Workflow Soluiton

 View Only
  • 1.  w3wp process spikes the CPU at 100% utilization

    Posted Sep 21, 2009 03:02 PM
    I have a published Workflow project that we've had in place since this past June. When it is running, it is causing the w3wp process to spike the CPU utilization at 100% for the majority of the time with some fairly regular dips (see image below)

    TaskMan143.PNG

    The project has 4 Wait (Merge) components due to the need for some parallel logic flows in the project. I ran the logger for two minutes against this project and when the log was complete, there were over 8,400+ entries for "storing object as XML" and 2,500+ entries for "Trigger: ESCALATION date: <date><time> processed: True" (where <date> is some date and <time> is some time - Example: "Trigger: ESCALATION date: 7/7/2009 9:58:41 AM processed: True"). All the "dates" listed for those 2,500+ "Trigger: ESCALATION ..." entries are between 06/30/09 and 07/07/09.

    Right now there are only 12 active processes and 14 active tasks associated with this project when viewing the workflowmanagement page (I'm using Workflow Basic 6.5) so I'm wondering why the log file shows 8,400+ entries for "storing object as XML" and 2,500+ entries for "Trigger: ESCALATION date: <date><time> processed: True".

    Can anyone provide some ideas on what I should do to resolve the high CPU utilization for this project? I did put the logging tool level back to just 'Error' for this project so it shouldn't be the logging tool causing the issue. Thanks for any ideas/suggestions!


  • 2.  RE: w3wp process spikes the CPU at 100% utilization
    Best Answer

    Posted Sep 22, 2009 02:24 PM
    Started going through the files in the \Workflow Designer\\Data\MQWorkflowFileStorage\localworkflowfilestorage-<guid>.tasks folder and I found some old tasks that were not showing up in the WorkflowManagement folder but were nonetheless still active and being checked. Once I deleted these old tasks and restarted server extensions, the CPU utilization went down to a spikes of between 50-70% with durations of less than a second (see attached). Not sure why those old tasks were still there and not in the WorkflowManagement page. Situation seems resolved for now though!

    TaskMan_09-22.PNG