[ 
https://issues.apache.org/jira/browse/SOLR-9689?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15641712#comment-15641712
 ] 

Pushkar Raste commented on SOLR-9689:
-------------------------------------

[~ichattopadhyaya] - 

*  Even for normal in normal operations, updates for the leader can arrive at 
the replica in a different order and we already have a way to handle it.  We 
currently store 100 DBQs, to handle reordered updates. If reordered DBQs are 
detected, DBQs are applied along with a add, 
https://github.com/apache/lucene-solr/blob/master/solr/core/src/java/org/apache/solr/update/DirectUpdateHandler2.java#L201

* I think even for partial updates, corresponding full update is stored in the 
tlog. I don't think tlog ever stores partial updates like inc value of a field 
or set value of a field. It always contains entire document with updated values.

* I am creating a batch of only 100 updates and only 100 updates in the batch 
will be applied concurrently. I don't think there will be any issues. We can 
make size of  DBQ list in the DirectUpdateHandler2 configurable as well

> Process updates concurrently during PeerSync
> --------------------------------------------
>
>                 Key: SOLR-9689
>                 URL: https://issues.apache.org/jira/browse/SOLR-9689
>             Project: Solr
>          Issue Type: Improvement
>      Security Level: Public(Default Security Level. Issues are Public) 
>            Reporter: Pushkar Raste
>         Attachments: SOLR-9689.patch, SOLR-9689.patch2
>
>
> This came up during discussion with [~shalinmangar]
> During {{PeerSync}}, updates are applied one a time by looping through the 
> updates received from the leader. This is slow and could keep node in 
> recovery for a long time if number of updates to apply were large. 
> We can apply updates concurrently, this should be no different than what 
> could happen during normal indexing (we can't really ensure that a replica 
> will process updates in the same order as the leader or other replicas).
> There are few corner cases around dbq we should be careful about. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to