Details
-
Bug
-
Resolution: Fixed
-
Major
-
6.6.0
-
Untriaged
-
1
-
No
Description
If an error occurs during a read only restore, the error handler will attempt to remove the lockfile by calling the 'Unlock' function. Naturally this function will fail to remove the lockfile (if the bucket is actually in read-only mode) because we never created it.
The cleanest way to solve this will be to have the lockfile implementation detect that we haven't actually run the 'Lock' function (and therefore the lockfile was not created) and return nil when the exit handler attempts to remove the lockfile. This avoiding leaking read only logic into the exit handler (which would ultimately end up producing a large (ish) code change due to 'exitIfError' being used all over the place).
Attachments
For Gerrit Dashboard: MB-39697 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
129601,2 | MB-39697 Avoid removing a lockfile that was never created | mad-hatter | backup | Status: MERGED | +2 | +1 |
130113,2 | Merge branch 'mad-hatter' into master | master | backup | Status: MERGED | +2 | +1 |