't want to report
// that we are primary in the new term.
if (_iAmPrimary()) {
return TopologyCoordinator::UpdateTermResult::kTriggerStepDown;
}
LOG(1) << "Updating term from " << _term << " to " << term;
_term = term;
return TopologyCoordinator::UpdateTermResult::kUpdatedTerm;
}
long long TopologyCoordinator::getTerm() const {
return _term;
}
// TODO(siyuan): Merge _hddata into _slaveInfo, so that we have a single view of the
// replset. Passing metadata is unnecessary.
bool TopologyCoordinator::shouldChangeSyncSource(
const HostAndPort& currentSource,
const rpc::ReplSetMetadata& replMetadata,
boost::optional<rpc::OplogQueryMetadata> oqMetadata,
Date_t now) const {
// Methodology:
// If there exists a viable sync source member other than currentSource, whose oplog has
// reached an optime greater than _options.maxSyncSourceLagSecs later than currentSource's,
// return true.
// If the currentSource has the same replication progress as we do and has no source for further
// progress, return true.
if (_selfIndex == -1) {
log() << "Not choosing new sync source because we are not in the config.";
return false;
}
// If the user requested a sync source change, return true.
if (_forceSyncSourceIndex != -1) {
log() << "Choosing new sync source because the user has requested to use "
<< _rsConfig.getMemberAt(_forceSyncSourceIndex).getHostAndPort()
<< " as a sync source";
return true;
}
if (replMetadata.getConfigVersion() != _rsConfig.getConfigVersion()) {
log() << "Choosing new sync source because the config version supplied by " << currentSource
<< ", " << replMetadata.getConfigVersion() << ", does not match ours, "
<< _rsConfig.getConfigVersion();
return true;
}
const int currentSourceIndex = _rsConfig.findMemberIndexByHostAndPort(currentSource);
if (currentSourceIndex == -1) {
log() << "Choosing new sync source because " << currentSource.toString()
<< " is not in our config";
return true;
}
invariant(currentSourceIndex != _selfIndex);
// If OplogQueryMetadata was provided, use its values, otherwise use the ones in
// ReplSetMetadata.
OpTime currentSourceOpTime;
int syncSourceIndex = -1;
int primaryIndex = -1;
if (oqMetadata) {
currentSourceOpTime =
std::max(oqMetadata->getLastOpApplied(),
_memberData.at(currentSourceIndex).getHeartbeatAppliedOpTime());
syncSourceIndex = oqMetadata->getSyncSourceIndex();
primaryIndex = oqMetadata->getPrimaryIndex();
} else {
currentSourceOpTime =
std::max(replMetadata.getLastOpVisible(),
_memberData.at(currentSourceIndex).getHeartbeatAppliedOpTime());
syncSourceIndex = replMetadata.getSyncSourceIndex();
primaryIndex = replMetadata.getPrimaryIndex();
}
if (currentSourceOpTime.isNull()) {
// Haven't received a heartbeat from the sync source yet, so can't tell if we should
// change.
return false;
}
// Change sync source if they are not ahead of us, and don't have a sync source,
// unless they are primary.
