stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1520926805
@tflobbe CHANGES file updated
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific c
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1518426686
I spent some more time on this class and have identified 2 problematic
things I wanted to share:
* this looks like a NPE waiting to happen
https://github.com/apache/solr/blob/db4cb6627
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1518101663
> That is very similar to my other PR that does the same thing for a similar
case. I might just list both together, which we do sometimes.
The CHANGES.txt proposal sounds good to me,
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1503952520
Status on current PR:
- added the version check on additions to fail in case we are not leader
and version = 0. (to match delete flows)
- changed error status from BAD_REQUEST to IN
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1503929909
Leaving this here for future reference. I think we could consider allowing
doc updates based on the `isSubShardLeader` but this is tricky to verify, and
well beyond my knowledge of this co
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1502421101
> Where is this in the PR? I at least like the sound of it :-)
you can find the logic here, basically not leader logic and zero version
will trigger an exception:
https://github.com
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1500709938
I have spent some more time unpacking this test.
Test setup is: shard `shard1` is split into 2 new slices `shard1_0` and
`shard1_1`, concurrently there are additions happening.
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1496270123
@dsmiley thanks for taking a look. I think we are looking at the same
problem area but different lifecycles.
Your summary on #1484 applies here as well:
```
A shard being split
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1493173156
updated the patch with a better fix. I'm now using the `isSubShardLeader`
flag which seems to have been introduced for the split shard scenario (not
sure). will be running some more tests
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1492480511
seeing some local failures due to the newly added 0 version check
```
org.apache.solr.search.TestRecovery >
testLogReplayWithReorderedDBQByAsterixAndChildDocs FAILED
java.l
stillalex commented on PR #1504:
URL: https://github.com/apache/solr/pull/1504#issuecomment-1492441639
I have pushed a tentative patch for review and more importantly for
discussion.
I have ran this test extensively (hundreds of beast jobs) and it looks like
this is the way to go, al
11 matches
Mail list logo