[jira] [Commented] (OAK-4371) Overly zealous warning about checkpoints on compaction

["Davide Giannella (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/OAK-4371?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15288964#comment-15288964 ] 

Davide Giannella commented on OAK-4371:
---------------------------------------

as of oak 1.3.something we added the ability to configure the async
index to run on different threads (lanes).

So it will have arbitrary multiple checkpoints depending on how many
threads are configured for the async index.

There's no single central point for the configuration, it will
depends by what you're going to set in {{async=<threadname>}}. Of
course it could change over the time.

I think the ideal check, for the log, is to check whether there are
multiple checkpoints associated to the same async thread. That is an
hint of problems.



> Overly zealous warning about checkpoints on compaction 
> -------------------------------------------------------
>
>                 Key: OAK-4371
>                 URL: https://issues.apache.org/jira/browse/OAK-4371
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: segment-tar, segmentmk
>            Reporter: Michael Dürig
>              Labels: compaction, gc, logging
>             Fix For: 1.6
>
>
> {{FileStore.compact}} logs a warning {{TarMK GC #{}: compaction found {} checkpoints, you might need to run checkpoint cleanup}} if there is more than a single checkpoints. 
> AFIK this is now the norm as async indexing has uses 2 checkpoints ([~chetanm], [~edivad] please clarify). 
> In any case should we improve this and not hard code any number of expected checkpoints. Maybe make the threshold configurable?



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