Automic Workload Automation

 View Only
  • 1.  V12.3.0 AWI/Tomcat support question

    Posted Oct 21, 2019 05:57 PM
    We implemented V12.3.0 into production on 10/19/2019, and we are really happy with it.  This version has far fewer bugs that impact our day to day operations than V11.2.1.  However being that today (10/21/2019) is our first full business day running on the AWI, I'm beginning to wonder a bit about its stability, and hoping to collect some wisdom from others here since I've never supported a tomcat solution before.

    Part way through the afternoon everyone's AWI locked up and I had to recycle tomcat.  Does this happen to others?  How does one mitigate this issue?   Do we need to schedule a tomcat recycle to occur periodically?  Or are there some tomcat settings I should be looking into?

    We implemented DEV several weeks ago and its tomcat seems to be running fine, but has less challenging work to perform.

    Our installation is supported by;
       Window2016
       SQLServer2017
       Tomcat9
       Java11.0.4
    We are running everything from the same server with a CPU that is usually below 10% and lots of RAM headroom. (We are a small shop, and keeping everything on one server simplifies DR)

    ------------------------------
    Pete
    ------------------------------


  • 2.  RE: V12.3.0 AWI/Tomcat support question

    Posted Oct 21, 2019 07:39 PM
    Ahhh... never mind.

    I found a local tomcat expert to look at this issue for me.  It was running out of RAM and we have been configured to only allow for a very small amount.  We are bumping it way up.

    ------------------------------
    Pete
    ------------------------------



  • 3.  RE: V12.3.0 AWI/Tomcat support question
    Best Answer

    Posted Oct 22, 2019 05:03 AM
    Was just about to suggest this.

    Java heap (the -Xmm ​style parameters to the JVM) is the only thing we ever bumped into really. Should give it a few gigabyte, I think we're giving it a maximum of 24 at this time.

    Apart from that, Tomcat and AWI is rock solid for us, especially since 12.3. I think we had only one true node freeze over something like 18 months.

    That is, however, without considering engine or client lockups. Users can still use the AWI to happily saturate CP processes and DB connections and lock up the CP, thus locking everyone out of AWI. And they can use the Process Monitoring filters to lock themselves individually out of AWI quite severely. It's not entirely free of some facepalm design.

    Br,