Import wiredtiger: 28109e1c6a2444d1ebcf8ef87ddc6944cf88eb61 from branch mongodb-master

ref: 5f92dc4c80..28109e1c6a
for: 5.2.0

WT-8346       Migrate Jenkins “wiredtiger-perf-stress” remaining tests to Evergreen

5 files changed, 117 insertions, 44 deletions

diff --git a/src/third_party/wiredtiger/bench/wtperf/runners/evict-btree-stress-multi.json b/src/third_party/wiredtiger/bench/wtperf/runners/evict-btree-stress-multi.json
new file mode 100644
index 00000000000..586b7a2e396
--- /dev/null
+++ b/src/third_party/wiredtiger/bench/wtperf/runners/evict-btree-stress-multi.json
@@ -0,0 +1,6 @@
+[
+    {
+      "arguments": null,
+      "operations": ["warnings", "max_latencies"]
+    }
+]
diff --git a/src/third_party/wiredtiger/bench/wtperf/runners/evict-fairness.json b/src/third_party/wiredtiger/bench/wtperf/runners/evict-fairness.json
new file mode 100644
index 00000000000..68b96f67d4c
--- /dev/null
+++ b/src/third_party/wiredtiger/bench/wtperf/runners/evict-fairness.json
@@ -0,0 +1,6 @@
+[
+    {
+      "arguments": ["-C statistics_log=(wait=10000,on_close=true,json=false,sources=[file:])", "-o reopen_connection=false"],
+      "operations": ["eviction_page_seen"]
+    }
+]
diff --git a/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat.py b/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat.py
index 7aab7d20b68..51787e75f9d 100644
--- a/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat.py
+++ b/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat.py
@@ -27,20 +27,25 @@
 # 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 glob, re, os
+import json
+
 
 class PerfStat:
     def __init__(self,
                  short_label: str,
-                 pattern: str,
-                 input_offset: int,
                  output_label: str,
+                 input_offset: int = 0,
                  output_precision: int = 0,
+                 pattern: str = None,
+                 stat_file: str = 'test.stat',
                  conversion_function=int):
         self.short_label: str = short_label
-        self.pattern: str = pattern
-        self.input_offset: int = input_offset
         self.output_label: str = output_label
+        self.input_offset: int = input_offset
         self.output_precision: int = output_precision
+        self.pattern: str = pattern
+        self.stat_file = stat_file
         self.conversion_function = conversion_function
         self.values = []
 
@@ -49,6 +54,14 @@ class PerfStat:
             converted_value = self.conversion_function(val)
             self.values.append(converted_value)
 
+    def find_stat(self, test_stat_path: str):
+        matches = []
+        for line in open(test_stat_path):
+            match = re.search(self.pattern, line)
+            if match:
+                matches.append(float(line.split()[self.input_offset]))
+        return matches
+
     def average(self, vals):
         return self.conversion_function(sum(vals) / len(vals))
 
@@ -60,6 +73,15 @@ class PerfStat:
         else:
             return self.average(self.values)
 
+    def get_value_list(self, brief: bool):
+        as_dict = {
+                    'name': self.output_label,
+                    'value': self.get_value()
+                }
+        if not brief:
+            as_dict['values'] = self.values
+        return [as_dict]
+
     def are_values_all_zero(self):
         result = True
         for value in self.values:
@@ -74,8 +96,58 @@ class PerfStatMin(PerfStat):
         min_3_vals = sorted(self.values)[:3]
         return self.average(min_3_vals)
 
+
 class PerfStatMax(PerfStat):
     def get_value(self):
         """Return the averaged maximum of all gathered values"""
         max_3_vals = sorted(self.values)[-3:]
         return self.average(max_3_vals)
+
+
+class PerfStatCount(PerfStat):
+    def find_stat(self, test_stat_path: str):
+        """Return the total number of times a pattern matched"""
+        total = 0
+        test_stat_path = glob.glob(test_stat_path)[0]
+        for line in open(test_stat_path):
+            match = re.search(self.pattern, line)
+            if match:
+                total += 1
+        return [total]
+
+
+class PerfStatLatency(PerfStat):
+    def __init__(self,
+                 short_label: str,
+                 stat_file:str,
+                 output_label: str,
+                 num_max: int):
+                 super().__init__(short_label=short_label,
+                                  stat_file=stat_file,
+                                  output_label=output_label)
+                 self.num_max = num_max
+
+    def find_stat(self, test_stat_path: str):
+        values = []
+        if os.path.isfile(test_stat_path):
+            for line in open(test_stat_path):
+                as_dict = json.loads(line)
+                values.append(as_dict["wtperf"]["read"]["max latency"])
+                values.append(as_dict["wtperf"]["update"]["max latency"])
+        return values
+
+    def get_value(self, nth_max: int):
+        """Return the nth maximum number from all the gathered values"""
+        return sorted(self.values)[-nth_max]
+
+    def get_value_list(self, brief: bool):
+        as_list = []
+        for i in range(1, self.num_max + 1):
+            as_dict = {
+                'name': self.output_label + str(i),
+                'value': self.get_value(i)
+            }
+            if not brief:
+                as_dict['values'] = self.values
+            as_list.append(as_dict)
+        return as_list
diff --git a/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat_collection.py b/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat_collection.py
index db3652fcab9..d62441f69a0 100644
--- a/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat_collection.py
+++ b/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/perf_stat_collection.py
@@ -28,19 +28,13 @@
 # ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 # OTHER DEALINGS IN THE SOFTWARE.
 
