License errors

I'm getting lots of intermittent performance degradation on our SM 9.34 system.  It's been running for a few years now and consists of two servers load balanced, with Web Client and SRC setup and lots of external tools calling the REST and SOAP APIs to raise and close tickets.

Looking in the log file, I noticed the following:

96( 6172) 05/29/2017 10:47:19 RTE W License Module( Change Management ) usage exceeded its limit. Licensed: 210, Active:546
96( 6172) 05/29/2017 10:47:19 RTE W License Module( Request Management ) usage exceeded its limit. Licensed: 125, Active:569
96( 6172) 05/29/2017 10:47:19 RTE W License Module( Service Desk ) usage exceeded its limit. Licensed: 200, Active:565
96( 6172) 05/29/2017 10:47:19 RTE W License Module( Problem Management ) usage exceeded its limit. Licensed: 200, Active:583

We are licensed for 240 Floating and 75 Named login licenses. Running reportlic, shows :

Login License        17(      75)       106(     240)

Checking the logged in users via 'status' in SM returns the same numbers.

I'm concerned the Active count in the logs might be contributing to performance degradation.  My guess is it's due to all the tools we have that hit the REST/SOAP APIs that are closing the sessions properly.  I have a webservices_timeout value set to 60 in the sm.ini file and the 'status' command shows the expected list of users connected.  There is only 120ish, not 560 .  

Does anyone have any ideas how the list reported in the log file works and how to clear those properly without a restart?

Tags:

Parents
  • Hello!

    Try sm -reportlbstatus. There may be a problem with a specific servlet. See if there is one servlet with a very high number of connections, compared to the others. If this is the case, this could also explain the performance issues you are dealing with. Kill the problematic servlet, although you should keep in mind that anyone connected to that particular servlet will lose their connection.

Reply
  • Hello!

    Try sm -reportlbstatus. There may be a problem with a specific servlet. See if there is one servlet with a very high number of connections, compared to the others. If this is the case, this could also explain the performance issues you are dealing with. Kill the problematic servlet, although you should keep in mind that anyone connected to that particular servlet will lose their connection.

Children
  • Thanks for replying. Load Balancer doesn't show any abnormal amount of sessions. Most serverlets have 25/40 users on them.  The total of users in this report matches what 'status' shows in SM. It's just the module level tracking that is off:

     

    Incident Management(HelpDesk) 30( 75) 550( 200)
    Service Desk(HelpDesk) 25( 75) 556( 200)
    Problem Management(HelpDesk) 4( 75) 575( 200)
    Scheduled Maintenance(HelpDesk) 0( 75) 0( 200)

    Service Catalog 0( 290100) 0( 0)
    Change Management 44( 65) 538( 210)
    Request Management 0( 40) 560( 125)
    Service Level Management 0( 40) 11( 130)
    Contract Management 0( 35) 0( 118)
    Asset Contracts Management 0( 35) 11( 118)
    Knowledge Management 0( 40) 95( 120)
    Knowledge Management ESS 0( 10000) 0( 0)

    HP SCAuto SDK for Unix/Windows(SCAuto) Enabled

    Login License 14( 75) 105( 240)

    I may have resolved my perf. issues by clearing out the msg and syslog tables. No job was setup to empty older records, so the 2million in there wasn't helping anything.

  • We have a daily "background purge/archive" job as a SM schedule record, which deletes all the msglog and syslog records older than a week. It was a OOB schedule record and we haven't modified it a bit.

    Other logs to look for:

    https://docs.software.hpe.com/SM/9.51/Hybrid/Content/admin_tasks_guide/cleanup/purge_logs.htm