Loading ...
Sorry, an error occurred while loading the content.
 

SQL 2005 sp3 Question

Expand Messages
  • Good, Mike
    If I remember correctly, last time I asked about this, the consensus advice was to immediately install Cumulative Update #1 after applying sp3. Is this still
    Message 1 of 15 , Jan 7, 2010

      If I remember correctly, last time I asked about this, the consensus advice was to immediately install Cumulative Update #1 after applying sp3.  Is this still true? 

      http://www.hsn.com

    • Jorge Segarra
      This is one of those fun it depends questions. Remember that technically speaking the cumulative updates aren t official so in theory you should only be
      Message 2 of 15 , Jan 7, 2010
        This is one of those fun "it depends" questions. Remember that technically speaking the cumulative updates aren't "official" so in theory you should only be installing those CU's if they specifically fix a problem listed. Usually it doesn't hurt to install but if you're going to just test it on a non-prod box/database first and make sure it doesn't do anything you're not expecting it to.





        i'm EMAILING FOR THE GREATER GOOD
        Join me



        To: tampasql@yahoogroups.com
        From: mike.good@...
        Date: Thu, 7 Jan 2010 18:05:00 -0500
        Subject: [tampasql] SQL 2005 sp3 Question

         

        If I remember correctly, last time I asked about this, the consensus advice was to immediately install Cumulative Update #1 after applying sp3.  Is this still true? 

        http://www.hsn. com
        0


      • Jonathan Kehayias
        Mike, I think you might be confusing SP3 with SP2 which had a major Maintenance Plan bug in it that was fixed in the CU1 and then again in CU2 for SP2. The
        Message 3 of 15 , Jan 7, 2010
          Mike,
           
          I think you might be confusing SP3 with SP2 which had a major Maintenance Plan bug in it that was fixed in the CU1 and then again in CU2 for SP2.  The only reason you would need to apply CU1 to SP3 is if you had CU10 or CU11 for SP2 previously installed on SQL 2005 since those were released while SP3 was in the regression testing process and were not included in the SP3 code base. 
           
          CU's are still official releases and are fully supported by Microsoft as a part of the Incremental Servicing Model for SQL Server (http://support.microsoft.com/kb/935897).  The hot fixes are tested but would receive additional tests when rolled up into a service pack.  I am from the "If it ain't broke, don't fix it" school of thought most times, but if there is a fix listed in the latest CU release that seems important enough to me, I will apply it after testing it against development servers, for example CU4 to SP3 has a fix related to Read Committed Snapshot returning uncommitted data http://support.microsoft.com/kb/970279/en-us, and was applied to two servers in my environment that use Read Committed Snapshot.
           
          On Thu, Jan 7, 2010 at 6:40 PM, Jorge Segarra <jorge.segarra@...> wrote:
           

          This is one of those fun "it depends" questions. Remember that technically speaking the cumulative updates aren't "official" so in theory you should only be installing those CU's if they specifically fix a problem listed. Usually it doesn't hurt to install but if you're going to just test it on a non-prod box/database first and make sure it doesn't do anything you're not expecting it to.





          i'm EMAILING FOR THE GREATER GOOD
          Join me



          To: tampasql@yahoogroups.com
          From: mike.good@...
          Date: Thu, 7 Jan 2010 18:05:00 -0500
          Subject: [tampasql] SQL 2005 sp3 Question


           

          If I remember correctly, last time I asked about this, the consensus advice was to immediately install Cumulative Update #1 after applying sp3.  Is this still true? 

          http://www.hsn.com
          0



        • Good, Mike
          Thanks, guys. I probably did ask this same question for sp2 as well, but that would have been a year before; I m pretty sure I brought this up re sp3 at a
          Message 4 of 15 , Jan 7, 2010

            Thanks, guys.  I probably did ask this same question for sp2 as well, but that would have been a year before; I'm pretty sure I brought this up re sp3 at a meeting w/in the last six months.  Just can't find my notes.

             

            Aaron Bertrand's Blog seems to imply CU#1  important.  But maybe the advice I heard was to install KB 970892 this instead?  That would bring the version up to 4053. 

             

            If you have any 2005 boxes, what version are they on?

             

            Reason I'm asking is that it maybe many months before I get to do this again, need to do anything important this go-around.

             

            From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
            Sent: Thursday, January 07, 2010 7:50 PM
            To: tampasql@yahoogroups.com
            Subject: Re: [tampasql] SQL 2005 sp3 Question

             

             

            Mike,

             

            I think you might be confusing SP3 with SP2 which had a major Maintenance Plan bug in it that was fixed in the CU1 and then again in CU2 for SP2.  The only reason you would need to apply CU1 to SP3 is if you had CU10 or CU11 for SP2 previously installed on SQL 2005 since those were released while SP3 was in the regression testing process and were not included in the SP3 code base. 

