Details
-
Bug
-
Resolution: Cannot Reproduce
-
Test Blocker
-
6.5.0
-
Centos 7.6
-
Untriaged
-
Yes
Description
Install Couchbase server 6.5.0-4052 on a centos 7.6 server.
Initial server with services kv,index,n1ql,fts,eventing
Create a default bucket and load 600 docs to default bucket.
Create a backup dir and config this dir to backup repo
Run backup. Backup will take a lot of time to run (like hang) which will timeout in automation test.
Base on this observation, if we load only 1 docs to default bucket, it will take more time to do backup since it will wait for timeout at each vbucket
|
2019-08-16T14:46:40.578-07:00 WARN: Could not create stat file for shard: shard_392.sqlite.0 -- storage.(*DataBackup).closeWorker() at databackup.go:524
|
2019-08-16T14:46:40.579-07:00 WARN: Could not create stat file for shard: shard_394.sqlite.0 -- storage.(*DataBackup).closeWorker() at databackup.go:524
|
2019-08-16T14:46:40.579-07:00 WARN: Could not create stat file for shard: shard_396.sqlite.0 -- storage.(*DataBackup).closeWorker() at databackup.go:524
|
2019-08-16T14:46:40.580-07:00 WARN: Could not create stat file for shard: shard_398.sqlite.0 -- storage.(*DataBackup).closeWorker() at databackup.go:524
|
|
I will find what the last stable build is.
To verify this theory, I load more than 10K docs to this default bucket, backup finishes in less than 40 seconds
Attachments
Issue Links
- relates to
-
MB-35595 Improve shard stat logic to avoid warnings on backups smaller than the number of vbuckets
- Closed