const OpTime myLastOpTime = getMyLastAppliedOpTime();
if (syncSourceIndex == -1 && currentSourceOpTime <= myLastOpTime &&
primaryIndex != currentSourceIndex) {
std::stringstream logMessage;
logMessage << "Choosing new sync source. Our current sync source is not primary and does "
"not have a sync source, so we require that it is ahead of us. "
<< "Current sync source: " << currentSource.toString()
<< ", my last fetched oplog optime: " << myLastOpTime
<< ", latest oplog optime of sync source: " << currentSourceOpTime;
if (primaryIndex >= 0) {
logMessage << " (" << _rsConfig.getMemberAt(primaryIndex).getHostAndPort() << " is)";
} else {
logMessage << " (sync source does not know the primary)";
}
log() << logMessage.str();
return true;
}
if (MONGO_unlikely(disableMaxSyncSourceLagSecs.shouldFail())) {
log() << "disableMaxSyncSourceLagSecs fail point enabled - not checking the most recent "
"OpTime, "
<< currentSourceOpTime.toString() << ", of our current sync source, " << currentSource
<< ", against the OpTimes of the other nodes in this replica set.";
} else {
unsigned int currentSecs = currentSourceOpTime.getSecs();
unsigned int goalSecs = currentSecs + durationCount<Seconds>(_options.maxSyncSourceLagSecs);
for (std::vector<MemberData>::const_iterator it = _memberData.begin();
it != _memberData.end();
++it) {
const int itIndex = indexOfIterator(_memberData, it);
const MemberConfig& candidateConfig = _rsConfig.getMemberAt(itIndex);
if (it->up() && (candidateConfig.isVoter() || !_selfConfig().isVoter()) &&
(candidateConfig.shouldBuildIndexes() || !_selfConfig().shouldBuildIndexes()) &&
it->getState().readable() && !_memberIsBlacklisted(candidateConfig, now) &&
goalSecs < it->getHeartbeatAppliedOpTime().getSecs()) {
log() << "Choosing new sync source because the most recent OpTime of our sync "
"source, "
<< currentSource << ", is " << currentSourceOpTime.toString()
<< " which is more than " << _options.maxSyncSourceLagSecs
<< " behind member " << candidateConfig.getHostAndPort().toString()
<< " whose most recent OpTime is "
<< it->getHeartbeatAppliedOpTime().toString();
invariant(itIndex != _selfIndex);
return true;
}
}
}
return false;
}
rpc::ReplSetMetadata TopologyCoordinator::prepareReplSetMetadata(
const OpTime& lastVisibleOpTime) const {
return rpc::ReplSetMetadata(_term,
_lastCommittedOpTimeAndWallTime,
lastVisibleOpTime,
_rsConfig.getConfigVersion(),
_rsConfig.getReplicaSetId(),
_currentPrimaryIndex,
_rsConfig.findMemberIndexByHostAndPort(getSyncSourceAddress()));
}
rpc::OplogQueryMetadata TopologyCoordinator::prepareOplogQueryMetadata(int rbid) const {
return rpc::OplogQueryMetadata(_lastCommittedOpTimeAndWallTime,
getMyLastAppliedOpTime(),
rbid,
_currentPrimaryIndex,
_rsConfig.findMemberIndexByHostAndPort(getSyncSourceAddress()));
}
void TopologyCoordinator::processReplSetRequestVotes(const ReplSetRequestVotesArgs& args,
ReplSetRequestVotesResponse* response) {
response->setTerm(_term);
if (MONGO_unlikely(voteNoInElection.shouldFail())) {
log() << "failpoint voteNoInElection enabled";
response->setVoteGranted(false);
response->setReason(str::stream() << "forced to vote no during dry run election due to "
"failpoint voteNoInElection set");
return;
}
if (MONGO_unlikely(voteYesInDryRunButNoInRealElection.shouldFail())) {
log() << "failpoint voteYesInDryRunButNoInRealElection enabled";
if (args.isADryRun()) {
response->setVoteGranted(true);
response->setReason(str::stream() << "forced to vote yes in dry run due to failpoint "
"voteYesInDryRunButNoInRealElection set");
} else {
response->setVoteGranted(false);
response->setReason(str::stream()
<< "forced to vote no in real election due to failpoint "
"voteYesInDryRunButNoInRealElection set");
}
return;
}
if (args.getTerm() < _term) {
response->setVoteGranted(false);
response->setReason(str::stream() << "candidate's term (" << args.getTerm()
