Pushlog and Treeherder do not show a pushed backout
Categories
(Developer Services :: Mercurial: Pushlog, defect)
Tracking
(Not tracked)
People
(Reporter: RaulG, Unassigned)
References
Details
Attachments
(1 file)
|
1.88 KB,
image/png
|
Details |
A backout was pushed to autoland and we received an error, something related to "database locked". Unfortunately i don't have the full error line anymore.
Autoland was closed and #vcs, #treeherder channels were pinged.
Checked the tip of mercurial, locally and the backout is there. This is the revision: https://hg.mozilla.org/integration/autoland/rev/c1170685b543
Note that there isn't any push id, push user or push date.
Sheriffs decided to open the tree and push a revision. It showed up in pushlog and TH. The backout which was pushed before, is still not showing up.
Pushlog link: https://hg.mozilla.org/integration/autoland/pushloghtml
It's interesting that the push metadata is missing for that commit. The parent and child commits seem correct, though.
Comment 2•7 years ago
•
|
||
The commit is also missing from pushlog: https://hg.mozilla.org/integration/autoland/pushloghtml Just realized that was mentioned in the summary, oops.
Comment 3•7 years ago
|
||
When Cristina pushed, the first attempt got sqlite3.OperationalError: database is locked and after that abort: stream ended unexpectedly [...]. She tried to push again. Looks like the first attempt got recorded.
Updated•7 years ago
|
Comment 4•7 years ago
|
||
This looks like another instance of the issue reported in bug 1511543.
This is causing issues unifying mozilla-unified.
Comment 5•7 years ago
|
||
Looking at the commit from bug 1511543, the pushlog info was never fixed. Not sure what to do here.
Comment 6•7 years ago
|
||
Somehow mozilla-unified has managed to move past the error and is now caught up just fine. When I took a look at the bookmarks via https://hg.mozilla.org/mozilla-unified/bookmarks, there was no autoland bookmark. F5'd and now it shows up... Very strange, but at least the main "this is causing something to break" problem is averted.
I tried parsing through the hg.log and hg-push.log files on hgssh1 looking for exceptions/errors (like "database is locked", "sqlite3" and "OperationalError") but found nothing. This is very strange, especially since we found the source quite easily in bug 1511543.
Description
•