lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jon Hawkesworth <jon.hawkeswo...@MEDQUIST.onmicrosoft.com>
Subject RE: solcloud; collection reload, core Statistics 'optimize now'
Date Fri, 26 Aug 2016 16:44:48 GMT
Many thanks for this, that's really useful.

We're feeding in documents all the time so makes sense that optimizing the index would just
be overhead.

We just have one collection that we care about at the moment so I can't see us using Reload
very often either.

Jon


-----Original Message-----
From: Erick Erickson [mailto:erickerickson@gmail.com] 
Sent: Friday, August 26, 2016 4:46 PM
To: solr-user
Subject: Re: solcloud; collection reload, core Statistics 'optimize now'

First of all, please have them pretty much ignore the cores admin page.
That's mostly a remnant of the non-SolrCloud days and largely is used for troubleshooting
and the like. Most of all, assuming your index changes reasonably frequently (i.e. less than
once a day) optimizing is unnecessary and should be avoided.

As far as the reload command on a collection, it finds all of the cores that make up a collection
and issues a core reload on all of them. This:
> reloads the config and schema files
> throws out all the cached data
> opens new searchers

There's no reason to reload unless you've changed the config files and pushed them to Zookeeper
in the normal course of events. For your ops people, reload should be about on par with "restart
Solr". Think of reloading a collection as bouncing the JVM except only for a single collection.

Best,
Erick

On Fri, Aug 26, 2016 at 12:47 AM, Jon Hawkesworth < jon.hawkesworth@medquist.onmicrosoft.com>
wrote:

> Hi,
>
>
>
> I'd like to understand a bit more about some of the admin options in 
> solrcloud admin interface.
>
>
>
> Can anyone point me at something which tells me what hit Reload for a 
> given collection actually does, whether it is safe to do at any time 
> and/or under what circumstances it should/shouldn't be used?
>
>
>
> Also, poking around the UI I noticed that if you select a core, on the 
> Overview page there is a Statistics panel and in it a button entitled 
> 'optimize now'.  Again I'd like to understand what this does, when it 
> should/shouldn't be used and whether optimising statistics is 
> something that should scheduled.
>
>
>
> The background to this is that I'm trying to provide operations team 
> members with instructions about what, if anything, needs to be done to 
> keep our production clusters in good working order.  Obviously my 
> preference is for things to be automatic where possible but if things 
> can't be automated then I want to be able to provide operations team 
> members clear guidance about what needs to be done and when and why.
>
>
>
> Many thanks,
>
>
>
> Jon
>
>
>
>
>
> *Jon Hawkesworth*
> Software Developer
>
>
>
>
>
> Hanley Road, Malvern, WR13 6NP. UK
>
> O: +44 (0) 1684 312313
>
> *jon.hawkesworth@mmodal.com <jon.hawkesworth@mmodal.com> 
> www.mmodal.com
> <http://www.medquist.com/>*
>
>
>
> *This electronic mail transmission contains confidential information 
> intended only for the person(s) named. Any use, distribution, copying 
> or disclosure by another person is strictly prohibited. If you are not 
> the intended recipient of this e-mail, promptly delete it and all 
> attachments.*
>
>
>
Mime
View raw message