Three things.

1) make your runs last longer than a few seconds, a sustained load over a
longer period will average out some of jitter..

2) You don't have enough log lines.  Is this a single replica cluster?  How
does that work?

3) If you're benchmarking with the background daemons on, you might try
turning them off as they're going to introduce some unpredictability
between runs - it may be more realistic, but you'll need ever *longer* runs
to smooth out the spotty load over time.


On Fri, Aug 9, 2013 at 7:17 AM, Snider, Tim <tim.sni...@netapp.com> wrote:

>  When running ssbench  I’ve consistently gotten inconsistent results –
> 67% difference between the fastest and slowest CREATE times. Nothing
> (scenario, parameters, cluster, etc…) was varied between the runs.  I’m
> using tempAuth and not Keystone authentication.****
>
> ** **
>
> Ssbench was run on a node  using the internal (10GbE) network that isn’t
> currently part of the Swift cluster.****
>
> Is this commonly seen with Swift? ****
>
> CREATE       Count:   406 (    0 error;     0 retries:  0.00%)  Average
> requests per second:  79.6****
>
> CREATE       Count:   389 (    0 error;     0 retries:  0.00%)  Average
> requests per second:  99.8****
>
> CREATE       Count:   412 (    0 error;     0 retries:  0.00%)  Average
> requests per second:  65.2****
>
> CREATE       Count:   416 (    0 error;     0 retries:  0.00%)  Average
> requests per second:  58.6****
>
> CREATE       Count:   406 (    0 error;     0 retries:  0.00%)  Average
> requests per second:  41.6****
>
> CREATE       Count:   409 (    0 error;     0 retries:  0.00%)  Average
> requests per second:  71.3****
>
> CREATE       Count:   395 (    0 error;     0 retries:  0.00%)  Average
> requests per second:  32.3****
>
> ** **
>
> /var/log/swift entries for a PUT with a large latency doesn’t seem to
> reveal much:****
>
> root@swift21:/home/swift/bin# grep txeaf22b65bbc442d091373399a931340e
> /var/log/swift/*****
>
> /var/log/swift/container.log:Aug  9 06:47:40 swift21 container-server
> 192.168.10.208 - - [09/Aug/2013:13:47:40 +0000] "PUT
> /sdc/8738/AUTH_test/ssbench_000074/100k.1_000180" 201 -
> "txeaf22b65bbc442d091373399a931340e" "-" "-" 0.0005****
>
> /var/log/swift/object.log:Aug  9 06:47:40 swift21 object-server
> 192.168.10.209 - - [09/Aug/2013:13:47:40 +0000] "PUT
> /sdd/12336/AUTH_test/ssbench_000074/100k.1_000180" 201 - "-"
> "txeaf22b65bbc442d091373399a931340e" "-" 5.3650****
>
> ** **
>
> I’d like to investigate the possible source of the large result variation
> and requests with large latencies. Suggestions on where / what to look
> at/for are appreciated.****
>
> ** **
>
> Details in: http://paste.openstack.org/show/43703 ****
>
> ** **
>
> Thanks,****
>
> Tim****
>
> ** **
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev@lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
_______________________________________________
OpenStack-dev mailing list
OpenStack-dev@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Reply via email to