             

            CU's are still official releases and are fully supported by Microsoft as a part of the Incremental Servicing Model for SQL Server (http://support.microsoft.com/kb/935897).  The hot fixes are tested but would receive additional tests when rolled up into a service pack.  I am from the "If it ain't broke, don't fix it" school of thought most times, but if there is a fix listed in the latest CU release that seems important enough to me, I will apply it after testing it against development servers, for example CU4 to SP3 has a fix related to Read Committed Snapshot returning uncommitted data http://support.microsoft.com/kb/970279/en-us, and was applied to two servers in my environment that use Read Committed Snapshot.

             

            On Thu, Jan 7, 2010 at 6:40 PM, Jorge Segarra <jorge.segarra@...> wrote:

             

            This is one of those fun "it depends" questions. Remember that technically speaking the cumulative updates aren't "official" so in theory you should only be installing those CU's if they specifically fix a problem listed. Usually it doesn't hurt to install but if you're going to just test it on a non-prod box/database first and make sure it doesn't do anything you're not expecting it to.




            Error! Filename not specified.EMAILING FOR THE GREATER GOOD
            Join me

             


            To: tampasql@yahoogroups.com
            From: mike.good@...
            Date: Thu, 7 Jan 2010 18:05:00 -0500
            Subject: [tampasql] SQL 2005 sp3 Question



             

             

            If I remember correctly, last time I asked about this, the consensus advice was to immediately install Cumulative Update #1 after applying sp3.  Is this still true? 

            http://www.hsn.com
            0

             

             

          • Jonathan Kehayias
            Aaron s blog post makes the point I made about CU10 or 11 for SP2. Build 3294 was the CU10 for SP2 build. If you were under this number, then you could
            Message 5 of 15 , Jan 7, 2010
              Aaron's blog post makes the point I made about CU10 or 11 for SP2.  Build 3294 was the CU10 for SP2 build.  If you were under this number, then you could immediately apply SP3 without risking losing a functionality or fix in CU10 or 11.  If you were above this, to keep the CU10 and 11 fixes, you need C1 for SP3. 
               
              (I am sure Microsoft could have made this more confusing if they actually tried to, but it is what it is.)
               
              I had to log in to work to check this, but most of my servers are on build 4226 (CU4), a few are 4207 (CU1) still, and our reporting services servers are 4266 which is CU6 for SP3, which has the MS09-062 GFE security update included in it.
               
               
              If you only get to do this once, I'd probably just do the latest CU or the one immediately prior to it.  They have been out long enough that stability issues would be known for them and I haven't heard of any on the Forums.

              Jonathan Kehayias, MCITP
              jmkehayias@...
              SQL Server MVP
              http://sqlblog.com/blogs/jonathan_kehayias/
              http://twitter.com/sqlsarg




              On Thu, Jan 7, 2010 at 8:19 PM, Good, Mike <mike.good@...> wrote:
               

              Thanks, guys.  I probably did ask this same question for sp2 as well, but that would have been a year before; I'm pretty sure I brought this up re sp3 at a meeting w/in the last six months.  Just can't find my notes.

               

              Aaron Bertrand's Blog seems to imply CU#1  important.  But maybe the advice I heard was to install KB 970892 this instead?  That would bring the version up to 4053. 

               

              If you have any 2005 boxes, what version are they on?

               

              Reason I'm asking is that it maybe many months before I get to do this again, need to do anything important this go-around.

               

              From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
              Sent: Thursday, January 07, 2010 7:50 PM
              To: tampasql@yahoogroups.com
              Subject: Re: [tampasql] SQL 2005 sp3 Question

               

               

              Mike,

               

              I think you might be confusing SP3 with SP2 which had a major Maintenance Plan bug in it that was fixed in the CU1 and then again in CU2 for SP2.  The only reason you would need to apply CU1 to SP3 is if you had CU10 or CU11 for SP2 previously installed on SQL 2005 since those were released while SP3 was in the regression testing process and were not included in the SP3 code base. 

               

              CU's are still official releases and are fully supported by Microsoft as a part of the Incremental Servicing Model for SQL Server (http://support.microsoft.com/kb/935897).  The hot fixes are tested but would receive additional tests when rolled up into a service pack.  I am from the "If it ain't broke, don't fix it" school of thought most times, but if there is a fix listed in the latest CU release that seems important enough to me, I will apply it after testing it against development servers, for example CU4 to SP3 has a fix related to Read Committed Snapshot returning uncommitted data http://support.microsoft.com/kb/970279/en-us, and was applied to two servers in my environment that use Read Committed Snapshot.

