cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Ellis <jbel...@gmail.com>
Subject Re: Persistently increasing read latency
Date Thu, 03 Dec 2009 22:06:10 GMT
if you have a backlog of several hundred files it will take a while to
get back down to normal.

On Thu, Dec 3, 2009 at 4:00 PM, B. Todd Burruss <bburruss@real.com> wrote:
> i saw those too, but the number of files don't decrease, should they?
> and the read performance still isn't good.
>
> i'll run the test again tonight with the latest from trunk
>
>
> On Thu, 2009-12-03 at 15:56 -0600, Jonathan Ellis wrote:
>> the lines like
>>
>> INFO [COMPACTION-POOL:1] [ColumnFamilyStore.java:932] Compacted to
>> /home/bburruss/cassandra.884232-0.5.0-beta1/btoddb/data/uds/bucket-443-Data.db.
>>
>> would seem to indicate that it's working.
>>
>> On Thu, Dec 3, 2009 at 3:00 PM, B. Todd Burruss <bburruss@real.com> wrote:
>> > correct.  i just bounced again, ran:
>> >
>> >   bin/nodeprobe -host gen-app02.dev.real.com -port 8080 compact
>> >
>> > this time i waited long enough to let it work.  a lot of logging went by
>> > so i've attached it.  this log covers starting the server until after
>> > nodeprobe was run.
>> >
>> > lemme know you got it.
>> >
>> >
>> >
>> > On Thu, 2009-12-03 at 14:51 -0600, Jonathan Ellis wrote:
>> >> On Thu, Dec 3, 2009 at 2:33 PM, B. Todd Burruss <bburruss@real.com>
wrote:
>> >> > nothing - i have DEBUG level set and bounced the server.  i'll restart
>> >> > again.
>> >>
>> >> just to verify: you bounced, then did nodeprobe compact, and nothing was
logged?
>> >>
>> >> > cassandra spends a lot of time loading INDEXes
>> >>
>> >> that's normal, especially w/ a lot of uncompacted sstables
>> >
>> >
>
>
>

Mime
View raw message