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

Issue 870803 link

Starred by 1 user
Status: Duplicate
Owner:
no...@chromium.org
Closed: Jan 4
Cc:
garymm@google.com
█ kjharland@google.com
tandrii@chromium.org
no...@chromium.org
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug



Sign in to add a comment

CQ shows stale status after merging change

Project Member Reported by kjharland@google.com, Aug 3 
This is copied from a bug in Fuchsia's Jira queue at https://fuchsia.atlassian.net/browse/INTK-488, which readers may or may not have access to.

Original bug: 

   Example change: https://fuchsia-review.googlesource.com/c/garnet/+/180402

  Here's the sequence as I remember it:
  Start CQ dry run.
  All "tryjobs" passed except the first one.
  First "tryjob" failed twice (doesn't matter why).
  Rebase button.
  Start CQ dry run.
  First "tryjob" passed, shows green box over two previous fails.
  Submit to CQ.
  Merged.
  No longer shows green box over two previous fails (unexpected).

  (I'm not sure what the priority of this ought to be - it doesn't seem 
  to impact the workflow all that much, but it might cause some 
  confusion re. how a CL got checked in, if it looks like it was never 
  all green.)

Comment 1 by kjharland@google.com, Aug 3

Description: Show this description

Comment 2 by dtapu...@chromium.org, Aug 24

Components: Infra>Codereview>Gerrit

Comment 3 by ajp@chromium.org, Nov 14

Components: -Infra>Codereview>Gerrit Infra>Platform>Buildbucket>Gerrit

Comment 4 by mar...@chromium.org, Nov 15

Cc: tandrii@chromium.org
Status: Available (was: Untriaged)

Comment 5 by tandrii@google.com, Nov 15

Cc: no...@chromium.org
Ah, that's the problem in plugin which assumes that last patchset of a merged CL is a generated patchset (like during rebase always OR cherry-pick strategy).
Thus, plugin only asks server for PS#1 and #2, skipping #3 **after CL status changes to "merged"**:

https://cr-buildbucket.appspot.com/_ah/api/buildbucket/v1/search?max_builds=500&fields=builds%28bucket%2Cfailure_reason%2Cid%2Cparameters_json%2Cresult%2Cstatus%2Ctags%2Curl%29&tag=buildset%3Apatch%2Fgerrit%2Ffuchsia-review.googlesource.com%2F180402%2F1
https://cr-buildbucket.appspot.com/_ah/api/buildbucket/v1/search?max_builds=500&fields=builds%28bucket%2Cfailure_reason%2Cid%2Cparameters_json%2Cresult%2Cstatus%2Ctags%2Curl%29&tag=buildset%3Apatch%2Fgerrit%2Ffuchsia-review.googlesource.com%2F180402%2F2


that said, I thought nodir@ has actually tried to fix this already, but apparently it didn't work as intended.

Comment 6 by garymm@google.com, Jan 4

Cc: kjharland@google.com garymm@google.com
This bug was filed way back in August. Was nodir's fix rolled out after that? If so this is probably fixed?

Comment 7 by no...@chromium.org, Jan 4 

is https://fuchsia-review.googlesource.com/c/garnet/+/180402 supposed to repro the problem? This CL does not have "MERGED" patchset, so I don't see how the reverted fix https://chromium-review.googlesource.com/c/infra/gerrit-plugins/buildbucket/+/1225117 is relevant

i currently cannot repro

Comment 8 by no...@chromium.org, Jan 4

Status: Unconfirmed (was: Available)

Comment 9 by garymm@google.com, Jan 4 

Yes it reproduces the problem.
It is showing tryjob results from patch sets 1-2, but not from patch set 3.

Comment 10 by no...@chromium.org, Jan 4

Mergedinto: 809182
Owner: no...@chromium.org
Status: Duplicate (was: Unconfirmed)
thanks, yeah it is a bug
►

Sign in to add a comment