Details
-
Bug
-
Resolution: Fixed
-
Major
-
4.0.0, 4.1.2, 4.5.1, 4.6.5, 5.0.1, 5.5.4, 6.0.2, 6.5.0
-
Untriaged
-
No
Description
Easy reproduction. All it needs is any activity that refreshes the namespace.
We get a new pool and silently discard the old.
Issue is, couchbase pools don't have a cleanup method, so they go away without closing buckets, which leaves the connection pools dangling and the connection closers running.
Symptom is loads of [WARN] Finalizing a bucket with active connections. messages in the log.
Need a pool closer method, defensive fixes in go-couchbase, and closing the pool in datastore/couchbase.
Attachments
Issue Links
- causes
-
MB-33208 Queries failing with bulkget exceeded MaxBulkRetries for default(vbid:858,keys:<ud>[query-testemployee11295.7170176-0]</ud>) error
- Closed
For Gerrit Dashboard: MB-33185 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
105476,3 | MB-33185 pools and connection pool leak | master | query | Status: MERGED | +2 | +1 |
105477,3 | MB-33185 pools and connection pool leak | master | go-couchbase | Status: MERGED | +2 | +1 |
105532,2 | backport of MB-33185 | vulcan | query | Status: MERGED | +2 | +1 |
105533,2 | backport of MB-33185 | vulcan | go-couchbase | Status: MERGED | +2 | +1 |
105534,2 | backport of MB-33185 | alice | go-couchbase | Status: MERGED | +2 | +1 |
105535,2 | backport of MB-33185 | alice | query | Status: MERGED | +2 | +1 |
114748,3 | MB-35811 connection pool leak querying system:indexes | master | query | Status: MERGED | +2 | +1 |
114842,2 | MB-35811 connection pool leak querying system:indexes | alice | query | Status: MERGED | +2 | +1 |
114893,2 | MB-35811 connection pool leak querying system:indexes | vulcan | query | Status: MERGED | +2 | +1 |