[
https://issues.apache.org/jira/browse/ACCUMULO-3617?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14367679#comment-14367679
]
Josh Elser commented on ACCUMULO-3617:
--------------------------------------
IMO, is there a purpose to auditing unloadTablet with the understand that the
master will repeatedly spam it until the server actually responds to the
request? Without more information, I don't think it makes sense to log it at
all. If it actually included the extent in question, I could see it at TRACE.
Given that it's in a checkPermission block, I think the intent was not to debug
the fact that the server saw an unloadMessage, but more that it allowed an
unloadMessage which, IMO, is useless.
> unloadTablet log message in TabletServer is spammy w/o purpose
> --------------------------------------------------------------
>
> Key: ACCUMULO-3617
> URL: https://issues.apache.org/jira/browse/ACCUMULO-3617
> Project: Accumulo
> Issue Type: Improvement
> Components: tserver
> Affects Versions: 1.5.2, 1.6.2
> Reporter: Josh Elser
> Priority: Minor
> Labels: newbie
> Fix For: 1.7.0
>
>
> Looking at TabletServer logs, the following message can be overwhelming:
> {noformat}
> 2015-02-24 01:57:34,723 [tserver.TabletServer] DEBUG: Got unloadTablet
> message from user: !SYSTEM
> {noformat}
> This message alone tells us nothing (no context about the tablet(s) requested
> to be unloaded, only that there was a request to unload a tablet), and can
> appear in the log upwards of 20 times a millisecond.
> It might be better if the TabletServer did...
> * Batched log messages "N unloadTablet requests in $duration"
> * Include extent to unload
> * Kill the log message entirely.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
