sailfin
  1. sailfin
  2. SAILFIN-2025

Too many timeout errors on the SIPp client. These extend for close to a minute

    Details

    • Type: Bug Bug
    • Status: Open
    • Priority: Minor Minor
    • Resolution: Unresolved
    • Affects Version/s: 2.0
    • Fix Version/s: milestone 1
    • Component/s: session_replication
    • Labels:
      None
    • Environment:

      Operating System: All
      Platform: All

      Description

      Build 31e
      Parent Issue - 1983

      There were ~400 timeouts observed in the conference app 24x1 with soft-failure.
      These are all for the 200 ok for the BYE request. A timeout could possibly occur
      when the BYE was lost in the pipeline at the instant of the failure. At 37 cps
      there should not be that many timeouts. Also the timeouts seems to keep on
      appearing about a minute after failure was introduced.

        Issue Links

          Activity

          Hide
          varunrupela added a comment -

          SIPp logs are attached to issue 2016.

          Show
          varunrupela added a comment - SIPp logs are attached to issue 2016.
          Hide
          varunrupela added a comment -

          Added keyword, updated dependencies

          Show
          varunrupela added a comment - Added keyword, updated dependencies
          Hide
          Mahesh Kannan added a comment -

          The Bye messages cause removal of the SIP artifacts. As we know, SSR batches
          remove calls and these are transmitted every 15 seconds. It has been observed on
          Vivek's 4 core 10 instance setup that even at 37cps, the number of ids that are
          batched is ~ 500 and in some cases even 600+. This means at the time of instance
          kill there could be as many as 600+ outstanding remove calls. This explains why
          we see too many timeouts.

          Since these are for BYE messages there is no session loss and hence I think this
          can be P4d.

          I agree that we need to investigate why the timeouts keep on coming. We have
          seen in 2016 that sessionexpired messages appear right after a kill within a
          minute. It is possible that these two could be related (or not).

          Again since the timeouts are related to BYE messages, this is not a show stopper.

          Show
          Mahesh Kannan added a comment - The Bye messages cause removal of the SIP artifacts. As we know, SSR batches remove calls and these are transmitted every 15 seconds. It has been observed on Vivek's 4 core 10 instance setup that even at 37cps, the number of ids that are batched is ~ 500 and in some cases even 600+. This means at the time of instance kill there could be as many as 600+ outstanding remove calls. This explains why we see too many timeouts. Since these are for BYE messages there is no session loss and hence I think this can be P4d. I agree that we need to investigate why the timeouts keep on coming. We have seen in 2016 that sessionexpired messages appear right after a kill within a minute. It is possible that these two could be related (or not). Again since the timeouts are related to BYE messages, this is not a show stopper.
          Hide
          Mahesh Kannan added a comment -

          Since the timeouts are related to BYE messages and do not result in session
          losses, this is not a show stopper. Marking this as a P4 for this release but
          needs investigation for the next release.

          Show
          Mahesh Kannan added a comment - Since the timeouts are related to BYE messages and do not result in session losses, this is not a show stopper. Marking this as a P4 for this release but needs investigation for the next release.

            People

            • Assignee:
              Mahesh Kannan
              Reporter:
              varunrupela
            • Votes:
              0 Vote for this issue
              Watchers:
              0 Start watching this issue

              Dates

              • Created:
                Updated: