-
Type: Bug
-
Resolution: Done
-
Priority: Major - P3
-
Affects Version/s: 3.0.6, 3.0.7, 3.2.0-rc0
-
Component/s: WiredTiger
-
None
-
Fully Compatible
-
ALL
-
Under a workload with a high volume of collection (and index) creation over a number of database then the dropping of these collections will cause listDatabases command to run longer on a WT instance.
Using the reproduction script shows the following execution time for listDatabases per lap in seconds:
"took: 1.705329872" "took: 1.328905619" "took: 1.55636039" "took: 1.221779232" "took: 1.772605573" "took: 1.689325783" "took: 1.15700596" "took: 1.470458173" "took: 1.255596249" "took: 1.455441504" "took: 8.487761074" "took: 13.784377947" "took: 15.079964722" "took: 13.892499541" "took: 15.179368285" "took: 14.683478172" "took: 14.776398389" "took: 14.481528086" "took: 14.688469802" "took: 13.738872847" "took: 14.298077263"
- depends on
-
WT-2174 Enhance size statistics to not wait for the table lock
- Closed
- is related to
-
SERVER-20716 Slow listDatabases with WT
- Closed
-
SERVER-20877 Under cache-full conditions serverStatus can become stuck
- Closed