Issue Status as of Jun 9, 2014
ISSUE SUMMARY
When a count() operation takes longer than the specified value of slowms (100ms by default), the corresponding message in the log file contains information about the plan summary used in the query. Here are two sample log lines:
2014-06-09T17:30:10.759-0400 [conn4] command test.$cmd command: count { count: "foo", query: { x: { $gt: 12345.0 } }, fields: {} } planSummary: COLLSCAN keyUpdates:0 numYields:287 locks(micros) r:253415 reslen:48 154ms 2014-06-09T17:30:58.876-0400 [conn4] command test.$cmd command: count { count: "foo", query: { x: { $gt: 12345.0 } }, fields: {} } planSummary: COUNT { x: 1.0 } keyUpdates:0 numYields:165 locks(micros) r:133823 reslen:48 78ms
The first line of the log above shows a slow count() operation doing a collection scan:
planSummary: COLLSCAN
while the second shows a slow count() operation using the index {x: 1}:
planSummary: COUNT { x: 1.0 }
RELEASE VERSION
This improvement is included in the MongoDB 2.6.2 production release.
Original description
planSummary information is very useful in determining index usage when mongod prints slow operations to the logs. Please add to count command logging which is not present in 2.6.0.