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
|
#!/usr/bin/env python
#
# Public Domain 2014-present MongoDB, Inc.
# Public Domain 2008-2014 WiredTiger, Inc.
#
# This is free and unencumbered software released into the public domain.
#
# Anyone is free to copy, modify, publish, use, compile, sell, or
# distribute this software, either in source code form or as a compiled
# binary, for any purpose, commercial or non-commercial, and by any
# means.
#
# In jurisdictions that recognize copyright laws, the author or authors
# of this software dedicate any and all copyright interest in the
# software to the public domain. We make this dedication for the benefit
# of the public at large and to the detriment of our heirs and
# successors. We intend this dedication to be an overt act of
# relinquishment in perpetuity of all present and future rights to this
# software under copyright law.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
# IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR
# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
# OTHER DEALINGS IN THE SOFTWARE.
import time, re
import wttest
from wtdataset import SimpleDataSet
from wiredtiger import stat
from wtscenario import make_scenarios
# test_hs21.py
# Test we don't lose any data when idle files with an active history are closed/sweeped.
# Files with active history, ie content newer than the oldest timestamp can be closed when idle.
# We want to ensure that when an active history file is idle closed we can continue reading the
# correct version of data and their base write generation hasn't changed (since we haven't
# restarted the system).
class test_hs21(wttest.WiredTigerTestCase):
# Configure handle sweeping to occur within a specific amount of time.
conn_config = 'file_manager=(close_handle_minimum=0,close_idle_time=2,close_scan_interval=1),' + \
'statistics=(all),operation_tracking=(enabled=false)'
file_name = 'test_hs21'
numfiles = 10
nrows = 1000
format_values = [
('column', dict(key_format='r', key1=1, key2=2, value_format='S')),
('column-fix', dict(key_format='r', key1=1, key2=2, value_format='8t')),
('string-row', dict(key_format='S', key1=str(0), key2=str(1), value_format='S')),
]
scenarios = make_scenarios(format_values)
def large_updates(self, uri, value, ds, nrows, commit_ts):
# Update a large number of records, we'll hang if the history store table isn't working.
session = self.session
cursor = session.open_cursor(uri)
session.begin_transaction()
for i in range(1, nrows + 1):
cursor[ds.key(i)] = value
session.commit_transaction('commit_timestamp=' + self.timestamp_str(commit_ts))
cursor.close()
def check(self, session, check_value, uri, nrows, read_ts=-1, flcs_nrows=None):
if self.value_format != '8t':
flcs_nrows = None
# Validate we read an expected value (optionally at a given read timestamp).
if read_ts != -1:
session.begin_transaction('read_timestamp=' + self.timestamp_str(read_ts))
cursor = session.open_cursor(uri)
count = 0
for k, v in cursor:
if flcs_nrows is not None and count >= nrows:
self.assertEqual(v, 0)
else:
self.assertEqual(v, check_value)
count += 1
if read_ts != -1:
session.rollback_transaction()
self.assertEqual(count, flcs_nrows if flcs_nrows is not None else nrows)
cursor.close()
def parse_run_write_gen(self, uri):
meta_cursor = self.session.open_cursor('metadata:')
config = meta_cursor[uri]
meta_cursor.close()
# The search string will look like: 'run_write_gen=<num>'.
# Just reverse the string and take the digits from the back until we hit '='.
write_gen = re.search('run_write_gen=\d+', config)
self.assertTrue(write_gen is not None)
write_gen_str = str()
for c in reversed(write_gen.group(0)):
if not c.isdigit():
self.assertEqual(c, '=')
break
write_gen_str = c + write_gen_str
return int(write_gen_str)
def test_hs(self):
active_files = []
if self.value_format == '8t':
value1 = 97
value2 = 100
else:
value1 = 'a' * 500
value2 = 'd' * 500
# Set up 'numfiles' with 'numrows' entries. We want to create a number of files that
# contain active history (content newer than the oldest timestamp).
for f in range(self.numfiles):
table_uri = 'table:%s.%d' % (self.file_name, f)
file_uri = 'file:%s.%d.wt' % (self.file_name, f)
# Create a small table.
ds = SimpleDataSet(
self, table_uri, 0, key_format=self.key_format, value_format=self.value_format,
config='log=(enabled=false)')
ds.populate()
# Checkpoint to ensure we write the files metadata checkpoint value.
self.session.checkpoint()
# Get the base write gen of the file so we can compare after the handles get closed.
base_write_gen = self.parse_run_write_gen(file_uri)
active_files.append((base_write_gen, ds))
