IT Monkey:   Click Here to Help Me Build the Agenda for Upcoming Cireson Webinars!

How to make SMA pick up more work faster

Stephane_BouillonStephane_Bouillon Customer Adept IT Monkey ✭✭
Hi,

My SMA server is sitting mostly at low CPU usage, but I notice that scripts (runbooks) are not executed immediately. Is there a way to configure the batch size and polling interval so that it processes more in one go and also poll faster ?

Stephane

Best Answers

Answers

  • Adam_DzyackyAdam_Dzyacky Customer Contributor Monkey ✭✭✭✭✭
    So there are two possible things happening here and the good news both should be within your control:

    From an SMA perspective, there really isn't a lot going on. It is genuinely a PowerShell repo of sorts waiting to be invoked. However when it comes to SMA performance there is a silent killer out there which is failed grooming jobs. By default, the SQL DB that SMA was installed to should also get a SQL Job deployed that ensures your SMA database is groomed daily and retains at most 30 days worth of jobs. If this job starts failing and isn't caught your SMA database balloons with data based on usage and will get worse as time goes on. You can check up on the "SMA Database Purge Job" history and make sure you aren't seeing errors on said SQL server.

    From an SCSM perspective, Activities created with the SMA connector and then placed within a Incident, Change, or Service Request are more subject to the SCSM workflow engine realizing the parent item is Active/In-Progress and then kicking off the corresponding Activity as opposed to SMA infrastructure. Because as you may or may not know, the general workflow for a SR/CR is something to the effect of:

    Parent SR/CR New, all Activities New
    Parent SR/CR flips to In-Progress, all Activities to Pending
    First Activity flips to In-Progress

    The SCSM workflow engine really only has two modes - on a schedule (e.g. every 30 seconds) or event based (e.g. Incident went from New -> Active). Event based triggers are more instantaneous and means you have less noise in SCSM since they aren't always running on a schedule. It's not a "a lot" of noise, but it is still technically more than nothing. Given that the SMA Connector is in a sealed management pack, our options seem to be limited. But given the entire SCSM architecture is taken directly from SCOM it means we should be able to create an Override for the Workflow and change how the values in a sealed MP are interpreted. Fortunately, there is such an article on the topic -

    https://blogs.technet.microsoft.com/servicemanager/2011/07/07/disabling-workflows-with-overrides/

    I myself have not done this for modifying a Workflow's schedule, but it feels like this should be possible.
  • Stephane_BouillonStephane_Bouillon Customer Adept IT Monkey ✭✭
    How can I verify the SMA Database Purge Job ?
  • Stephane_BouillonStephane_Bouillon Customer Adept IT Monkey ✭✭
    The reason I ased is because I noticed that when I have a SR with two SMA activities, it takes quite long for the first one to go to "in progress", and there is quite a long pause between the first one going to "Completed" and the second one to go to "in progress". Given that my server has capacity to spare, I don't think it's a performance issue and therefore I'd like to somehow fasten the cycles of progressing through the activities.
  • Stephane_BouillonStephane_Bouillon Customer Adept IT Monkey ✭✭
    Thanks, the last time it ran was on Feb 5th. It looked like the SQL Server Agent was not running. I started it, and then the purge job executed.

    However, out of curiosity I also checked my SCSM database server, and there the SQL Server Agent is not running either: it shows Agent XPs disabled should I start it ?
  • Stephane_BouillonStephane_Bouillon Customer Adept IT Monkey ✭✭
    Thanks Adam, I started the agent and ran all the scheduled jobs
Sign In or Register to comment.