Layer 7 Access Management

Expand all | Collapse all

Policy Server STATS

Jump to Best Answer
  • 1.  Policy Server STATS

    Posted 07-10-2019 03:34 AM
    [23646/91][Mon Jul 08 2019 15:15:01][CServer.cpp:4636][INFO][sm-Server-02000] System Statistics
    [23646/91][Mon Jul 08 2019 15:15:01][CServer.cpp:4640][INFO][sm-Server-02010] Available file descriptors: 4096
    [23646/91][Mon Jul 08 2019 15:15:01][CServer.cpp:4653][INFO][sm-Server-02020] Thread pool limit: 54
    [23646/91][Mon Jul 08 2019 15:15:01][CServer.cpp:4673][INFO][sm-Server-02030] Thread pool: Msgs=91073937 Waits=41590741 Misses=233924149 Max HP Msg=331 Max NP Msg=115 Current Depth=307 Max Depth=334 Current High Depth=307 Current Norm Depth=0 Current Threads=54 Max Threads=54
    [23646/91][Mon Jul 08 2019 15:15:01][CServer.cpp:4681][INFO][sm-Server-02040] Connections: Current=1385 Max=1736 Limit=3328 Exceeded limit=0

    Hi all,
    i am reviewing stats on one of the policy server in the customer production environments, and in particular i am looking at:
    Msgs= 91073937 (about 91M)
    Waits= 41590741 (about 41M)
    Misses= 233924149 (about 233M)

    Due to the huge number of Misses, probably the 54 threads are too much.
    Moreover, what sounds strange to me is the big difference (about 50 millions) between the Msgs and the Waits. I read somewhere that a policy server normal behaviour is number of Msgs about the same number of Waits coupled with a low number of Waits.
    So how came up to this big difference?
    I know that If a worker went to get a request from the queue to process, but had to wait for one to arrive , the Waits is increased by one.
    But, in case the worker went to the queue and immediatly (no wait) find a request to process, is the request processed and the Waits is NOT updated? Am i right?
    Otherwise, what could be another reason to explain the big difference between Msgs and Waits?

    Best Regards
    Claudio



  • 2.  RE: Policy Server STATS
    Best Answer

    Posted 07-11-2019 02:57 AM
    Hi Claudio,
     
    You mentioned :
     
    Msgs= 91073937 (about 91M)
    Waits= 41590741 (about 41M)
    Misses= 233924149 (about 233M)
     
    You asked :
     
    "what sounds strange to me is the big difference (about 50 millions)
    between the Msgs and the Waits. I read somewhere that a policy server
    normal behaviour is number of Msgs about the same number of Waits
    coupled with a low number of Waits. So how came up to this big
    difference?"
     
    The difference is that the Policy Server recieves too few requests for
    the amount of threads it runs. When a thread comes to check if there's
    a request, and it gets a request, the Waits will not be updated.
     
    "Msgs=91,073,937" - This is the total number of requests that have been
    received.
     
    "Waits=41,590,741" - This is the number of times that a thread went to
    get a request from the queue to process, but had to wait for one to
    arrive.
     
    "Misses=233,924,149" - This is the number of times that a thread went to get
    a request from the queue to process, had to wait for one to arrive,
    but none came in before the thread went back to sleep.
     
    So there 49483196 requests that have been served immediatly
    (91073937 - 41590741).
     
    I hope this helps,
     
    Best Regards,
    Patrick


  • 3.  RE: Policy Server STATS

    Posted 07-11-2019 09:02 AM
    Hi Patrick thank you for your explanation.
    I was not sure on how to read the data, because on other post it is stated that Messages and Waits should be about the same, so i was worrying some messages went lost.
    This is not the case, and so i can conclude that "policy threads" are overdimensioned, but apart from this policy server is working as expected.
    Best Regards
    Claudio


  • 4.  RE: Policy Server STATS

    Posted 07-11-2019 09:02 AM
    Hi Patrick thank you for your explanation.
    I was not sure on how to read the data, because on other post is stated that Messages and Waits should be about the same, so i was worrying some messages went lost.
    This is not the case, and so i can conclude that "policy threads" are overdimensioned, but apart from this policy server is working as expected.
    Best Regards