[prev in list] [next in list] [prev in thread] [next in thread] 

List:       solr-user
Subject:    Re: Solr resource recommendations
From:       ufuk_yılmaz <uyilmaz () vivaldi ! net ! INVALID>
Date:       2024-05-11 8:37:30
Message-ID: AF998CDB-ACBE-4B6D-B045-3469D307BC3D () vivaldi ! net
[Download RAW message or body]

Hi Damjan,

Just a reminder, you need to reindex if you change an existing field definition to \
have docValues, since they are created and written to disk during indexing. 

Sincerely,
Ufuk

—

> On May 11, 2024, at 4:52, Damjan Cvetan <damjan@cvetan.si.invalid> wrote:
> 
> Hi Ufuk,
> 
> Thank you for pointing me to the docValues. I have some fields there added
> later in the project used for faceting and not having docValues enabled.
> I'll try that out.
> 
> I'm not using the JSON Facet API but a traditional faceting approach. I see
> there are facet.method options (enum, fc, fcs), I'm using the fc option as
> default. I see that JSON Facet API provides more methods. I'll test them
> out on my collection before doing any code changes in the project.
> 
> I'll report back with my findings. Thank you.
> 
> Regards,
> Damjan
> 
> On Fri, 10 May 2024 at 19:24, ufuk yılmaz <uyilmaz@vivaldi.net.invalid>
> wrote:
> 
> > Hi Damjan,
> > 
> > Do you make proper use of docvalues in your schema?
> > 
> > Also for faceting, did you check out the "method" parameter of the json
> > facet api?
> > 
> > Those two made the biggest impact for me in a situation similar to yours
> > in the past.
> > 
> > Regards,
> > Ufuk
> > —
> > 
> > > On May 10, 2024, at 17:29, Damjan Cvetan <damjan@cvetan.si.invalid>
> > wrote:
> > > 
> > > Dear Solr users,
> > > 
> > > I'm reaching out for any recommendation on efficient resource needs for
> > the
> > > collection I'm running. I hope some of you can share some info on your
> > > resources and collection size which I can compare to.
> > > 
> > > I'm trying to optimize the setup with various settings as suggested
> > within
> > > "Solr optimization" articles but I'm wondering how much CPUs, RAM and
> > Solr
> > > nodes should have been enough.
> > > 
> > > I have 2 collections, each of 175 GB in size where the average
> > > document size is 180kB. One collection is using stemming filters while
> > the
> > > other does not. I'm running all of it on 2 nodes, each having 6 CPUs and
> > > 80GB of RAM. Collections are split into 2 shards, one shard per node. I'm
> > > doing query, filter query, faceeting, sorting and range queries. It was
> > all
> > > working fine but last year the collection got bigger and response time
> > got
> > > up to 30s.
> > > 
> > > Thank you all.
> > > 
> > > Regards,
> > > Damjan
> > 
> > 


[prev in list] [next in list] [prev in thread] [next in thread]