<< ") is lower than mine (" << _term << ")");
} else if (args.getConfigVersion() != _rsConfig.getConfigVersion()) {
response->setVoteGranted(false);
response->setReason(str::stream()
<< "candidate's config version (" << args.getConfigVersion()
<< ") differs from mine (" << _rsConfig.getConfigVersion() << ")");
} else if (args.getSetName() != _rsConfig.getReplSetName()) {
response->setVoteGranted(false);
response->setReason(str::stream()
<< "candidate's set name (" << args.getSetName()
<< ") differs from mine (" << _rsConfig.getReplSetName() << ")");
} else if (args.getLastDurableOpTime() < getMyLastAppliedOpTime()) {
response->setVoteGranted(false);
response
->setReason(str::stream()
<< "candidate's data is staler than mine. candidate's last applied OpTime: "
<< args.getLastDurableOpTime().toString()
<< ", my last applied OpTime: " << getMyLastAppliedOpTime().toString());
} else if (!args.isADryRun() && _lastVote.getTerm() == args.getTerm()) {
response->setVoteGranted(false);
response->setReason(str::stream()
<< "already voted for another candidate ("
<< _rsConfig.getMemberAt(_lastVote.getCandidateIndex()).getHostAndPort()
<< ") this term (" << _lastVote.getTerm() << ")");
} else {
int betterPrimary = _findHealthyPrimaryOfEqualOrGreaterPriority(args.getCandidateIndex());
if (_selfConfig().isArbiter() && betterPrimary >= 0) {
response->setVoteGranted(false);
response->setReason(str::stream()
<< "can see a healthy primary ("
<< _rsConfig.getMemberAt(betterPrimary).getHostAndPort()
<< ") of equal or greater priority");
} else {
if (!args.isADryRun()) {
_lastVote.setTerm(args.getTerm());
_lastVote.setCandidateIndex(args.getCandidateIndex());
}
response->setVoteGranted(true);
}
}
LOG_FOR_ELECTION(0) << "Received vote request: " << args.toString();
LOG_FOR_ELECTION(0) << "Sending vote response: " << response->toString();
LOG_FOR_ELECTION(4) << _getReplSetStatusString();
}
void TopologyCoordinator::loadLastVote(const LastVote& lastVote) {
_lastVote = lastVote;
}
void TopologyCoordinator::voteForMyselfV1() {
_lastVote.setTerm(_term);
_lastVote.setCandidateIndex(_selfIndex);
}
void TopologyCoordinator::setPrimaryIndex(long long primaryIndex) {
_currentPrimaryIndex = primaryIndex;
}
Status TopologyCoordinator::becomeCandidateIfElectable(const Date_t now,
StartElectionReasonEnum reason) {
if (_role == Role::kLeader) {
return {ErrorCodes::NodeNotElectable, "Not standing for election again; already primary"};
}
if (_role == Role::kCandidate) {
return {ErrorCodes::NodeNotElectable, "Not standing for election again; already candidate"};
}
const UnelectableReasonMask unelectableReason = _getMyUnelectableReason(now, reason);
if (unelectableReason) {
return {ErrorCodes::NodeNotElectable,
str::stream() << "Not standing for election because "
<< _getUnelectableReasonString(unelectableReason)};
}
// All checks passed, become a candidate and start election proceedings.
_role = Role::kCandidate;
return Status::OK();
}
void TopologyCoordinator::setStorageEngineSupportsReadCommitted(bool supported) {
_storageEngineSupportsReadCommitted =
supported ? ReadCommittedSupport::kYes : ReadCommittedSupport::kNo;
}
void TopologyCoordinator::restartHeartbeats() {
for (auto& hb : _memberData) {
hb.restart();
}
}
OpTime TopologyCoordinator::latestKnownOpTime() const {
OpTime latest = getMyLastAppliedOpTime();
for (std::vector<MemberData>::const_iterator it = _memberData.begin(); it != _memberData.end();
++it) {
// Ignore self
if (it->isSelf()) {
continue;
}
// Ignore down members
if (!it->up()) {
continue;
}
// Ignore removed nodes (not in config, so not valid).
if (it->getState().removed()) {
continue;
}
OpTime optime = it->getHeartbeatAppliedOpTime();
if (optime > latest) {
latest = optime;
}
}
return latest;
}
boost::optional<OpTime> TopologyCoordinator::latestKnownOpTimeSinceHeartbeatRestart() const {
// The smallest OpTime in PV1.
