Test mock ontology term replacement for new SOP (II)
kltm opened this issue · 17 comments
A follow on from #476
A place to put results and discussion of the second mock run of the new minerva outage SOP.
Related to #456
Tagging @vanaukenk @balhoff
Possible future here: https://github.com/geneontology/noctua-models/commits/issue-minerva-479-test-replacement-2
Pinging @vanaukenk and @balhoff
@kltm - there are other aspects of the diffs that I'm not clear on. If you have time to review today, that'd be helpful. Either during the meeting break of after 3:30pm EDT is good.
Hmmmmm. Seems like a problem:
EMAPA:16627 somite 27 sclerotome
EMAPA:16647 is an xref to UBERON:0034876 and as near as I can see should not have been replaced. It's 'live' in MGI.
Sorry, my mistake. It's EMAPA:16647 but the question still stands. I'll edit above.
As near as I can tell, it should not have been replaced.
Perhaps a rule to never replace a term from one ontology with another unless reviewed by a curator.
This term was deprecated within Uberon for some reason. This is definitely a bug in Uberon and shouldn't have happened. I'll file a bug there. We shouldn't run this update.
@balhoff Possibly related question: does minerva do replace operations when replaced_by is present during any other operations, or is it only considered when we do the bulk updates? For example, if during "normal" editorial operations I did something with an obsolete term, would minerva ever attempt to replace it?
No. It used to do some automatic replacement to calculate root types for nodes (but not changing the data), but I removed this in the recent performance-improvement PRs.
@balhoff Okay great! Something was kicking around in my head about that--thank you for clarifying.
After talking to @vanaukenk we'll go ahead with a reduced SOP today, just: power down, flush, save, remove ontology journal, restart (and have ontology journal build back up).
I've added this issue to the 2022-05-02 ontology call; we may also want to discuss on a manager's call.
@balhoff @kltm
I'm checking in about our Noctua outage this coming Thursday, the 14th.
From the ticket in the UBERON tracker, it looks like the specific EMAPA issue we encountered has now been fixed.
Is there anything more we need to do on our end?
I'm assuming we'll do another draft replacement sometime this week so I can check the pending changes before we move ahead with an update on production Noctua.
Thanks.
I think we need to wait for the next Uberon release. I can ask when the next will be.