4083: crunchstat-summary imports _strptime.

[arvados.git] / tools / crunchstat-summary / crunchstat_summary / summarizer.py

diff --git a/tools/crunchstat-summary/crunchstat_summary/summarizer.py b/tools/crunchstat-summary/crunchstat_summary/summarizer.py
index d058c232ac3c4ff177e11aa29a3590a8be4a833a..a88e4d5c41f06f017863965d43a9b60818426130 100644 (file)
--- a/tools/crunchstat-summary/crunchstat_summary/summarizer.py
+++ b/tools/crunchstat-summary/crunchstat_summary/summarizer.py
@@ -11,6 +11,7 @@ import math
 import re
 import sys
 import threading
+import _strptime
 
 from arvados.api import OrderedJsonModel
 from crunchstat_summary import logger
@@ -21,6 +22,11 @@ from crunchstat_summary import logger
 AVAILABLE_RAM_RATIO = 0.95
 
 
+# Workaround datetime.datetime.strptime() thread-safety bug by calling
+# it once before starting threads.  https://bugs.python.org/issue7980
+datetime.datetime.strptime('1999-12-31_23:59:59', '%Y-%m-%d_%H:%M:%S')
+
+
 class Task(object):
     def __init__(self):
         self.starttime = None
@@ -336,19 +342,20 @@ class Summarizer(object):
                 int(math.ceil(nearlygibs(used_mib))*AVAILABLE_RAM_RATIO*1024))
 
     def _recommend_keep_cache(self):
-        """Recommend increasing keep cache if miss rate is above 0.2%"""
-        if self.job_tot['keepcalls']['get'] == 0:
+        """Recommend increasing keep cache if utilization < 80%"""
+        if self.job_tot['net:keep0']['rx'] == 0:
             return
-        miss_rate = float(self.job_tot['keepcache']['miss']) / float(self.job_tot['keepcalls']['get']) * 100.0
+        utilization = (float(self.job_tot['blkio:0:0']['read']) /
+                       float(self.job_tot['net:keep0']['rx']))
         asked_mib = self.existing_constraints.get('keep_cache_mb_per_task', 256)
 
-        if miss_rate > 0.2:
+        if utilization < 0.8:
             yield (
-                '#!! {} Keep cache miss rate was {:.2f}% -- '
-                'try runtime_constraints "keep_cache_mb_per_task":{}'
+                '#!! {} Keep cache utilization was {:.2f}% -- '
+                'try runtime_constraints "keep_cache_mb_per_task":{} (or more)'
             ).format(
                 self.label,
-                miss_rate,
+                utilization * 100.0,
                 asked_mib*2)