Closed
Bug 1276355
Opened 9 years ago
Closed 9 years ago
IntegrityError for key 'idx_revision_hash' with pulse ingestion
Categories
(Tree Management :: Treeherder: Data Ingestion, defect)
Tree Management
Treeherder: Data Ingestion
Tracking
(Not tracked)
RESOLVED
FIXED
People
(Reporter: camd, Assigned: camd)
References
Details
Attachments
(2 files)
We need a better check for creating resultsets on the fly wrt the revision_hash. Though this field is on its way out, it is still being used as we transition to just using the revision.
We still have a unique index for revision_hash, and therefore need to check for duplicates of that value when we insert a new resultset record.
Found during testing for pulse data ingestion.
Assignee | ||
Updated•9 years ago
|
Assignee: nobody → cdawson
Comment 1•9 years ago
|
||
Assignee | ||
Updated•9 years ago
|
Attachment #8757476 -
Flags: review?(emorley)
Comment 2•9 years ago
|
||
Comment on attachment 8757476 [details] [review]
[treeherder] mozilla:revision-hash-integrity-error > mozilla:master
I'm not sure I understand why this change is being made?
(The commit message is just a single line)
Could you add a multi-line commit message to help me review and also for future git log benefit? :-)
Attachment #8757476 -
Flags: review?(emorley)
Assignee | ||
Comment 3•9 years ago
|
||
Sorry about that, Ed. Yeah, that was too terse. I fixed the commit, the PR and I'll put it here for good measure: :)
When updating or inserting a resultset record, we need to check against
the revision_hash value because, while revision_hash is on its way out, we
do still use it as a fallback and there is a unique index for it.
This error can occur as a race condition when a job is being ingested that
does not yet have a resultset, but during ingestion, the resultset is
created in parallel. So the process ingesting the job tries to insert
the resultset and gets the index violation. This check will prevent that.
Assignee | ||
Comment 4•9 years ago
|
||
Comment on attachment 8757476 [details] [review]
[treeherder] mozilla:revision-hash-integrity-error > mozilla:master
Hopefully this clarifies things. :)
Attachment #8757476 -
Flags: review?(emorley)
Comment 5•9 years ago
|
||
Comment on attachment 8757476 [details] [review]
[treeherder] mozilla:revision-hash-integrity-error > mozilla:master
Looks good, thank you :-)
Attachment #8757476 -
Flags: review?(emorley) → review+
Comment 6•9 years ago
|
||
Commit pushed to master at https://github.com/mozilla/treeherder
https://github.com/mozilla/treeherder/commit/3642e03307428ef92ce665d267cbc144fdb61d09
Bug 1276355 - Fix IntegrityError for idx_revision_hash on resultset creation
When updating or inserting a resultset record, we need to check against
the revision_hash value because, while revision_hash is on its way out, we
do still use it as a fallback and there is a unique index for it.
This error can occur as a race condition when a job is being ingested that
does not yet have a resultset, but during ingestion, the resultset is
created in parallel. So the process ingesting the job tries to insert
the resultset and gets the index violation. This check will prevent that.
Assignee | ||
Updated•9 years ago
|
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•