Implement Delete-by-Query operation after Reshard #125519
Conversation
…eteByQuery Refresh
…eteByQuery Refresh branch
elasticsearchmachine
added
v9.1.0
serverless-linked
…eteByQuery Refresh branch
elasticsearchmachine
added
the
needs:triage
ankikuma
added
:Distributed Indexing/Distributed
elasticsearchmachine
removed
the
needs:triage
|
Pinging @elastic/es-distributed-indexing (Team:Distributed Indexing) |
|
|
||
| public void deleteByQuery(ShardSplittingQuery query) throws Exception { | ||
| // System.out.println("Delete documents using ShardSplitQuery"); | ||
| indexWriter.deleteDocuments(query); | ||
| indexWriter.flush(); | ||
| indexWriter.commit(); | ||
| } |
There was a problem hiding this comment.
I think we can probably do this in stateless and avoid the need to publish this API here while we're incubating.
There was a problem hiding this comment.
also I'd probably move the flush/commit out of the delete operation itself and let something higher up decide when it needs to schedule those.
There was a problem hiding this comment.
There was a problem hiding this comment.
By the way, I don't know if the flush and commit at this level is sufficient to allow us to move the split on this shard to DONE. When we do that, we're also going to drop the search filter for unowned documents, which means the search nodes need to be using the state we've just flushed. I think that implies we should be doing a refresh.
No description provided.