[
https://issues.apache.org/jira/browse/CASSANDRA-1991?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Peter Schuller updated CASSANDRA-1991:
--------------------------------------
Attachment: 1991-trunk-v2.txt
Attaching a potential alternative that changes getContext() to
getDurableContext(), and makes the CommitLog keep a
CommitLogSegment.CommitLogContext() continuously up to date on every
synchronization to disk. Given that it only happens on disk synch, I'm not
worried about the cost of the measly allocation. This way of doing it means
that no additional concurrency semantics are required w.r.t keeping track of
the BRAF's current position.
The main semantic difference between this and trunk is that the per-column
family flush indicators will be *conservative* instead of precise; in other
words, data will have been flushed that is not recorded as having done so.
One consequence of this is that there may be more log replay than strictly
necessary on start-up; particularly in periodic commit mode.
What I'm not sure of is whether this will have any actual detrimental effects.
Is there anything relying on the flush positions not being conservative? If
not, I like this solution because of its simplicity.
> 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,
> 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.
