Re: SolrCloud Performance Issue
Shamik: You're right, the use of NOW shouldn't be making that much of a difference between versions. FYI, though, here's a way to use NOW and re-use fq clauses: http://searchhub.org/2012/02/23/date-math-now-and-filter-queries/ It may well be this setting: 1000 Every second (assuming you're indexing), you're throwing away all your top-level caches and executing any autowarm queries etc. And if you _don't_ have any autowarming queries, you may not be filling caches, an expensive process. Try lengthening that out to, say, a minute (6) or even longer and see if that makes a difference. If that's the culprit, you at least have a place to start. If that's not it, it's also possible you're seeing decompression. How many documents are you returning and how big are they? There's some anecdotal comments that the default stored field decompression for either a large number of doc or very large docs may be playing a role here. Try setting fl=id (don't return any stored fields). If that is faster, this might be your problem. queryResultCache is often not very high re: hit ratio. It's usually used for paging, so if your users aren't hitting the "next" page you may not hit many. Best, Erick On Sat, Oct 19, 2013 at 4:12 AM, Otis Gospodnetic < otis.gospodne...@gmail.com> wrote: > Hi, > > What happens if you have just 1 shard - no distributed search, like > before? SPM for Solr or any other monitoring tool that captures OS and > Solr metrics should help you find the source of the problem faster. > Is disk IO the same? utilization of caches? JVM version, heap, etc.? > CPU usage? network? I'd look at each of these things side by side and > look for big differences. > > Otis > -- > Solr & ElasticSearch Support -- http://sematext.com/ > SOLR Performance Monitoring -- http://sematext.com/spm > > > > On Fri, Oct 18, 2013 at 1:38 AM, shamik wrote: > > I tried commenting out NOW in bq, but didn't make any difference in the > > performance. I do see minor entry in the queryfiltercache rate which is a > > meager 0.02. > > > > I'm really struggling to figure out the bottleneck, any known pain > points I > > should be checking ? > > > > > > > > -- > > View this message in context: > http://lucene.472066.n3.nabble.com/SolrCloud-Performance-Issue-tp4095971p4096277.html > > Sent from the Solr - User mailing list archive at Nabble.com. >
Re: SolrCloud Performance Issue
Hi, What happens if you have just 1 shard - no distributed search, like before? SPM for Solr or any other monitoring tool that captures OS and Solr metrics should help you find the source of the problem faster. Is disk IO the same? utilization of caches? JVM version, heap, etc.? CPU usage? network? I'd look at each of these things side by side and look for big differences. Otis -- Solr & ElasticSearch Support -- http://sematext.com/ SOLR Performance Monitoring -- http://sematext.com/spm On Fri, Oct 18, 2013 at 1:38 AM, shamik wrote: > I tried commenting out NOW in bq, but didn't make any difference in the > performance. I do see minor entry in the queryfiltercache rate which is a > meager 0.02. > > I'm really struggling to figure out the bottleneck, any known pain points I > should be checking ? > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/SolrCloud-Performance-Issue-tp4095971p4096277.html > Sent from the Solr - User mailing list archive at Nabble.com.
