[ 
https://issues.apache.org/jira/browse/SOLR-8030?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15220137#comment-15220137
 ] 

David Smiley commented on SOLR-8030:
------------------------------------

I'm still experiencing a problem and then I more carefully read the earlier 
comments here by Hoss etc.  I originally just thought the docs were added to 
the UpdateLog in DirectUpdateHandler2 (which makes sense to me) but now I see 
that DistributedURP has code paths to add to the UpdateLog too.  That's very 
confusing to me -- that we do it in two places; the code in DistributedURP is 
very confusing too.  And with custom URPs & URP chains, I'm sure there are 
gotchas like this (issue) waiting to surprise you.  Given DistributedURP adds 
the doc to the log, I can understand the desire to record the URP chain, since 
we need to know where to continue from on replay.  Perhaps DistributedURP's 
design could be changed to _not_ write to the UpdateLog?  That would add some 
sanity to being able to reason about things more easily and, I think, simplify 
the code.  But DistributedURP is writing to the UpdateLog now for some reason.  
[[email protected]] (or someone else familiar with the design) can you 
please comment?

As a practical matter right now given these issues, I think I'm going to have 
to have just one URP chain, and ensure each of the URPs can deal with the 
possibility of a doc flowing through twice.

> Transaction log does not store the update chain used for updates
> ----------------------------------------------------------------
>
>                 Key: SOLR-8030
>                 URL: https://issues.apache.org/jira/browse/SOLR-8030
>             Project: Solr
>          Issue Type: Bug
>          Components: SolrCloud
>    Affects Versions: 5.3
>            Reporter: Ludovic Boutros
>         Attachments: SOLR-8030.patch
>
>
> Transaction Log does not store the update chain used during updates.
> Therefore tLog uses the default update chain during log replay.
> If we implement custom update logic with multiple update chains, the log 
> replay could break this logic.



--
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