Details
Major Description
There seems to be a requirement for the cbbackupmgr tool to be able to
create a Bucket at the destination cluster/node during the restore path.
This will make it easier for customers during the restore process.
Attachments
Issue Links
- relates to
-
DOC-5501 Bucket creation unsupported/undocumented during cbbackumgr restore
- Resolved
| For Gerrit Dashboard: MB-26916 | ||||||
|---|---|---|---|---|---|---|
| # | Subject | Branch | Project | Status | CR | V |
| 101597,3 | MB-26916 Allow restore to create missing buckets | master | backup | Status: MERGED | +2 | +1 |
Activity
Build couchbase-server-6.5.0-1607 contains backup commit e7eae0d with commit message:
MB-26916 Allow restore to create missing buckets
What is the current default behavior? Is it a flag to be set? Are instructions documented anywhere?
By default cbbackupmgr restore won't create the buckets on a restore. A flag --auto-create-buckets was added to the command that will create remote buckets if they do not exist. This was documented in the man page of the command.
Verified on build 6.5.0-3458 with no travel-sample in restore cluster
[root@localhost ~]# /opt/couchbase/bin/cbbackupmgr restore -r backup -a /tmp/backup/ -c 12.11.10.129 -u Administrator -p password --vbucket-filter 1-10 --auto-create-buckets
|
|
|
(1/1) Restoring backup 2019-06-07T14_28_55.849194657-07_00
|
Copied all data in 8s (Avg. 28.16KB/Sec) 305 items / 225.25KB
|
travel-sample [=================================================================================================================================================================================] 100.00%
|
|
|
Restore completed successfully
|
[root@localhost ~]#
|
|
We did this on purpose for two reasons.
1) We did not want to backup bucket password
2) Restoring to a cluster that is not the same size may fail due to the server not having enough memory
Since the first issue is no longer a problem in 5.0 because we have RBAC I'm open to considering this, but I wouldn't want to make it the default behavior.