Re: SolrCloud Performance Issue
I tried commenting out NOW in bq, but didn't make any difference in the performance. I do see minor entry in the queryfiltercache rate which is a meager 0.02. I'm really struggling to figure out the bottleneck, any known pain points I should be checking ? -- View this message in context: http://lucene.472066.n3.nabble.com/SolrCloud-Performance-Issue-tp4095971p4096277.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: SolrCloud Performance Issue
Thanks Primoz, I was suspecting that too. But then, its hard to imagine that query cache is only contributing to the big performance hit. The setting applies to the old configuration, and it works pretty well even with the query cache low hit rate. -- View this message in context: http://lucene.472066.n3.nabble.com/SolrCloud-Performance-Issue-tp4095971p4096123.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: SolrCloud Performance Issue
Query result cache hit might be low due to using NOW in bf. NOW is always translated to current time and that of course changes from ms to ms... :) Primoz From: Shamik Bandopadhyay To: solr-user@lucene.apache.org Date: 17.10.2013 00:14 Subject:SolrCloud Performance Issue Hi, I'm in the process of transitioning to SolrCloud from a conventional Master-Slave model. I'm using Solr 4.4 and has set-up 2 shards with 1 replica each. I've 3 zookeeper ensemble. All the nodes are running on AWS EC2 instances. Shards are on m1.xlarge and sharing a zookeeper instance (mounted on a separate volume). 6 gb memory is allocated to each solr instance. I've around 10 million documents in index. With the previous standalone model, the queries avg around 100 ms. The SolrCloud query response have been abysmal so far. The query response time is over 1000ms, reaching 2000ms often. I expected some surge due to additional servers, network latency, etc. but this difference is really baffling. The hardware is similar in both cases, except for the fact that couple of SolrCloud node is sharing zookeeper as well. m1x.large I/O is high, so shouldn't be a bottleneck as well. The other difference from old setup is that I'm using the new CloudSolrServer class which is having the 3 zookeeper reference for load balancing. But I don't think it has any major impact as the queries executed from Solr admin query panel confirms the slowness. Here are some of my configuration setup: 3 false 1000 1024 true 200 400 line xref draw line draw linelanguage:english lineSource2:documentation lineSource2:CloudHelp drawlanguage:english drawSource2:documentation drawSource2:CloudHelp 2 The custom request handler : explicit 0.01 velocity browse text/html;charset=UTF-8 layout cloudhelp edismax *:* 15 id,url,Description,Source2,text,filetype,title,LastUpdateDate,PublishDate,ViewCount,TotalMessageCount,Solution,LastPostAuthor,Author,Duration,AuthorUrl,ThumbnailUrl,TopicId,score text^1.5 title^2 IndexTerm^.9 keywords^1.2 ADSKCommandSrch^2 ADSKContextId^1 Source2:CloudHelp^3 Source2:youtube^0.85 recip(ms(NOW,PublishDate),3.16e-11,1,1)^2.0 text on 1 100 language Source2 DocumentationBook ADSKProductDisplay audience true text title 250 ShortDesc true default true false false 1 spellcheck One thing I've noticed is that the queryresultcache hit rate is really low, not sure our queries are always that unique. I'm using edismax and there's a recip(ms(NOW,PublishDate),3.16e-11,1,1)^2.0 , can this contribute ? Sorry about the long post, but I'm struggling to nail down the issue here, especially when queries are running fine in a master-slave environment with similar hardware and network. Any pointers will be highly appreciated. Regards, Shamik
SolrCloud Performance Issue
Hi, I'm in the process of transitioning to SolrCloud from a conventional Master-Slave model. I'm using Solr 4.4 and has set-up 2 shards with 1 replica each. I've 3 zookeeper ensemble. All the nodes are running on AWS EC2 instances. Shards are on m1.xlarge and sharing a zookeeper instance (mounted on a separate volume). 