               

              On Thu, Jan 7, 2010 at 6:40 PM, Jorge Segarra <jorge.segarra@...> wrote:

               

              This is one of those fun "it depends" questions. Remember that technically speaking the cumulative updates aren't "official" so in theory you should only be installing those CU's if they specifically fix a problem listed. Usually it doesn't hurt to install but if you're going to just test it on a non-prod box/database first and make sure it doesn't do anything you're not expecting it to.




              Error! Filename not specified.EMAILING FOR THE GREATER GOOD
              Join me

               


              To: tampasql@yahoogroups.com
              From: mike.good@...
              Date: Thu, 7 Jan 2010 18:05:00 -0500
              Subject: [tampasql] SQL 2005 sp3 Question



               

               

              If I remember correctly, last time I asked about this, the consensus advice was to immediately install Cumulative Update #1 after applying sp3.  Is this still true? 

              http://www.hsn.com
              0

               

               


            • Good, Mike
              Very good. Thanks for checking your boxes. I just got a pm from Gary, who espouses same philosophy. Unless someone makes good counter argument, this is path
              Message 6 of 15 , Jan 7, 2010

                Very good.  Thanks for checking your boxes.  I just got a pm from Gary, who espouses same philosophy. 

                 

                Unless someone makes good counter argument, this is path I'm going to try to pursue.  I will have time to run this in non-prod environments prior to next prod install.

                 

                From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
                Sent: Thursday, January 07, 2010 9:33 PM
                To: tampasql@yahoogroups.com
                Subject: Re: [tampasql] SQL 2005 sp3 Question

                 

                 

                Aaron's blog post makes the point I made about CU10 or 11 for SP2.  Build 3294 was the CU10 for SP2 build.  If you were under this number, then you could immediately apply SP3 without risking losing a functionality or fix in CU10 or 11.  If you were above this, to keep the CU10 and 11 fixes, you need C1 for SP3. 

                 

                (I am sure Microsoft could have made this more confusing if they actually tried to, but it is what it is.)

                 

                I had to log in to work to check this, but most of my servers are on build 4226 (CU4), a few are 4207 (CU1) still, and our reporting services servers are 4266 which is CU6 for SP3, which has the MS09-062 GFE security update included in it.

                 

                 

                If you only get to do this once, I'd probably just do the latest CU or the one immediately prior to it.  They have been out long enough that stability issues would be known for them and I haven't heard of any on the Forums.

                Jonathan Kehayias, MCITP
                jmkehayias@...
                SQL Server MVP
                http://sqlblog.com/blogs/jonathan_kehayias/
                http://twitter.com/sqlsarg



                On Thu, Jan 7, 2010 at 8:19 PM, Good, Mike <mike.good@...> wrote:

                 

                Thanks, guys.  I probably did ask this same question for sp2 as well, but that would have been a year before; I'm pretty sure I brought this up re sp3 at a meeting w/in the last six months.  Just can't find my notes.

                 

                Aaron Bertrand's Blog seems to imply CU#1  important.  But maybe the advice I heard was to install KB 970892 this instead?  That would bring the version up to 4053. 

                 

                If you have any 2005 boxes, what version are they on?

                 

                Reason I'm asking is that it maybe many months before I get to do this again, need to do anything important this go-around.

                 

                From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
                Sent: Thursday, January 07, 2010 7:50 PM
                To: tampasql@yahoogroups.com
                Subject: Re: [tampasql] SQL 2005 sp3 Question

                 

                 

                Mike,

                 

                I think you might be confusing SP3 with SP2 which had a major Maintenance Plan bug in it that was fixed in the CU1 and then again in CU2 for SP2.  The only reason you would need to apply CU1 to SP3 is if you had CU10 or CU11 for SP2 previously installed on SQL 2005 since those were released while SP3 was in the regression testing process and were not included in the SP3 code base. 

                 

                CU's are still official releases and are fully supported by Microsoft as a part of the Incremental Servicing Model for SQL Server (http://support.microsoft.com/kb/935897).  The hot fixes are tested but would receive additional tests when rolled up into a service pack.  I am from the "If it ain't broke, don't fix it" school of thought most times, but if there is a fix listed in the latest CU release that seems important enough to me, I will apply it after testing it against development servers, for example CU4 to SP3 has a fix related to Read Committed Snapshot returning uncommitted data http://support.microsoft.com/kb/970279/en-us, and was applied to two servers in my environment that use Read Committed Snapshot.

                 

                On Thu, Jan 7, 2010 at 6:40 PM, Jorge Segarra <jorge.segarra@...> wrote:

                 

                This is one of those fun "it depends" questions. Remember that technically speaking the cumulative updates aren't "official" so in theory you should only be installing those CU's if they specifically fix a problem listed. Usually it doesn't hurt to install but if you're going to just test it on a non-prod box/database first and make sure it doesn't do anything you're not expecting it to.



