Which version of MarkLogic is this? How many CPU cores are on each host?
How much RAM is on each host? How are you measuring CPU utilization? Try to identify a particular query that reproduces the behavior. If you have it, HOST:8002/dashboard may help by identifying long-running requests. The server status pages report similar information. Or you may have to add some simple logging with xdmp:elapsed-time. Once you have identified a test query, profile it to see where it is spending its time. As a wild guess, I suspect the combination of an expensive wildcard query term and an older release. That might give you some idea which queries to test, but it could be something completely different. -- Mike On 29 Sep 2013, at 10:48 , indar verma <[email protected]> wrote: > Hi All, > > We have a 5 nodes clustered environment and data is in TBs. Ml1 to Ml3 are > evaluator nodes & data nodes and Ml4 & Ml5 are data nodes only for now. > > Once loads increase the Ml4 CPU utilization going to become maximum and other > 4 nodes are normal around 30% utilization. > > Due to Ml4 CPU maxed out, page is not rendering, some queries are giving time > out. > > Also one thing, more data processing is not being happen in Ml4 than other > nodes. > > Could you please somebody help me to resolve this? what can be the root cause > for this problem? what should we improve? > > Thanking in advance!!! > > JJ > > _______________________________________________ > General mailing list > [email protected] > http://developer.marklogic.com/mailman/listinfo/general _______________________________________________ General mailing list [email protected] http://developer.marklogic.com/mailman/listinfo/general
