New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 874267 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Last visit > 30 days ago
Closed: Aug 15
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: ----



Sign in to add a comment

Auditing on chromium/src was stuck on a commit landed during Gerrit outage

Project Member Reported by st...@chromium.org, Aug 14

Issue description

The last relevant commit was 5 days ago https://cr-audit-commits.appspot.com/view/status

And it was stuck at https://chromium.googlesource.com/chromium/src/+/a944b0d7 which was landed during Gerrit outage and got purged because chromium/src master branch was reset to very last commit before the Gerrit outage.

Two actions needed here:
1) Reset the checking point of audit-app to the last "good" revision. One thing to be careful here is a burst of commits to audit.
2) Set up an alert for auditing on chromium/src master branch (and maybe master branches of other audited repos): if there is a large gap between last scanned commit and the current scanning time, fire an alert.
 
Note: All other audited repos don't have the same problem as chromium/src.
Owner: robert...@chromium.org
Status: Assigned (was: Available)
Thanks for bringing this to my attention.
Ah, sorry for not PSA-ing you earlier. I forgot to do this on Thursday :(
The plan is as follows:
Manually set the status of these 3 'RelevantCommit' entities to the new
auditSkipped (i.e. 4) see this cl:
https://chromium-review.googlesource.com/c/infra/infra/+/1174987

These are the commits:
[hi on] (master<):robertocn@traal:~/chromium/src$ git show a944b0d7f322400a54def81b313f16b9e125167b
fatal: bad object a944b0d7f322400a54def81b313f16b9e125167b
[hi on] (master<):robertocn@traal:~/chromium/src$ git show 224eb4663f904ffc0c1356fe5f1d83fc4afc2c90
fatal: bad object 224eb4663f904ffc0c1356fe5f1d83fc4afc2c90
[hi on] (master<):robertocn@traal:~/chromium/src$ git show ec978fdcd324d32e1dd1ccec03900084652ebe26
fatal: bad object ec978fdcd324d32e1dd1ccec03900084652ebe26

Next, manually edit the RepoState for chromium/src to set _both_
LastKnownCommit and LastRelevantCommit to 86f8273e2fd5cd87e70f3711250d6edc12da2823  , which is the last commit audited that had a commit position before the outage began.

As discussed on the code review mentioned in #4, we'll instead delete the entries that no longer exist.
Status: Fixed (was: Assigned)
This is fixed.

BTW the audit app didn't even blink when catching up. It took it about 9 seconds to scan ~1184 commits and audit 89 of them.

I will file a separate bug to add an alert that will monitor the http statuses of these tasks.

Sign in to add a comment