                Error! Filename not specified.EMAILING FOR THE GREATER GOOD
                Join me

                 


                To: tampasql@yahoogroups.com
                From: mike.good@...
                Date: Thu, 7 Jan 2010 18:05:00 -0500
                Subject: [tampasql] SQL 2005 sp3 Question



                 

                 

                If I remember correctly, last time I asked about this, the consensus advice was to immediately install Cumulative Update #1 after applying sp3.  Is this still true? 

                http://www.hsn.com
                0

                 

                 

                 

              • Good, Mike
                Thanks for replies to my previous questions. We ve begun implementing sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe
                Message 7 of 15 , Jan 25, 2010

                  Thanks for replies to my previous questions.  We've begun implementing sp3, cu6 on our SQL 2005 servers.  Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).

                   

                  When installing sp3 on one of our clusters, it only installed on one of the two nodes.  At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running."  All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.

                   

                  If we'd known of this pending reboot requirement in advance, we could have failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly.   Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same.  But sp3 didn't tell us about this until after the fact.  Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again. 

                   

                  We now know how to look for pending file rename operations in  "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this  check is part of our sp install process going forward. 

                   

                  The question is:  Is that it?  Or do we need to check other things too?  Does anyone know what is checked when determining a box must be rebooted before sp3 can be installed?

                  http://www.hsn.com

                • MikeG
                  I realize it s not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite
                  Message 8 of 15 , Jan 26, 2010
                    I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                    Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                    --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:
                    >
                    > Thanks for replies to my previous questions. We've begun implementing sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                    >
                    > When installing sp3 on one of our clusters, it only installed on one of the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                    >
                    > If we'd known of this pending reboot requirement in advance, we could have failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn't tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                    >
                    > We now know how to look for pending file rename operations in "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                    >
                    > The question is: Is that it? Or do we need to check other things too? Does anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                    > http://www.hsn.com/
                    >
                  • Jorge Segarra
                    Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be
                    Message 9 of 15 , Jan 26, 2010

                      Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be your biggest hurdle going into it. One way to ensure that before you start anything like an upgrade on a cluster, reboot the non-active node failover and reboot the other one that way both are restarted fresh before upgrades. Also it lets you make sure your clustering and failover are working properly before you’re met with other surprises. I’d also say make sure both nodes are at the same major service pack level but I’m assuming in your case they are already.

                       

                      From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of MikeG
                      Sent: Tuesday, January 26, 2010 12:50 PM
                      To: tampasql@yahoogroups.com
                      Subject: [tampasql] Re: SQL 2005 sp3 Question

                       

                       



                      I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                      Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                      --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:
                      >
                      > Thanks for replies to my previous questions. We've begun implementing sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                      >
                      > When installing sp3 on one of our clusters, it only installed on one of the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                      >
                      > If we'd known of this pending reboot requirement in advance, we could have failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn't tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                      >
                      > We now know how to look for pending file rename operations in "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                      >
                      > The question is: Is that it? Or do we need to check other things too? Does anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                      > http://www.hsn.com/
                      >

                    • Good, Mike
                      Thanks, Jorge. Yes, this is pretty much our plan. So far we ve not been able to get permission for a failover outage immediately in advance of the sp3
                      Message 10 of 15 , Jan 26, 2010

                        Thanks, Jorge.  Yes, this is pretty much our plan.  So far we've not been able to get permission for a failover outage immediately in advance of the sp3 install, so we can really only be sure one (inactive) node has been freshly bounced--not both.    

                         

                        But cluster failovers here are pretty quick, so this is a point worth pursuing further.  Being able to guarantee success has value, management may find it worth the extra small outage time.  Good answer.

                         

                        From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jorge Segarra
                        Sent: Tuesday, January 26, 2010 1:19 PM
                        To: tampasql@yahoogroups.com
                        Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                         

                         

                        Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be your biggest hurdle going into it. One way to ensure that before you start anything like an upgrade on a cluster, reboot the non-active node failover and reboot the other one that way both are restarted fresh before upgrades. Also it lets you make sure your clustering and failover are working properly before you’re met with other surprises. I’d also say make sure both nodes are at the same major service pack level but I’m assuming in your case they are already.

                         

                        From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of MikeG
                        Sent: Tuesday, January 26, 2010 12:50 PM
                        To: tampasql@yahoogroups.com
                        Subject: [tampasql] Re: SQL 2005 sp3 Question

                         

                         



                        I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                        Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                        --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:

                        >
                        > Thanks for replies to my prev ious questions. We've begun implementing
                        sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                        >
                        > When installing sp3 on one of our clusters, it only installed on one of
                        the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                        >
                        > If we'd known of this pending reboot requirement in advance, we could have
                        failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn' t tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                        >
                        > We now know how to look for pending file rename operations in
                        "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                        >
                        > The question is: Is that it? Or do we need to check other things too? Does
                        anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                        > http://www.hsn.com/
                        >

                      • Jorge Segarra
                        You re welcome! One thing (and this is kind of a hack so do so with caution). If you have one of those rename operations pending and you look at the registry
                        Message 11 of 15 , Jan 26, 2010

                          You’re welcome! One thing (and this is kind of a hack so do so with caution). If you have one of those rename operations pending and you look at the registry key and it seems to be only something in a temp directory, what I’ve done in the past is deleted those references from the registry key. This allows you to rerun the SQL setup check and it will pass and you won’t have to reboot the server prior to your install. Again, hacky and dicey, but an option.

