Oks I understand walter and what about the number of shards? In an index about 20GB must be a good number 2 shards or more? There are any way to calculate it?
The idea of avoiding distributed request was, we have all the data replicated in all the nodes, to ensure that the node that receive the request was the only node who works to response to it, but maybe was a wrong idea. Thanks a lot. ________________________________ David Martín Nieto E-mail: dmart...@viewnext.com | Web: www.viewnext.com ________________________________ [https://mail.google.com/mail/u/0?ui=2&ik=72317294cd&attid=0.0.2&permmsgid=msg-f:1662155651369049897&th=171129c229429f29&view=fimg&sz=s0-l75-ft&attbid=ANGjdJ_o0Ds8_P8d7W-csq2mmc6mBGQy9hSjXsGEv15RXUutalCYzg3HQB3CByE2swcJkH3yRaLwrXkr1G81F9FpfqcPlbpRoZcainmsJjviLoypusuKOxCnOw97zuo&disp=emb] ________________________________ De: Walter Underwood <wun...@wunderwood.org> Enviado: miércoles, 16 de noviembre de 2022 18:32 Para: users@solr.apache.org <users@solr.apache.org> Asunto: Re: Doubt about number of shards Do those limiters apply only to external requests or also to internal requests to shards? Killing a request to a shard kills the whole request, so it has a magnified effect, probably larger than intended. That is why I suggested using nginx to limit request rates before they hit the cluster. That is more predictable. wunder Walter Underwood wun...@wunderwood.org <mailto:wun...@wunderwood.org> http://observer.wunderwood.org/ (my blog) > On Nov 16, 2022, at 3:55 AM, Jan Høydahl <jan....@cominvent.com > <mailto:jan....@cominvent.com>> wrote: > > Also see the Ref Guide about Request Rate Limiting: > https://solr.apache.org/guide/solr/latest/deployment-guide/rate-limiters.html > <https://solr.apache.org/guide/solr/latest/deployment-guide/rate-limiters.html> > > Jan > >> 15. nov. 2022 kl. 12:49 skrev DAVID MARTIN NIETO <dmart...@viewnext.com >> <mailto:dmart...@viewnext.com>>: >> >> hello solr users >> >> We have a production cluster of six machines using solr 8.2 and I had a >> question about whether or not changing the number of shards could improve >> the performance of collection queries. Specifically, we have several >> collections, some of which are several gigabytes (20 GB, 10, 2 GB) and other >> very small collections that have 40% of the incoming queries. All the solr >> nodes have all de shards replicateds now. >> >> Any recommendations on how to improve performance? >> >> Additionally several questions: >> - Is there any way for Solr to discard requests that take more than X >> seconds to respond? >> - Is there any way to limit Solr's response rate, with the idea of not >> responding if more than 100 requests arrive per second, for example, and >> discarding the rest so as not to affect performance? >> - Is there any way for Solr to avoid distributed request between the six >> nodes of the cluster? >> >> Thank you very much for the help. >> All the best. >> >> >> ________________________________ >> David Martín Nieto >> E-mail: dmart...@viewnext.com <mailto:dmart...@viewnext.com> | Web: >> www.viewnext.com<http://www.viewnext.com> <http://www.viewnext.com/> >> ________________________________ >> >> >> >> >> --------------------------------------------------------------------------------------------------------------- >> Este mensaje va dirigido únicamente a la(s) persona(s) y/o entidad(es) >> arriba relacionada(s). Puede contener información confidencial o legalmente >> protegida. Si no es usted el destinatario señalado, le rogamos borre del >> sistema inmediatamente el mensaje y sus copias. Asimismo le informamos que >> cualquier copia, divulgación, distribución o uso de los contenidos está >> prohibida. >> --------------------------------------------------------------------------------------------------------------- >> This message is addressed only to the person (people) and / or entities >> listed above. It may contain confidential or legally protected information. >> If you are not the recipient indicated, please delete the message and its >> copies immediately from the system. We also inform that any copy, >> disclosure, distribution or use of the contents is forbidden >> --------------------------------------------------------------------------------------------------------------- >> Viewnext, S.A. Domicilio Social: Avda. de Burgos 8-A 28036 de Madrid. telf: >> 913834060, Fax: 913834090. Reg. M. Madrid: Tomo 3238, Libro:0, Folio: 78, >> Seccion: 8ª, Hoja M-55112, N.I.F.: A-80157746 >
