shared memory

Hi,

We have observed that everyday the free shared memory keeps on reducing for unknown reasons in our PROD batch region. There are other regions in our environment in which the free shared memory doesnt change. Due to this issue we are forced to recycle the region every week atleast so the free memory comes back to normal (114,216) k,

What could be the reason ?

 

Attached screenshot which is of todays

 

 

  • Have you opened an incident with support. If not that would be the best place to start


  • No steve , incident is not yet opened. But do you have an idea what could be like if any batch job is holding the memory.

  • That's why support will need to work with you to turn on appropriate diagnostics to see what may be happening. The screenshot mentioned in the original post was not attached anywhere I could see. Is it possible you forgot to upload it.



    You may also want to pull down the MFESDIAGS tool from the support website. The run it in the region. It will gather all lthe different diagnostic files support will likely need wot help determine what is happening.

    https://supportline.microfocus.com/examplesandutilities/troubleshootingpack/MFESDIAGS.zip


  • I have attached , which is of today's shared memory reading. region recycle was performed 4 days back.

    We will be doing region recycle again after 2 days since earlier we werent aware of our environment and because of that we were missing to do weekly recycling and that was leading to region hung issues.

     

  • It's hard to diagnosis a possible problem without any context or diagnostic artifacts. Let's start with:

    What operating system?

    What product version?

    You mention the region hangs and for some reason you seen to equate this with Share Memory use, what has lead you to that conclusion?

    Have you ever seen short on storage messages or share memory cushion released messages in the console logs, in particular when the region hangs?

    What do you mean when you say the regions hangs? You can't get in via the web interface, your scheduler can't submit jobs,.....?

    You mention this is the batch region, are there other online regions? How many, are you sharing a catalog, data, using Fileshare?

    About how many jobs per day are submitted? How many run concurrently?

    What security is being used, how many rules?

    How many abends occur daily if any?

    How did you arrive at a perceived need to allocate 30,000 (120M) of shared memory in a batch region, than number seems excessive on the surface since much of what goes into the calculation is driven by use in an online environment.

    https://www.microfocus.com/documentation/enterprise-developer/ed60/ED-Eclipse/BKCACACONFS011.html





    As Michael mentioned in another of you posts region recycles depend very much on workload, I've run product environments where we recycled about one a month to coincide with Windows patch and had no issues. While it possible this may stop the region hangs, you are really just making the problem. If you are serious about trying to identify the cause and resolving it, opening a support incident is the way to go


  • What operating system?
    windows server 2016 hosted in AWS cloud.

    What product version?
    4.0 PU14

    You mention the region hangs and for some reason you seen to equate this with Share Memory use, what has lead you to that conclusion?
    We are getting shared memory constraint messages in the console log at the time of region going to hung.


    Have you ever seen short on storage messages or share memory cushion released messages in the console logs, in particular when the region hangs?
    We used to get shared memory constraint ended messages also along with shared memory constraint but still the region goes to hung after sometime.


    What do you mean when you say the regions hangs? You can't get in via the web interface, your scheduler can't submit jobs,.....?
    The region doesnt respond any of the mouse clicks. It keeps refreshing and then finally batch region goes hung.


    You mention this is the batch region, are there other online regions? How many, are you sharing a catalog, data, using Fileshare?
    Yes we have 1 online region.  the fileshare is configured on the same server.


    About how many jobs per day are submitted? How many run concurrently?
    1000 . Some are concurrent.

    How did you arrive at a perceived need to allocate 30,000 (120M) of shared memory in a batch region, than number seems excessive on the surface since much of what goes into the calculation is driven by use in an online environment.
    It was suggested by the Support team to avoid shared memory constratint issues..

  • Issue is still persists, can somebody help?

  • Issue is still persists. Can somebody help?

  • Suggested Answer

    Hi Noel, I recommend you to collate the above information and raise a support case to seek further assistance. My colleagues can then review the diagnostics collection and provide feedback on their findings via the support case.