                           

                          From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Good, Mike
                          Sent: Tuesday, January 26, 2010 1:51 PM
                          To: tampasql@yahoogroups.com
                          Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                           

                           

                          Thanks, Jorge.  Yes, this is pretty much our plan.  So far we've not been able to get permission for a failover outage immediately in advance of the sp3 install, so we can really only be sure one (inactive) node has been freshly bounced--not both.    

                           

                          But cluster failovers here are pretty quick, so this is a point worth pursuing further.  Being able to guarantee success has value, management may find it worth the extra small outage time.  Good answer.

                           

                          From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jorge Segarra
                          Sent: Tuesday, January 26, 2010 1:19 PM
                          To: tampasql@yahoogroups.com
                          Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                           

                           

                          Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be your biggest hurdle going into it. One way to ensure that before you start anything like an upgrade on a cluster, reboot the non-active node failover and reboot the other one that way both are restarted fresh before upgrades. Also it lets you make sure your clustering and failover are working properly before you’re met with other surprises. I’d also say make sure both nodes are at the same major service pack level but I’m assuming in your case they are already.

                           

                          From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of MikeG
                          Sent: Tuesday, January 26, 2010 12:50 PM
                          To: tampasql@yahoogroups.com
                          Subject: [tampasql] Re: SQL 2005 sp3 Question

                           

                           



                          I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                          Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                          --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:
                          >
                          > Thanks for replies to my prev ious questions. We've begun implementing sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                          >
                          > When installing sp3 on one of our clusters, it only installed on one of the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                          >
                          > If we'd known of this pending reboot requirement in advance, we could have failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn' t tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                          >
                          > We now know how to look for pending file rename operations in "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                          >
                          > The question is: Is that it? Or do we need to check other things too? Does anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                          > http://www.hsn.com/
                          >

                        • Jonathan Kehayias
                          If you copy the key value into a text file, you can always paste the values to be deleted back into the key once SQL Setup completes. I ve done this a couple
                          Message 12 of 15 , Jan 26, 2010
                            If you copy the key value into a text file, you can always paste the values to be deleted back into the key once SQL Setup completes.  I've done this a couple of times as a hack around the issue.  The key doesn't have hard bound rules associated with its values.
                            On Tue, Jan 26, 2010 at 3:13 PM, Jorge Segarra <jorge.segarra@...> wrote:
                             

                            You’re welcome! One thing (and this is kind of a hack so do so with caution). If you have one of those rename operations pending and you look at the registry key and it seems to be only something in a temp directory, what I’ve done in the past is deleted those references from the registry key. This allows you to rerun the SQL setup check and it will pass and you won’t have to reboot the server prior to your install. Again, hacky and dicey, but an option.

                             

                            From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Good, Mike
                            Sent: Tuesday, January 26, 2010 1:51 PM


                            To: tampasql@yahoogroups.com
                            Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                             

                             

                            Thanks, Jorge.  Yes, this is pretty much our plan.  So far we've not been able to get permission for a failover outage immediately in advance of the sp3 install, so we can really only be sure one (inactive) node has been freshly bounced--not both.    

                             

                            But cluster failovers here are pretty quick, so this is a point worth pursuing further.  Being able to guarantee success has value, management may find it worth the extra small outage time.  Good answer.

                             

                            From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jorge Segarra
                            Sent: Tuesday, January 26, 2010 1:19 PM
                            To: tampasql@yahoogroups.com
                            Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                             

                             

                            Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be your biggest hurdle going into it. One way to ensure that before you start anything like an upgrade on a cluster, reboot the non-active node failover and reboot the other one that way both are restarted fresh before upgrades. Also it lets you make sure your clustering and failover are working properly before you’re met with other surprises. I’d also say make sure both nodes are at the same major service pack level but I’m assuming in your case they are already.

                             

                            From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of MikeG
                            Sent: Tuesday, January 26, 2010 12:50 PM
                            To: tampasql@yahoogroups.com
                            Subject: [tampasql] Re: SQL 2005 sp3 Question

                             

                             



                            I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                            Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                            --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:
                            >
                            > Thanks for replies to my prev ious questions. We've begun implementing sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                            >
                            > When installing sp3 on one of our clusters, it only installed on one of the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                            >
                            > If we'd known of this pending reboot requirement in advance, we could have failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn' t tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                            >
                            > We now know how to look for pending file rename operations in "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                            >
                            > The question is: Is that it? Or do we need to check other things too? Does anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                            > http://www.hsn.com/
                            >


                          • Good, Mike
                            There he is! Ok, I m on board with this hack if required. I ve done this before, just to eliminate failed entries that would not go away. But we re all
                            Message 13 of 15 , Jan 26, 2010

                              There he is!  Ok, I'm on board with this hack if required.  I've done this before, just to eliminate failed entries that would not go away. 

