Details
Improvement
Description
We need to harden collection deletion so that it's more difficult to end up in partial states or receive unhelpful errors.
Issue Links
- relates to
-
SOLR-8450
Our HttpClient retry policy is too permissive.
-
- Closed
-
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
sometimes you can get a fail of core already unloaded.
Why this is, I'm less sure. I suspect it may be related to HttpClient retries. I'm going to make sure we do not retry on Admin commands in SOLR-8450.
Commit 899f1fcf74d1562e4188073da53404b9f918032f in lucene-solr's branch refs/heads/master from markrmiller
[ https://git-wip-us.apache.org/repos/asf?p=lucene-solr.git;h=899f1fc ]
SOLR-8551: Make collection deletion more robust.
Commit 6af3d8ffd6b48b5aea017f660f0a108a167fcf22 in lucene-solr's branch refs/heads/branch_5x from markrmiller
[ https://git-wip-us.apache.org/repos/asf?p=lucene-solr.git;h=6af3d8f ]
SOLR-8551: Make collection deletion more robust.
Working towards some improvements.
One of the cases I'd like to fix is that sometimes you can get a fail of core already unloaded. Since that is what we want, we should not fail in that case.
I've added some support for getting exception class names for remote exceptions - I'd like to use that with a new NonExistentCore SolrException so that we can specifically ignore core already unloaded exceptions when trying to unload a core when deleting a collection.