Very long warmup query vs. frequent soft commit with new searcher

[Otis Gospodnetic <ot...@gmail.com>]

Hi,

What happens when one has a *single* veeeery long *warming* running
query that takes, say, 10 minutes, and a soft commit that opens a new
searcher happening every 1 minute?

Could one run into a situation where each soft commit triggers the
same long warming query, thus queueing them one after the other and
making the queue endless, so to speak?

Thanks,
Otis
--
Performance Monitoring * Log Analytics * Search Analytics
Solr & Elasticsearch Support * http://sematext.com/

Re: Very long warmup query vs. frequent soft commit with new searcher

[Shawn Heisey <so...@elyograg.org>]

On 11/15/2013 11:59 PM, Otis Gospodnetic wrote:
> What happens when one has a *single* veeeery long *warming* running
> query that takes, say, 10 minutes, and a soft commit that opens a new
> searcher happening every 1 minute?
> 
> Could one run into a situation where each soft commit triggers the
> same long warming query, thus queueing them one after the other and
> making the queue endless, so to speak?

That's essentially what happens, but it won't be an endless queue.
You'll eventually hit your configured maxWarmingSearchers value and Solr
will stop making new searchers until it is under the limit.

http://wiki.apache.org/solr/FAQ#What_does_.22exceeded_limit_of_maxWarmingSearchers.3DX.22_mean.3F

If your described 10 minute warming query is describing a real setup and
isn't exaggerated for discussion purposes, the system in question is
probably resource-starved.  The max heap and/or total system RAM would
likely be the best resource(s) to increase.

Thanks,
Shawn