                               

                              But we're all talking about the one thing I know to check.  What I'm more concerned about is anything else.  So far can't find anyone that can state if this is only factor in the service pack installer's determination of "reboot pending before sp can be installed."  The fact that we're all familiar with this one setting makes me think this might be the only factor?

                               

                              On the cautionary front, am currently trying to get approval for a proactive failover, perhaps hours or days in advance of the sp3 install--this would allow us to ensure both nodes rebooted prior to the sp3 install, and make this discussion moot.

                               

                              From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
                              Sent: Tuesday, January 26, 2010 3:39 PM
                              To: tampasql@yahoogroups.com
                              Subject: Re: [tampasql] Re: SQL 2005 sp3 Question

                               

                               

                              If you copy the key value into a text file, you can always paste the values to be deleted back into the key once SQL Setup completes.  I've done this a couple of times as a hack around the issue.  The key doesn't have hard bound rules associated with its values.

                              On Tue, Jan 26, 2010 at 3:13 PM, Jorge Segarra <jorge.segarra@...> wrote:

                               

                              You’re welcome! One thing (and this is kind of a hack so do so with caution). If you have one of those rename operations pending and you look at the registry key and it seems to be only something in a temp directory, what I’ve done in the past is deleted those references from the registry key. This allows you to rerun the SQL setup check and it will pass and you won’t have to reboot the server prior to your install. Again, hacky and dicey, but an option.

                               

                              From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Good, Mike
                              Sent: Tuesday, January 26, 2010 1:51 PM


                              To: tampasql@yahoogroups.com
                              Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                               

                               

                              Thanks, Jorge.  Yes, this is pretty much our plan.  So far we've not been able to get permission for a failover outage immediately in advance of the sp3 install, so we can really only be sure one (inactive) node has been freshly bounced--not both.    

                               

                              But cluster failovers here are pretty quick, so this is a point worth pursuing further.  Being able to guarantee success has value, management may find it worth the extra small outage time.  Good answer.

                               

                              From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jorge Segarra
                              Sent: Tuesday, January 26, 2010 1:19 PM
                              To: tampasql@yahoogroups.com
                              Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                               

                               

                              Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be your biggest hurdle going into it. One way to ensure that before you start anything like an upgrade on a cluster, reboot the non-active node failover and reboot the other one that way both are restarted fresh before upgrades. Also it lets you make sure your clustering and failover are working properly before you’re met with other surprises. I’d also say make sure both nodes are at the same major service pack level but I’m assuming in your case they are already.

                               

                              From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of MikeG
                              Sent: Tuesday, January 26, 2010 12:50 PM
                              To: tampasql@yahoogroups.com
                              Subject: [tampasql] Re: SQL 2005 sp3 Question

                               

                               



                              I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                              Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                              --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:

                              >
                              > Thanks for replies to my prev ious questions. We've begun implementing
                              sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                              >
                              > When installing sp3 on one of our clusters, it only installed on one of
                              the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                              >
                              > If we'd known of this pending reboot requirement in advance, we could have
                              failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn' t tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                              >
                              > We now know how to look for pending file rename operations in
                              "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                              >
                              > The question is: Is that it? Or do we need to check other things too? Does
                              anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                              > http://www.hsn.com/
                              >

                               

                            • Jonathan Kehayias
                              Sorry Mike, I ve had a nasty stomach virus the last 48 hours and haven t been at work or in my personal email since SQL Saturday. Trying to catch up now. To
                              Message 14 of 15 , Jan 26, 2010

                                Sorry Mike, I've had a nasty stomach virus the last 48 hours and haven't been at work or in my personal email since SQL Saturday.  Trying to catch up now.
                                 
                                To my knowledge the pending file rename operations is the only thing that ever triggers failure of the reboot required setup pre-requisite check.  I haven't seen anything else mentioned on the Setup/Upgrade forums in going on two-half years answering on there beyond this registry key.  I can email a member of the product team to find out for certain, but I know that if I post the question incorrectly, I won't likely get a response.  The supported path for this would be to reboot the nodes and clear the pending file operations, and then apply the patches.  Anything else is a hack around the issue.
                                 
                                I've personally used the method I described in my previous email with great success in 2005 and 2008 and I have never had a issue popup after the fact.  Let me know if you want me to try and get more information about that check from the product team and I'll see what I can dig up.
                                 
