Bug #48671
closedsuite_sha1 and workunit:sha1 can get out of sync with sha1
0%
Description
https://tracker.ceph.com/issues/48663#note-1 indicates that a task was queued with sha 9b5210e497b0bf4322e569c37bf27a1c0d47f7dd and that between the time it was scheduled and the time the suite_sha1 was calculated the branch saw the merge of some code which caused the test to fail.
I think in create_initial_config we should check whether self.args.suite_repo is empty and self.args.suite_branch is equal to the --ceph branch being used or empty and, if so, set suite_hash to --sha1. If someone wants to use a different suite branch/hash they can accomplish that by passing suite_dir or suite_repo/suite_branch pair. I think this should be more inline with the way it is expected to work despite it having been the way it is for a considerable amount of time. This actually could explain many otherwise unexplained failures in a repo which is being updated all eh time.
I'll craft a patch for this and seek approval.
Updated by Brad Hubbard over 3 years ago
- Description updated (diff)
- Assignee set to Brad Hubbard
Updated by Sebastian Wagner over 3 years ago
- Related to Bug #47336: `orch device ls`: Unexpected argument '--wide' added
Updated by Sebastian Wagner over 3 years ago
- Related to Bug #48663: TypeError: ac_user_create_cmd() got an unexpected keyword argument 'inbuf' added
Updated by Josh Durgin about 3 years ago
- Status changed from New to Resolved