2 Replies Latest reply on Aug 17, 2017 11:17 AM by RichardLetts

    SQL PATCHES BEING APPLIED - BEST PRACTICE APPROACH

    riffster

      Our physical Solarwinds database server is having SQL (and other non-Solarwinds patches) applied this weekend.

       

      We have stopped our two VM Solarwinds application servers (one server has NPM, NCM and NTA on it and the other only manages the Solarwinds NetFlow database using NetFlow Configuration Manager) before but never stopped the SQL server. There will be at least two reboots of the Solarwinds database server during these patches,

       

      What I was going to do was stop the main VM (with NPM, NCM and NTA installed on it) using the "Shutdown Everything" on the Solarwinds Service Manager app just before the maintenance and upon receiving the "All Clear" after the end of patching/return to online status by the SQL server, click on the "Start Everything" button and test access to the Web Console and go through the various sections. Good so far? If not - what would you do differently/add/remove/modify?

       

      But is that enough? (The Netflow database is kept separately on the second VM server.) If it isn't enough, what more should be done?

       

      Am I missing some caveats, gotchas and other miscreant occurences? If so, which?

       

      I was thinking that it might be prudent to schedule a backup of the database server tomorrow with services stopped as described above - any feelings? The maintenance backups done by the SQL team before these patches are apparently normally done with Solarwinds services running and I am wondering if those backups be guaranteed good?

        • Re: SQL PATCHES BEING APPLIED - BEST PRACTICE APPROACH
          riffster

          Forgot to note versions of software running.

           

          NPM Version 12.0

          NCM Version 7.4.1

          NTA Version 4.1.2

           

          OS Version on main VM Server (NPM, NCM, NTA application server) - Windows Server 2008 R2 Enterprise, Service Pack 1

           

          OS Version on secondary VM Server (Netflow Database Configurator/Database) - Windows Server 2008 R2 Enterprise, Service Pack 1

           

          OS Version on SQL Server - Windows Server 2008 R2 Enterprise, Service Pack 1

           

          SQL Version on SQL Server - SQL Server 2008 R2 (SP2) - 10.50.4000.0 (X64)

          • Re: SQL PATCHES BEING APPLIED - BEST PRACTICE APPROACH
            RichardLetts

            [late reply I know]

             

            1. We run backups of our SQL database three times a day -- one reason I find the 'Have you saved a backup of your data' question so odd during an upgrade.

            [actually we run a full backup once/day and a save the transaction logs, but same deal]

             

            2. yes, SQLserver backups are [supposed] to be restorable. you should not have a problem with consistency between tables when restoring it.

             

            3. I do not normally worry about stopping and starting solarwinds services over backup -- they don't know/notice that the transactions are only hitting the log, not the actual database during the backup window. nor should they.

             

            You can also drop an "app_offline.htm" file into your solarwinds inetpub directory, and the server will serve that up instead of an error page.

            We have a pair of database servers mirroring data between them, so during maintenance on one server we switch over to the other server.

            Mirrored databases running in full recovery used to give solarwinds support conniptions and they point at our DB every time...except our DB servers are an identical pair of high-performance servers with over 24 spindles in the arrays rather than the common situation of the backupserver being woefully underpowered. [these servers are 6 years old now, due for replacement in the next FY -- PCIe SSD are now a cost-optimal choice for the database]

             

            Personally, as I do not see that anything bad would happen: I'd drop the app_offline.htm file in there

            then the maintenance happens without

            remove the app_offline file

            see what broke

            [open a support case if anything did break, because services should all recover if the database connectivity gets lost/broken anyway.]

            restart all of the Solarwinds Services to make sure

            1 of 1 people found this helpful