                                Jonathan Kehayias, MCITP
                                jmkehayias@...
                                SQL Server MVP
                                http://sqlblog.com/blogs/jonathan_kehayias/
                                http://twitter.com/sqlsarg




                                On Tue, Jan 26, 2010 at 3:50 PM, Good, Mike <mike.good@...> wrote:
                                 

                                There he is!  Ok, I'm on board with this hack if required.  I've done this before, just to eliminate failed entries that would not go away. 

                                 

                                But we're all talking about the one thing I know to check.  What I'm more concerned about is anything else.  So far can't find anyone that can state if this is only factor in the service pack installer's determination of "reboot pending before sp can be installed."  The fact that we're all familiar with this one setting makes me think this might be the only factor?

                                 

                                On the cautionary front, am currently trying to get approval for a proactive failover, perhaps hours or days in advance of the sp3 install--this would allow us to ensure both nodes rebooted prior to the sp3 install, and make this discussion moot.

                                 

                                From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
                                Sent: Tuesday, January 26, 2010 3:39 PM
                                To: tampasql@yahoogroups.com
                                Subject: Re: [tampasql] Re: SQL 2005 sp3 Question

                                 

                                 

                                If you copy the key value into a text file, you can always paste the values to be deleted back into the key once SQL Setup completes.  I've done this a couple of times as a hack around the issue.  The key doesn't have hard bound rules associated with its values.

                                On Tue, Jan 26, 2010 at 3:13 PM, Jorge Segarra <jorge.segarra@...> wrote:

                                 

                                You’re welcome! One thing (and this is kind of a hack so do so with caution). If you have one of those rename operations pending and you look at the registry key and it seems to be only something in a temp directory, what I’ve done in the past is deleted those references from the registry key. This allows you to rerun the SQL setup check and it will pass and you won’t have to reboot the server prior to your install. Again, hacky and dicey, but an option.

                                 

                                From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Good, Mike
                                Sent: Tuesday, January 26, 2010 1:51 PM


                                To: tampasql@yahoogroups.com
                                Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                                 

                                 

                                Thanks, Jorge.  Yes, this is pretty much our plan.  So far we've not been able to get permission for a failover outage immediately in advance of the sp3 install, so we can really only be sure one (inactive) node has been freshly bounced--not both.    

                                 

                                But cluster failovers here are pretty quick, so this is a point worth pursuing further.  Being able to guarantee success has value, management may find it worth the extra small outage time.  Good answer.

                                 

                                From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jorge Segarra
                                Sent: Tuesday, January 26, 2010 1:19 PM
                                To: tampasql@yahoogroups.com
                                Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                                 

                                 

                                Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be your biggest hurdle going into it. One way to ensure that before you start anything like an upgrade on a cluster, reboot the non-active node failover and reboot the other one that way both are restarted fresh before upgrades. Also it lets you make sure your clustering and failover are working properly before you’re met with other surprises. I’d also say make sure both nodes are at the same major service pack level but I’m assuming in your case they are already.

                                 

                                From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of MikeG
                                Sent: Tuesday, January 26, 2010 12:50 PM
                                To: tampasql@yahoogroups.com
                                Subject: [tampasql] Re: SQL 2005 sp3 Question

                                 

                                 



                                I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                                Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                                --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:
                                >
                                > Thanks for replies to my prev ious questions. We've begun implementing sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                                >
                                > When installing sp3 on one of our clusters, it only installed on one of the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                                >
                                > If we'd known of this pending reboot requirement in advance, we could have failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn' t tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                                >
                                > We now know how to look for pending file rename operations in "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                                >
                                > The question is: Is that it? Or do we need to check other things too? Does anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                                > http://www.hsn.com/
                                >

                                 


                              • Good, Mike
                                I m sorry to hear about the virus. 1st sentence paragraph 2 was the answer I was hoping for, especially with all your experience on those forums. Please
                                Message 15 of 15 , Jan 26, 2010

                                  I'm sorry to hear about the virus.  1st sentence paragraph 2 was the answer I was hoping for, especially with all your experience on those forums.  Please don't bother posting further.  I'm not really worried about handling the pending operations, I just need to make sure there aren't any before I start.  Or if there are, I'll postpone the sp3 install. 

                                   

                                  THANK YOU VERY MUCH!

                                   

                                  From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
                                  Sent: Tuesday, January 26, 2010 8:51 PM
                                  To: tampasql@yahoogroups.com
                                  Subject: Re: [tampasql] Re: SQL 2005 sp3 Question

                                   

                                   

                                  Sorry Mike, I've had a nasty stomach virus the last 48 hours and haven't been at work or in my personal email since SQL Saturday.  Trying to catch up now.
                                   
