Re: High 100% CPU usage with SOLR 1.4.1
On Wed, Jun 15, 2011 at 2:21 PM, pravesh wrote: > I would need some help in minimizing the CPU load on the new system. Could > possibly NIOFSDirectory attributes to high CPU? Yes, it's a feature! The CPU is only higher because the threads aren't blocked on IO as much. So the increase in CPU you are seeing is a good thing, not a bad thing (i.e. the number of requests processed in a given number of CPU busy CPU cycles should be greater or equal to the old release). -Yonik http://www.lucidimagination.com
Re: High 100% CPU usage with SOLR 1.4.1
Hi Yonik, Thanx for the prompt reply. This is a relief :) Just 1 more question. Wouldn't the 100% CPU load would affect the system, as system process would starve for the CPU? I tried the load test 1st with 4-cores and then with 8-cores, still the CPU usage was reaching 100% We have index of about 32GB with 100+ fields indexed,18 fields stored & using an optimized index for search Thanx Pravesh -- View this message in context: http://lucene.472066.n3.nabble.com/High-100-CPU-usage-with-SOLR-1-4-1-tp3068667p3068778.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: High 100% CPU usage with SOLR 1.4.1
Yes, 100% CPU utilization will affect other processes, but you've created an artificial situation with your load testing, so I don't think it counts... What kind of cpu utilization do you see when you simulate your actual load rather than querying as fast as you can? That's a more relevant number I think.. Best Erick On Wed, Jun 15, 2011 at 2:41 PM, pravesh wrote: > Hi Yonik, > > Thanx for the prompt reply. This is a relief :) > > Just 1 more question. Wouldn't the 100% CPU load would affect the system, as > system process would starve for the CPU? > > I tried the load test 1st with 4-cores and then with 8-cores, still the CPU > usage was reaching 100% > We have index of about 32GB with 100+ fields indexed,18 fields stored & > using an optimized index for search > > Thanx > Pravesh > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/High-100-CPU-usage-with-SOLR-1-4-1-tp3068667p3068778.html > Sent from the Solr - User mailing list archive at Nabble.com. >
Re: High 100% CPU usage with SOLR 1.4.1
Or another way of saying this is - what is the maximum throughput you get from the system (qps / indexing speed, etc) since that is what you really (should) care about - and how does it compare to the previous setup? -Mike On 6/15/2011 3:52 PM, Erick Erickson wrote: Yes, 100% CPU utilization will affect other processes, but you've created an artificial situation with your load testing, so I don't think it counts... What kind of cpu utilization do you see when you simulate your actual load rather than querying as fast as you can? That's a more relevant number I think.. Best Erick On Wed, Jun 15, 2011 at 2:41 PM, pravesh wrote: Hi Yonik, Thanx for the prompt reply. This is a relief :) Just 1 more question. Wouldn't the 100% CPU load would affect the system, as system process would starve for the CPU? I tried the load test 1st with 4-cores and then with 8-cores, still the CPU usage was reaching 100% We have index of about 32GB with 100+ fields indexed,18 fields stored& using an optimized index for search Thanx Pravesh -- View this message in context: http://lucene.472066.n3.nabble.com/High-100-CPU-usage-with-SOLR-1-4-1-tp3068667p3068778.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: High 100% CPU usage with SOLR 1.4.1
Yes Erick, I did create an artificial load test with 30 users concurrently doing search (around 28000 samples of actual queries). With 1.4.1, the test completes within 3hrs without any failures (with SOLR1.2.1 it wouldn't match with this performance, i.e., in 3 hrs it could only do 9700 samples). My actual production load is much less than that(3hrs cycle is actually spans to 24 hrs on production). I will repeat this with actual load now. Thanx all 4 ur time :) Pravesh -- View this message in context: http://lucene.472066.n3.nabble.com/High-100-CPU-usage-with-SOLR-1-4-1-tp3068667p3070663.html Sent from the Solr - User mailing list archive at Nabble.com.