-import re
-from typing import List
-
+import os
 from perf_stat import PerfStat
+from typing import List
 
 
-def find_stat(test_stat_path: str, pattern: str, position_of_value: int):
-    matches = []
-    for line in open(test_stat_path):
-        match = re.search(pattern, line)
-        if match:
-            matches.append(float(line.split()[position_of_value]))
-    return matches
+def create_test_stat_path(test_home_path: str, test_stats_file: str):
+    return os.path.join(test_home_path, test_stats_file)
 
 
 class PerfStatCollection:
@@ -50,23 +44,17 @@ class PerfStatCollection:
     def add_stat(self, perf_stat: PerfStat):
         self.perf_stats[perf_stat.short_label] = perf_stat
 
-    def find_stats(self, test_stat_path: str, operations: List[str]):
+    def find_stats(self, test_home: str, operations: List[str]):
         for stat in self.perf_stats.values():
             if not operations or stat.short_label in operations:
-                values = find_stat(test_stat_path=test_stat_path,
-                                   pattern=stat.pattern,
-                                   position_of_value=stat.input_offset)
+                test_stat_path = create_test_stat_path(test_home, stat.stat_file)
+                values = stat.find_stat(test_stat_path=test_stat_path)
                 stat.add_values(values=values)
 
     def to_value_list(self, brief: bool):
-        as_list = []
+        stats_list = []
         for stat in self.perf_stats.values():
             if not stat.are_values_all_zero():
-                as_dict = {
-                    'name': stat.output_label,
-                    'value': stat.get_value()
-                }
-                if not brief:
-                    as_dict['values'] = stat.values
-                as_list.append(as_dict)
-        return as_list
+                stat_list = stat.get_value_list(brief = brief)
+                stats_list.extend(stat_list)
+        return stats_list
diff --git a/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/wtperf_run.py b/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/wtperf_run.py
index dc84f31ac5d..0ae18756b53 100644
--- a/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/wtperf_run.py
+++ b/src/third_party/wiredtiger/bench/wtperf/wtperf_run_py/wtperf_run.py
@@ -29,23 +29,18 @@
 # OTHER DEALINGS IN THE SOFTWARE.
 
 import argparse
-import json
 import os.path
-import subprocess
-import sys
 import platform
 import psutil
+import subprocess
+import sys
+import json
+from perf_stat import PerfStat, PerfStatCount, PerfStatLatency, PerfStatMax, PerfStatMin
+from perf_stat_collection import PerfStatCollection
 from pygit2 import discover_repository, Repository
 from pygit2 import GIT_SORT_NONE
 from typing import List
-
 from wtperf_config import WTPerfConfig
-from perf_stat import PerfStat, PerfStatMax, PerfStatMin
-from perf_stat_collection import PerfStatCollection
-
-# the 'test.stat' file is where wt-perf.c writes out it's statistics
-# (within the directory specified by the 'home' parameter)
-test_stats_file = 'test.stat'
 
 
 def create_test_home_path(home: str, test_run: int, operations: List[str] = None):
@@ -56,10 +51,6 @@ def create_test_home_path(home: str, test_run: int, operations: List[str] = None
     return home_path
 
 
-def create_test_stat_path(test_home_path: str):
-    return os.path.join(test_home_path, test_stats_file)
-
-
 def get_git_info(git_working_tree_dir):
     repository_path = discover_repository(git_working_tree_dir)
     assert repository_path is not None
@@ -158,10 +149,9 @@ def run_test(config: WTPerfConfig, test_run: int, operations: List[str] = None,
 def process_results(config: WTPerfConfig, perf_stats: PerfStatCollection, operations: List[str] = None):
     for test_run in range(config.run_max):
         test_home = create_test_home_path(home=config.home_dir, test_run=test_run, operations=operations)
-        test_stats_path = create_test_stat_path(test_home)
         if config.verbose:
-            print('Reading test stats file: {}'.format(test_stats_path))
-        perf_stats.find_stats(test_stat_path=test_stats_path, operations=operations)
+            print('Reading stats from {} directory.'.format(test_home))
+        perf_stats.find_stats(test_home=test_home, operations=operations)
 
 
 def setup_perf_stats():
@@ -200,6 +190,17 @@ def setup_perf_stats():
                                     pattern=r'updates,',
                                     input_offset=8,
                                     output_label='Min update throughput'))
+    perf_stats.add_stat(PerfStatCount(short_label="warnings",
+                                      pattern='WARN',
+                                      output_label='Warnings'))
+    perf_stats.add_stat(PerfStatLatency(short_label="max_latencies",
+                                        stat_file='monitor.json',
+                                        output_label='Latency Max',
+                                        num_max = 5))
+    perf_stats.add_stat(PerfStatCount(short_label="eviction_page_seen",
+                                      stat_file='WiredTigerStat*',
+                                      pattern='[0-9].wt cache: pages seen by eviction',
+                                      output_label='Pages seen by eviction'))
     return perf_stats