cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jonathan Ellis (JIRA)" <j...@apache.org>
Subject [jira] Updated: (CASSANDRA-1991) CFS.maybeSwitchMemtable() calls CommitLog.instance.getContext(), which may block, under flusher lock write lock
Date Sun, 16 Jan 2011 04:00:47 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-1991?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Jonathan Ellis updated CASSANDRA-1991:
--------------------------------------

    Attachment: 1991-v4.txt

Yes, drain does need to include all the writes in what it marks as applied (and so does RecoveryManager2Test,
if run with periodic mode commitlog sync).  My initial analysis was incorrect; the old getContext
did provide this because the flusher writelock would prevent any other tasks on the commitlog
executor while it is held.

Renamed CommitLog.sync to syncInternal and added a threadsafe sync for drain and RM2T to call.

> 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,
1991-v4.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.


Mime
View raw message