I was hoping someone had seen a similar issue in the past. Normally the system will run find for 1-3 days before a spike occurs. By looking at the windows performance log, I can tell that it is an abrupt spike that never comes down. Each thread that spikes seems to take about 10% of the CPU. I have seen up to 7 threads that have spiked and have taken upwards of 70-80% of the CPU. Eventually the arserver will crash when this happens.
Nothing jumps out at me by looking at the arerror log. I have gone as far as setting up "log per thread" and enabling SQL, API, and Filter logging, then tracking down the exact time a spike occurs and on which thread, and cross referencing that to the SQL, API, and Filter logs for that thread, but still nothing jumps out at me. The API calls are the same ones that are called quite often but don't result in the same symptoms. I have tried several different combinations of min & max List threads. Currently the system has 9 min and 9 max list threads (this configuration was recommended in an AIE performance tuning white paper for a server with 3 processors). Max filters for an operation is 200000 and Max stack of filters is 10000. Is there any way to identify a filter loop other than by looking in the logs? Also another interesting thing to note is that actual end user performance does not seem to be affected, however the issue certianly detracts from overall system stability because the system crashes every couple of days. Also, when a spike has happened, any admin action such as saving a filter or form will usually crash the system. On Apr 17, 11:43 am, googerb <[EMAIL PROTECTED]> wrote: > Has anyone experienced issues with sustained, high CPU usage for the > arserver.exe process? > > Using: > ARServer V7.0.1 patch 006 > SQL server v9 > Windows Server 2003 Enterprise > 16 GB of RAM on server > 3 quad core processors on IBM 3850 xSeries > > Windows performance monitor with thread logging shows that some > arserver.exe list threads spike after an unknown amount of time and > never come down.The only way to resolve the CPU spike is to cycle the > arserver. Remedy support has been unable to identify a root cause. I > have had a ticket open with them for 6 months for the issue. Help! > Thanks! > > _______________________________________________________________________________ > UNSUBSCRIBE or access ARSlist Archives atwww.arslist.org > Platinum Sponsor:www.rmsportal.comARSlist: "Where the Answers Are" _______________________________________________________________________________ UNSUBSCRIBE or access ARSlist Archives at www.arslist.org Platinum Sponsor: www.rmsportal.com ARSlist: "Where the Answers Are"
