cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sanjeev Kulkarni <sanj...@locomatix.com>
Subject Re: Commitlog Disk Full
Date Fri, 13 May 2011 06:34:14 GMT
Hi Peter,
Thanks for the response.
I haven't explictly set a value for the memtable_flush_after_mins parameter.
Looks like the default is 60minutes.
I will try to play around this value to see if that fixes things.
Thanks again!

On Thu, May 12, 2011 at 11:41 AM, Peter Schuller <
peter.schuller@infidyne.com> wrote:

> > I understand that cassandra periodically cleans up the commitlog
> directories
> > by generating sstables in datadir. Is there any way to speed up this
> > movement from commitog to datadir?
>
> commitlog_rotation_threshold_in_mb could cause problems if it was set
> very very high, but with the default of 128mb it should not be an
> issue.
>
> I suspect the most likely reason is that you have a column family
> whose memtable flush settings are extreme. A commit log segment cannot
> be removed until the corresponding data has been flushed to an
> sstable. For high-throughput memtables where you flush regularly this
> should happen often. For idle or almost idle memtables you may be
> waiting on the timeout criteria to trigger. So in general, having a
> memtable with a long expiry time will have the potential to generate
> commit logs of whatever size is implied by the write traffic during
> that periods.
>
> The memtable setting in question is the "memtable_flush_after"
> setting. Do you have that set to something very high on one of your
> column families?
>
> You can use "describe keyspace name_of_keyspace" in cassandra-cli to
> check current settings.
>
> --
> / Peter Schuller
>

Mime
View raw message