[
https://issues.apache.org/jira/browse/CASSANDRA-813?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jonathan Ellis updated CASSANDRA-813:
-------------------------------------
Attachment: 813-v2.txt
cleaned up some.
unsolved problem: do we want to log ParNew as debug, while leaving
ConcurrentMarkSweep at INFO? ParNew is pretty aggressive in my testing,
spamming the logs with that isn't very useful. But, people may experiment w/
different gc types (e.g. G1) so hardcoding this seems Bad. But maybe we should
let the first person who wants to experiment w/ G1 deal with that. :)
semi-related interesting thread from Voldemort on GC settings:
http://groups.google.com/group/project-voldemort/browse_thread/thread/24230399a0d34dc8/d893fb5bcaf0d173?#d893fb5bcaf0d173
> add some kind of logging of GC activity
> ---------------------------------------
>
> Key: CASSANDRA-813
> URL: https://issues.apache.org/jira/browse/CASSANDRA-813
> Project: Cassandra
> Issue Type: Task
> Reporter: Jonathan Ellis
> Assignee: Brandon Williams
> Priority: Minor
> Fix For: 0.6
>
> Attachments: 0001_log_gc_stats.txt, 813-v2.txt
>
>
> "help, my node froze when i hammered it too hard" is a faq. before we have
> CASSANDRA-685 it would be nice to make it obvious what is going on w/ gc
> storms.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
