lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ryan Zezeski (JIRA)" <>
Subject [jira] [Commented] (SOLR-4509) Disable Stale Check - Distributed Search (Performance)
Date Wed, 27 Feb 2013 19:21:13 GMT


Ryan Zezeski commented on SOLR-4509:

> Would the 100ms latency you have described here show up in QTime values or be purely

So my latency measurements include end-to-end from client->yokozuna->solr coordinator->shards->solr
coordinator->yokozuna->client.  I didn't track the QTimes, but in benchmarks I ran months
ago I saw the same results whether I went through Yokozuna or hit Solr directly with the distributed
search.  I could probably re-run to report QTime changes.

> I ask because the median (not average)...

I'm also reporting median latency, not mean.  In my first link you can also find graphs with
95th, 99th, 99.9 percentiles.

> assumption of a server timeout of 50 milliseconds

The assumption is 50s, but yes, it is an assumption.  A better patch would make the idle timeout
configurable as well as the period.

> If these are not actual worries, then I can be quiet.

No, these are great points.  This patch, as it sits now, is for a specific context.  I did
what I needed to do to improve my specific situation.  Honestly, I would like to see another
person run this patch and verify at least some amount of speed-up.  Otherwise, it might best
stay downstream.
> Disable Stale Check - Distributed Search (Performance)
> ------------------------------------------------------
>                 Key: SOLR-4509
>                 URL:
>             Project: Solr
>          Issue Type: Improvement
>          Components: search
>         Environment: 5 node SmartOS cluster (all nodes living in same global zone - i.e.
same physical machine)
>            Reporter: Ryan Zezeski
>            Priority: Minor
>         Attachments: SOLR-4509.patch
> By disabling the Apache HTTP Client stale check I've witnessed a 2-4x increase in throughput
and reduction of over 100ms.  This patch was made in the context of a project I'm leading,
called Yokozuna, which relies on distributed search.
> Here's the patch on Yokozuna:
> Here's a write-up I did on my findings:
> I'm happy to answer any questions or make changes to the patch to make it acceptable.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message