6 gb memory is allocated to each solr instance. I've around 10 million documents in index. With the previous standalone model, the queries avg around 100 ms. The SolrCloud query response have been abysmal so far. The query response time is over 1000ms, reaching 2000ms often. I expected some surge due to additional servers, network latency, etc. but this difference is really baffling. The hardware is similar in both cases, except for the fact that couple of SolrCloud node is sharing zookeeper as well. m1x.large I/O is high, so shouldn't be a bottleneck as well. The other difference from old setup is that I'm using the new CloudSolrServer class which is having the 3 zookeeper reference for load balancing. But I don't think it has any major impact as the queries executed from Solr admin query panel confirms the slowness. Here are some of my configuration setup: 3 false 1000 1024 true 200 400 line xref draw line draw linelanguage:english lineSource2:documentation lineSource2:CloudHelp drawlanguage:english drawSource2:documentation drawSource2:CloudHelp 2 The custom request handler : explicit 0.01 velocity browse text/html;charset=UTF-8 layout cloudhelp edismax *:* 15 id,url,Description,Source2,text,filetype,title,LastUpdateDate,PublishDate,ViewCount,TotalMessageCount,Solution,LastPostAuthor,Author,Duration,AuthorUrl,ThumbnailUrl,TopicId,score text^1.5 title^2 IndexTerm^.9 keywords^1.2 ADSKCommandSrch^2 ADSKContextId^1 Source2:CloudHelp^3 Source2:youtube^0.85 recip(ms(NOW,PublishDate),3.16e-11,1,1)^2.0 text on 1 100 language Source2 DocumentationBook ADSKProductDisplay audience true text title 250 ShortDesc true default true false false 1 spellcheck One thing I've noticed is that the queryresultcache hit rate is really low, not sure our queries are always that unique. I'm using edismax and there's a recip(ms(NOW,PublishDate),3.16e-11,1,1)^2.0 , can this contribute ? Sorry about the long post, but I'm struggling to nail down the issue here, especially when queries are running fine in a master-slave environment with similar hardware and network. Any pointers will be highly appreciated. Regards, Shamik
SolrCloud Performance Issue
Hi, I'm in the process of transitioning to SolrCloud from a conventional Master-Slave model. I'm using Solr 4.4 and has set-up 2 shards with 1 replica each. I've 3 zookeeper ensemble. All the nodes are running on AWS EC2 instances. Shards are on m1.xlarge and sharing a zookeeper instance (mounted on a separate volume). 6 gb memory is allocated to each solr instance. I've around 10 million documents in index. With the previous standalone model, the queries avg around 100 ms. The SolrCloud query response have been abysmal so far. The query response time is over 1000ms, reaching 2000ms often. I expected some surge due to additional servers, network latency, etc. but this difference is really baffling. The hardware is similar in both cases, except for the fact that couple of SolrCloud node is sharing zookeeper as well. m1x.large I/O is high, so shouldn't be a bottleneck as well. The other difference from old setup is that I'm using the new CloudSolrServer class which is having the 3 zookeeper reference for load balancing. But I don't think it has any major impact as the queries executed from Solr admin query panel confirms the slowness. Here are some of my configuration setup: 3 false 1000 1024 true 200 400 line xref draw line draw linelanguage:english lineSource2:documentation lineSource2:CloudHelp drawlanguage:english drawSource2:documentation drawSource2:CloudHelp 2 The custom request handler : explicit 0.01 velocity browse text/html;charset=UTF-8 layout cloudhelp edismax *:* 15 id,url,Description,Source2,text,filetype,title,LastUpdateDate,PublishDate,ViewCount,TotalMessageCount,Solution,LastPostAuthor,Author,Duration,AuthorUrl,ThumbnailUrl,TopicId,score text^1.5 title^2 IndexTerm^.9 keywords^1.2 ADSKCommandSrch^2 ADSKContextId^1 Source2:CloudHelp^3 Source2:youtube^0.85 recip(ms(NOW,PublishDate),3.16e-11,1,1)^2.0 text on 1 100 language Source2 DocumentationBook ADSKProductDisplay audience true text title 250 ShortDesc true default true false false 1 spellcheck One thing I've noticed is that the queryresultcache hit rate is really low, not sure our queries are always that unique. I'm using edismax and there's a recip(ms(NOW,PublishDate),3.16e-11,1,1)^2.0 , can this contribute ? Sorry about the long post, but I'm struggling to nail down the issue here, especially when queries are running fine in a master-slave environment with similar hardware and network. Any pointers will be highly appreciated. Regards, Shamik -- View this message in context: http://lucene.472066.n3.nabble.com/SolrCloud-Performance-Issue-tp4095940.html Sent from the Solr - User mailing list archive at Nabble.com.