16 Replies Latest reply on Jul 10, 2008 5:21 AM by Krislee

    NPM 8.2 database  - nightly maintenance question

      Morning,

      Our NPM database can take an age to run it's 'nightly' maintenance.  It's usually about 28 hours but can sometimes be two and a half days!

      Anyone got any ideas why? It seems ludicrous.  How can this be brought down to nightly?

      Database size can vary between 800MB and 1.5GB.  PC spec is Pentium D 2.8Ghz (P4 Dual Core), 2GB RAM and 280GB SATA disk.
       

      Regards,

      Kris. 

        • Re: NPM 8.2 database  - nightly maintenance question
          mark wiggans

          This behavior might be caused by a full or corrupt DB. Have you attempted a compact and repair using Access? Generally the closer to the 2GB limit you are the longer the nightly maintenance will take to run through all the records. You might even have a corrupt DB which is not allowing the DB Maintenance to perform correctly. Here are the detailed steps for using the compact and repair within Access. http://support.solarwinds.com/kbase/ProblemDetail.cfm?ID=128

          You might consider adjusting your data retention settings and limit the amount of nodes you monitor to stem some of the DB growth. Also, many improvements have been made since 8.2, so I would also suggest upgrading if possible.


           

            • Re: NPM 8.2 database  - nightly maintenance question

              Thanks Mark.

              'Repair and compaction' is run regularly from within Solarwinds but I tried the first two methods using Access which are shown in the link you've given (the third was to delete the database and start again which I don't want to do).  It has made no difference I'm afraid.  The file size at the start was about 1.2GB (which had just had the Solarwinds Compaction tool run on it), after the Access tools it was still about 1.2GB.

              The nightly maintenance was run again and it took 66 hours to complete.  After compacting it again it dropper down to 850MB.


              Any other ideas?

              Cheers,

               

              Kris.
               


                • Re: NPM 8.2 database  - nightly maintenance question
                  cdoyle

                  Hi Kris,


                  Could you submit a support ticket so that we can look into this further with you?
                  You can submit it by logging into your customer portal and clicking "Submit a Ticket", or from here: http://www.solarwinds.com/support/


                   Caroline

                    • Re: NPM 8.2 database  - nightly maintenance question

                       Hi Caroline,

                      Ticket has been raised : Case #42287

                       
                      Regards,

                      Kris. 

                        • Re: NPM 8.2 database  - nightly maintenance question

                          Kris,


                               Has your issue been resolved? I've had to open a ticket myself (#48162)  for almost the same issue. Each and every night, the nightly db maintenance tries to run, but gets hung up sometime during the night. Windows 2003 Server Task manager marks it as "Not Responding". Strange thing is, if I close NetPerfMon and run it straight from the dbase utility, it doesn't hang up. I've applied SP1 hoping that would help, but it didnt have any affect.


                          I opened my ticket last week and have yet to hear anything from support. Are they usually this "responsive"? 

                            • Re: NPM 8.2 database  - nightly maintenance question

                              Forgot to mention that I am running 9.2 Engineers Toolset.

                                • Re: NPM 8.2 database  - nightly maintenance question
                                  mark wiggans

                                   The reply was sent an hour after the case was received, please review mail filters or spam folder perhaps. One suggestion for you to try would be to create a new database and test the Maintenance on it - if the maintenance runs ok, then the issue is with that database itself.

                                  Open the C:\Program Files\Solarwinds\Engineer's Toolset\ fodler, and look for the SWNetPerfMon.cfg file.
                                  Move this file to the desktop (deleting it from the folder.
                                  Run the NPM tool again - when it does not find the SWNetPerfMon.cfg file in that folder, it will create a new one. Add a few devices in, and then run the Nightly Maintenance again from File-> Database Maintenance. This should take just a few minutes to complete.

                                  If it does, then we need to start looking at your old SWNetPerfMon.cfg file. Move it back into the Engineer's Toolset folder.
                                  Check the size of your SWNetPerfMon.cfg file - if it is around the 1.5GB mark, it can impact the Nightly maintenance (as the Nightly maintenance needs 'room' in order to complete quickly. If the database has grown to around this size, try going to File -> NPM Settings, and examining the Statistics Summarization settings under the Database tab.

                                  Try reducing the length of time detailed, hourly and daily statistics are stored for. Once this is done, start the database maintenance off again. It may take some time the first time - but check the size of the file after completion to see if it has made a difference to the size.
                                  If the SWNetPerfMon.cfg file is small (under 1GB), then it might be corrupted beyond what the compact and repair could fix. If this is the case, you may need to start fresh with a new database.

                                  Let me know if this helps.

                                • Re: NPM 8.2 database  - nightly maintenance question

                                  Morning,

                                  I did not get the issue resolved to satisfaction but the conclusion was that we are running far too many devices/elements and keeping the data for too long.  We've since halved the retention time from 365 to 180 days.  This halved our Database size down to 500MB after compaction but it grows to 1.5GB in about a week.

                                  The nightly stats maintenance runs much better when you have closed NPM down but it takes ages for the program to quit properly - between 5 hours and 14 hours (we've not worked it out properly yet).  Again, this is due to the number of devices we monitor.

                                  Basically we need to upgrade to Orion...

                                  Responses from the helpdesk were pretty good.  I'd often get an email from them asking how the last step/idea had gone before I'd actually tried it.

                                  Regards,

                                  Kris.