4 Replies Latest reply on Jun 30, 2016 10:36 AM by jlm

    FIleMaker Server 14 error detection

    jlm

      I have a FileMaker Server 14 crash that has as consequence users with spinning wheels and unable to work and users building up on the admin console.

       

      I can't disconnect the users or solve this problem through the admin console.

       

      I created a batch file where I kill the Java and FileMaker Server processes and then restart the FileMaker service.

       

      What I need to know is if there is a way to detect the crash (no event created by FM until the number of users is exceeded) and thus run the batch file.

       

      Maybe there is a plug in that analyses the strings (duplicate users) in the console?

       

      Maybe a filemaker robot with system level scripts?

       

      Until I find what is causing the crashes I must create a mechanism so that my client doesn't have to call me.

       

      JLM

        • 1. Re: FIleMaker Server 14 error detection
          wimdecorte

          I don't think you'd want to automate it in quite that way because it means that you will be re-using the crashed FM files and you don't know whether slightly damage files is actually contributing to the crash.

           

          First thing to find out is whether FMS actually crashed or if it is just running out of resources.

          - check for DMP files in the FMS logs folder

          - check the FMS stats log (in the FMS logs folder) for signs that it is under strain.  Especially pay attention to the 'elapsed time per call'.  The stats log is not on by default so you will have to turn it on if it is not already collecting data

          - check the OS logs for any signs of FMS throwing warnings or errors

          • 2. Re: FIleMaker Server 14 error detection
            jlm

            I already sent log and dump files to FM tech support and nothing was found.

             

            FM testing has a copy of the file for more than a month, now. No crashes up to now.

             

            This approach is because I cannot be looking at the admin console for 24 hours to react to a crash nor can I have my client receiving calls from his clients saying they cannot work.

             

            I built a file from scratch and I'm replacing the actual (importing all data) next weekend. Maybe I'll have a good surprise...

             

            JLM

            • 3. Re: FIleMaker Server 14 error detection
              wimdecorte

              jlm wrote:

               

              I already sent log and dump files to FM tech support and nothing was found.

               

               

               

              So there are DMP files in the FMS logs?  Then that is a clear indication that something is definitely going wrong.  FMI's testing may not generate the same load on the file as your users do, or they may not be running on the same hardware.

               

              Have you looked at the stats.log file?

               

              Can you give us the specs of the server and the typical # of users on the system when the crash happened.

               

              I still maintain that you do not want to automate that kind of restart.  You are doing yourself and your customer a disservice in the long run.  Yes it is painful in the short term but the cause needs to be found, not patched over.

               

              Let us know what is in the stats.log file.

               

              I would probably find a spare server to run this on for a while, just to rule out the hardware.

              • 4. Re: FIleMaker Server 14 error detection
                jlm

                Thanks Wim,

                 

                Shall I send you the .dmp and log files?

                 

                I'm running FMS 14.0.4.412 on a webserver Intel Xeon E5-2630 v3 @ 2.49GHz / 32GB RAM / 2x 512GB SSD (dbs installed) + 3TB 7200rpm (installer + backups, etc...)

                 

                I have a 25 concurrent license + 10 fm pro licenses.

                 

                Usually I get 1 to 2 crashes a day. Average 7 fmpro and 8 webd users.

                 

                Yesterday I had 13 crashes. 10 fmpro and average of 15 webd users.

                 

                The solution has 2 files:

                the main where webd users edit text fields and upload files (.jpg) and fmpro users check the data and generate pdf files that are inserted in the db (all external containers)

                and

                log where fmpro users activity is registered.

                 

                I tested  for 4 months the solution and had no crashes. In the beginning of this year the client started using it and after a few days 701 error WPE process crash happened. (I have task manager initiating the process again)

                 

                I may have one or more WPE crashes in one day and no fmserver crash until next day. For 2 months I rebooted the server at 5:00AM believing I cleaned the errors and started fresh. No luck.