Details
-
Bug
-
Resolution: Fixed
-
Major
-
6.5.0
Description
In writing a revised test for full disk errors, I've made the following observation. If the disk fills as we're running backup, we get:
# /opt/couchbase/bin//cbbackupmgr backup --archive /cbqe3043/entbackup --repo backup --cluster http://10.112.191.101:8091 --username Administrator --password password --no-progress-bar
|
|
Error backing up cluster: EOF
|
If the disk is completely full, we get:
# /opt/couchbase/bin//cbbackupmgr backup --archive /cbqe3043/entbackup --repo backup --cluster http://10.112.191.101:8091 --username Administrator --password password --no-progress-bar
|
|
Error backing up cluster: write /cbqe3043/entbackup/200724775: no space left on device
|
And a similar error for the config command:
# /opt/couchbase/bin//cbbackupmgr config --archive /cbqe3043/entbackup --repo backup
|
|
Backup repository creation failed: mkdir /cbqe3043/entbackup: no space left on device
|
These seem to be generic OS level errors being propagated to the user, but it would be ideal, if possible, to create a consistent error message when disk is full. Particularly in the case of the EOF error, it isn't clear why the backup failed.
Attachments
For Gerrit Dashboard: MB-35106 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
114699,1 | MB-35106 More informative disk full errors | master | backup | Status: ABANDONED | 0 | 0 |
114700,2 | MB-35106 More informative disk full errors | mad-hatter | backup | Status: MERGED | +2 | +1 |
114738,2 | Merge branch 'mad-hatter' into master | master | backup | Status: MERGED | +2 | +1 |