# Pin oldest and stable to timestamp 1.
self.conn.set_timestamp('oldest_timestamp=' + self.timestamp_str(1) +
',stable_timestamp=' + self.timestamp_str(1))
# Perform a series of updates over our files at timestamp 2. This being data we can later assert
# to ensure the history store is working as intended.
for (_, ds) in active_files:
# Load data at timestamp 2.
self.large_updates(ds.uri, value1, ds, self.nrows // 2 , 2)
# We want to create a long running read transaction in a seperate session which we will persist over the closing and
# re-opening of handles. We want to ensure the correct data gets read throughout this time period.
session_read = self.conn.open_session()
session_read.begin_transaction('read_timestamp=' + self.timestamp_str(2))
# Check our inital set of updates are seen at the read timestamp.
for (_, ds) in active_files:
# Check that all updates at timestamp 2 are seen.
self.check(session_read, value1, ds.uri, self.nrows // 2)
# Perform a series of updates over files at a later timestamp. Checking the history store data is consistent
# with old and new timestamps.
for (_, ds) in active_files:
# Load more data with a later timestamp.
self.large_updates(ds.uri, value2, ds, self.nrows, 100)
# Check that the new updates are only seen after the update timestamp.
self.check(self.session, value1, ds.uri, self.nrows // 2, 2, flcs_nrows=self.nrows)
self.check(self.session, value2, ds.uri, self.nrows, 100)
# Set the stable timestamp to 100 to let checkpoint write all the stable data.
self.conn.set_timestamp('stable_timestamp=' + self.timestamp_str(100))
# Our sweep scan interval is every 1 second and the amount of idle time needed for a handle to be closed is 2 seconds.
# It should take roughly 3 seconds for the sweep server to close our file handles. Lets wait at least double
# that to be safe.
max = 6
sleep = 0
# After waiting for the sweep server to remove our idle handles, the only open
# handles that should be the metadata file, history store file and lock file.
final_numfiles = 3
# Open the stats cursor to collect the dhandle sweep status.
stat_cursor = self.session.open_cursor('statistics:', None, None)
while sleep < max:
# We continue doing checkpoints which as a side effect runs the session handle sweep. This encouraging the idle
# handles get removed.
# Note, though checkpointing blocks sweeping, the checkpoint should be fast and not add too much extra time to the
# overall test time.
self.session.checkpoint()
sleep += 0.5
time.sleep(0.5)
stat_cursor.reset()
curr_files_open = stat_cursor[stat.conn.file_open][2]
curr_dhandles_removed = stat_cursor[stat.conn.dh_sweep_remove][2]
curr_dhandle_sweep_closes = stat_cursor[stat.conn.dh_sweep_close][2]
self.printVerbose(3, "==== loop " + str(sleep))
self.printVerbose(3, "Number of files open: " + str(curr_files_open))
self.printVerbose(3, "Number of connection sweep dhandles closed: " + str(curr_dhandle_sweep_closes))
self.printVerbose(3, "Number of connection sweep dhandles removed from hashlist: " + str(curr_dhandles_removed))
# We've sweeped all the handles we can if we are left with the number of final dhandles
# that we expect to be always open.
if curr_files_open == final_numfiles and curr_dhandle_sweep_closes >= self.numfiles:
break
stat_cursor.reset()
final_dhandle_sweep_closes = stat_cursor[stat.conn.dh_sweep_close][2]
stat_cursor.close()
# We want to assert our active history files have all been closed.
self.assertGreaterEqual(final_dhandle_sweep_closes, self.numfiles)
# Using our long running read transaction, we want to now check the correct data can still be read after the
# handles have been closed.
for (_, ds) in active_files:
# Check that all updates at timestamp 2 are seen.
self.check(session_read, value1, ds.uri, self.nrows // 2, flcs_nrows=self.nrows)
session_read.rollback_transaction()
# Perform a series of checks over our files to ensure that our transactions have been written
# before the dhandles were closed/sweeped.
# Also despite the dhandle is being re-opened, we don't expect the base write generation
# to have changed since we haven't actually restarted the system.
for idx, (initial_base_write_gen, ds) in enumerate(active_files):
# Check that the most recent transaction has the correct data.
self.check(self.session, value2, ds.uri, self.nrows, 100)
file_uri = 'file:%s.%d.wt' % (self.file_name, idx)
# Get the current base_write_gen and ensure it hasn't changed since being
# closed.
base_write_gen = self.parse_run_write_gen(file_uri)
self.assertEqual(initial_base_write_gen, base_write_gen)
if __name__ == '__main__':
wttest.run()
|
