[
https://issues.apache.org/jira/browse/CASSANDRA-1991?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12982186#action_12982186
]
Peter Schuller commented on CASSANDRA-1991:
-------------------------------------------
Looks good to me - except v3 retains an unnecessary addition of the
AtomicReference import. Agreed regarding the lack of a need to update context
on new segment (becomes extra obvious when the contract of getContext() is put
in the way it is in your comment).
Regarding it not already being entirely precise - yes. However there would be
an observable difference with anything that relied on strict preciseness in the
event that incoming writes are lacking. That makes nodetool 'drain' a potential
candidate for regression. I did look at how it was implemented and it seemed
not to interact with the commit log.
I wonder if I'm missing something though. Would this in any way break e.g.
upgrades between versions that require 'drain' due to commit log compatibility
reasons? I think I recall reading some instructions somewhere that seemed to
imply that part of the contract of 'drain', or maybe 'flush', was to make
obsolete the commit log so that reply would not be attempted on restart.
> CFS.maybeSwitchMemtable() calls CommitLog.instance.getContext(), which may
> block, under flusher lock write lock
> ---------------------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-1991
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1991
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Peter Schuller
> Assignee: Peter Schuller
> Attachments: 1991-logchanges.txt, 1991-trunk-v2.txt, 1991-trunk.txt,
> 1991-v3.txt, trigger.py
>
>
> While investigate CASSANDRA-1955 I realized I was seeing very poor latencies
> for reasons that had nothing to do with flush_writers, even when using
> periodic commit log mode (and flush writers set ridiculously high, 500).
> It turns out writes blocked were slow because Table.apply() was spending lots
> of time (I can easily trigger seconds on moderate work-load) trying to
> acquire a flusher lock read lock ("flush lock millis" log printout in the
> logging patch I'll attach).
> That in turns is caused by CFS.maybeSwitchMemtable() which acquires the
> flusher lock write lock.
> Bisecting further revealed that the offending line of code that blocked was:
> final CommitLogSegment.CommitLogContext ctx =
> writeCommitLog ? CommitLog.instance.getContext() : null;
> Indeed, CommitLog.getContext() simply returns currentSegment().getContext(),
> but does so by submitting a callable on the service executor. So
> independently of flush writers, this can block all (global, for all cf:s)
> writes very easily, and does.
> I'll attach a file that is an independent Python script that triggers it on
> my macos laptop (with an intel SSD, which is why I was particularly
> surprised) (it assumes CPython, out-of-the-box-or-almost Cassandra on
> localhost that isn't in a cluster, and it will drop/recreate a keyspace
> called '1955').
> I'm also attaching, just FYI, the patch with log entries that I used while
> tracking it down.
> Finally, I'll attach a patch with a suggested solution of keeping track of
> the latest commit log with an AtomicReference (as an alternative to
> synchronizing all access to segments). With that patch applied, latencies are
> not affected by my trigger case like they were before. There are some
> sub-optimal > 100 ms cases on my test machine, but for other reasons. I'm no
> longer able to trigger the extremes.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
