tree 3dcfe664431595b6795ad731960da44ed683f425
parent 1df24549848cf15a077d54b9cea99ed8e6a8d07f
author Luca Milanesio <luca.milanesio@gmail.com> 1713648248 +0100
committer Luca Milanesio <luca.milanesio@gmail.com> 1713648543 +0000

Check for existence of change's target SHA1 for reindexing

When the target branch of a change was advancing rapidly,
the reindexing retries applied by the ForwardedIndexingHandlerWithRetries
were continuing potentially forever as they were waiting to
see the target branch's expected SHA1 for the event.

Whilst it is vital to wait for a target branch SHA1 to appear before
reindexing a change on a remote site, it isn't essential to have
that SHA1 being *exactly* the tip of the target branch, as that
moment can be easily missed.

Example:
- Create two changes C1 and C2 in parallel and push them with
  %submit option
- The changes C1 and C2 will trigger the rapid advance of their
  target SHA1
- The other nodes may reindex C1 but not C2 because the target branch
  has advanced more than expected

Checking simply the existence of the target SHA1 in the repository
is enough to satisfy the conditions for the change being reindexed.

Bug: Issue 335353379
Change-Id: Ife361ec3378a60c0d1eb15df0f9570d569adc07a