[ 
https://issues.apache.org/jira/browse/SOLR-8263?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15022474#comment-15022474
 ] 

Shalin Shekhar Mangar commented on SOLR-8263:
---------------------------------------------

[~rendel] - I went through the patch and it looks good.

If I understood the sequence of operations correctly -- the update log's 
recovery info (and hence the tlog file name and offset) is stored and the 
update log is initialised with these details. The old buffered tlog file is 
read and each and every operation starting from the offset is copied over to 
the new update logs. The old tlog file is discarded at this point. 
RecoverStrategy then proceeds to apply buffered updates as usual.

Can you please confirm or correct my understanding of the solution? If this is 
correct then we're good to commit this patch. I beasted the tests for a while 
and found no data loss.

The only negative to this solution is that the replica may have the same 
updates in the update log again (one copy as written on the leader and another 
copied over from the old buffered tlog) which means that they will be 
replicated twice (albeit harmlessly) to the peer cluster?

> Tlog replication could interfere with the replay of buffered updates
> --------------------------------------------------------------------
>
>                 Key: SOLR-8263
>                 URL: https://issues.apache.org/jira/browse/SOLR-8263
>             Project: Solr
>          Issue Type: Sub-task
>            Reporter: Renaud Delbru
>            Assignee: Erick Erickson
>         Attachments: SOLR-8263-trunk-1.patch, SOLR-8263-trunk-2.patch
>
>
> The current implementation of the tlog replication might interfere with the 
> replay of the buffered updates. The current tlog replication works as follow:
> 1) Fetch the the tlog files from the master
> 2) reset the update log before switching the tlog directory
> 3) switch the tlog directory and re-initialise the update log with the new 
> directory.
> Currently there is no logic to keep "buffered updates" while resetting and 
> reinitializing the update log.



--
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