OpTime latest(Timestamp(0, 0), 0);
for (size_t i = 0; i < _memberData.size(); i++) {
auto& peer = _memberData[i];
if (static_cast<int>(i) == _selfIndex) {
continue;
}
// If any heartbeat is not fresh enough, return none.
if (!peer.isUpdatedSinceRestart()) {
return boost::none;
}
// Ignore down members
if (!peer.up()) {
continue;
}
if (peer.getHeartbeatAppliedOpTime() > latest) {
latest = peer.getHeartbeatAppliedOpTime();
}
}
return latest;
}
std::map<MemberId, boost::optional<OpTime>>
TopologyCoordinator::latestKnownOpTimeSinceHeartbeatRestartPerMember() const {
std::map<MemberId, boost::optional<OpTime>> opTimesPerMember;
for (size_t i = 0; i < _memberData.size(); i++) {
auto& member = _memberData[i];
MemberId memberId = _rsConfig.getMemberAt(i).getId();
if (!member.isUpdatedSinceRestart()) {
opTimesPerMember[memberId] = boost::none;
continue;
}
if (!member.up()) {
opTimesPerMember[memberId] = boost::none;
continue;
}
opTimesPerMember[memberId] = member.getHeartbeatAppliedOpTime();
}
return opTimesPerMember;
}
bool TopologyCoordinator::checkIfCommitQuorumCanBeSatisfied(
const CommitQuorumOptions& commitQuorum, const std::vector<MemberConfig>& members) const {
if (!commitQuorum.mode.empty() && commitQuorum.mode != CommitQuorumOptions::kMajority) {
StatusWith<ReplSetTagPattern> tagPatternStatus =
_rsConfig.findCustomWriteMode(commitQuorum.mode);
if (!tagPatternStatus.isOK()) {
return false;
}
ReplSetTagMatch matcher(tagPatternStatus.getValue());
for (auto&& member : members) {
for (MemberConfig::TagIterator it = member.tagsBegin(); it != member.tagsEnd(); ++it) {
if (matcher.update(*it)) {
return true;
}
}
}
// Even if all the nodes in the set had a given write it still would not satisfy this
// commit quorum.
return false;
} else {
int nodesRemaining = 0;
if (!commitQuorum.mode.empty()) {
invariant(commitQuorum.mode == CommitQuorumOptions::kMajority);
nodesRemaining = _rsConfig.getWriteMajority();
} else {
nodesRemaining = commitQuorum.numNodes;
}
for (auto&& member : members) {
if (!member.isArbiter()) { // Only count data-bearing nodes
--nodesRemaining;
if (nodesRemaining <= 0) {
return true;
}
}
}
return false;
}
}
bool TopologyCoordinator::checkIfCommitQuorumIsSatisfied(
const CommitQuorumOptions& commitQuorum,
const std::vector<HostAndPort>& commitReadyMembers) const {
std::vector<MemberConfig> commitReadyMemberConfigs;
for (auto& commitReadyMember : commitReadyMembers) {
const MemberConfig* memberConfig = _rsConfig.findMemberByHostAndPort(commitReadyMember);
invariant(memberConfig);
commitReadyMemberConfigs.push_back(*memberConfig);
}
// Calling this with commit ready members only is the same as checking if the commit quorum is
// satisfied. Because the 'commitQuorum' is based on the participation of all the replica set
// members, and if the 'commitQuorum' can be satisfied with all the commit ready members, then
// the commit quorum is satisfied in this replica set configuration.
return checkIfCommitQuorumCanBeSatisfied(commitQuorum, commitReadyMemberConfigs);
}
} // namespace repl
} // namespace mongo
|