                                  To my knowledge the pending file rename operations is the only thing that ever triggers failure of the reboot required setup pre-requisite check.  I haven't seen anything else mentioned on the Setup/Upgrade forums in going on two-half years answering on there beyond this registry key.  I can email a member of the product team to find out for certain, but I know that if I post the question incorrectly, I won't likely get a response.  The supported path for this would be to reboot the nodes and clear the pending file operations, and then apply the patches.  Anything else is a hack around the issue.
                                   
                                  I've personally used the method I described in my previous email with great success in 2005 and 2008 and I have never had a issue popup after the fact.  Let me know if you want me to try and get more information about that check from the product team and I'll see what I can dig up.
                                   
                                  Jonathan Kehayias, MCITP
                                  jmkehayias@...
                                  SQL Server MVP
                                  http://sqlblog.com/blogs/jonathan_kehayias/
                                  http://twitter.com/sqlsarg



                                  On Tue, Jan 26, 2010 at 3:50 PM, Good, Mike <mike.good@...> wrote:

                                   

                                  There he is!  Ok, I'm on board with this hack if required.  I've done this before, just to eliminate failed entries that would not go away. 

                                   

                                  But we're all talking about the one thing I know to check.  What I'm more concerned about is anything else.  So far can't find anyone that can state if this is only factor in the service pack installer's determination of "reboot pending before sp can be installed."  The fact that we're all familiar with this one setting makes me think this might be the only factor?

                                   

                                  On the cautionary front, am currently trying to get approval for a proactive failover, perhaps hours or days in advance of the sp3 install--this would allow us to ensure both nodes rebooted prior to the sp3 install, and make this discussion moot.

                                   

                                  From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jonathan Kehayias
                                  Sent: Tuesday, January 26, 2010 3:39 PM
                                  To: tampasql@yahoogroups.com
                                  Subject: Re: [tampasql] Re: SQL 2005 sp3 Question

                                   

                                   

                                  If you copy the key value into a text file, you can always paste the values to be deleted back into the key once SQL Setup completes.  I've done this a couple of times as a hack around the issue.  The key doesn't have hard bound rules associated with its values.

                                  On Tue, Jan 26, 2010 at 3:13 PM, Jorge Segarra <jorge.segarra@...> wrote:

                                   

                                  You’re welcome! One thing (and this is kind of a hack so do so with caution). If you have one of those rename operations pending and you look at the registry key and it seems to be only something in a temp directory, what I’ve done in the past is deleted those references from the registry key. This allows you to rerun the SQL setup check and it will pass and you won’t have to reboot the server prior to your install. Again, hacky and dicey, but an option.

                                   

                                  From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Good, Mike
                                  Sent: Tuesday, January 26, 2010 1:51 PM


                                  To: tampasql@yahoogroups.com
                                  Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                                   

                                   

                                  Thanks, Jorge.  Yes, this is pretty much our plan.  So far we've not been able to get permission for a failover outage immediately in advance of the sp3 install, so we can really only be sure one (inactive) node has been freshly bounced--not both.    

                                   

                                  But cluster failovers here are pretty quick, so this is a point worth pursuing further.  Being able to guarantee success has value, management may find it worth the extra small outage time.  Good answer.

                                   

                                  From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of Jorge Segarra
                                  Sent: Tuesday, January 26, 2010 1:19 PM
                                  To: tampasql@yahoogroups.com
                                  Subject: RE: [tampasql] Re: SQL 2005 sp3 Question

                                   

                                   

                                  Lol sorry maybe I misunderstood what you were asking. The pending file rename is a giant PITA and should be checked. As far as I know that would probably be your biggest hurdle going into it. One way to ensure that before you start anything like an upgrade on a cluster, reboot the non-active node failover and reboot the other one that way both are restarted fresh before upgrades. Also it lets you make sure your clustering and failover are working properly before you’re met with other surprises. I’d also say make sure both nodes are at the same major service pack level but I’m assuming in your case they are already.

                                   

                                  From: tampasql@yahoogroups.com [mailto:tampasql@yahoogroups.com] On Behalf Of MikeG
                                  Sent: Tuesday, January 26, 2010 12:50 PM
                                  To: tampasql@yahoogroups.com
                                  Subject: [tampasql] Re: SQL 2005 sp3 Question

                                   

                                   



                                  I realize it's not the most interesting question, but I would have bet my new Hawaiian shirt that I would have seen at least one reply. I suspect my favorite local MVP is suddenly a lot busier???

                                  Anyway, I just re-posted this out on SQL Server Central, we'll see if anyone there has anything to contribute.

                                  --- In tampasql@yahoogroups.com, "Good, Mike" <mike.good@...> wrote:

                                  >
                                  > Thanks for replies to my prev ious questions. We've begun implementing
                                  sp3, cu6 on our SQL 2005 servers. Below is a follow-up problem/question (and maybe something you might want to consider before sp install on a critical system).
                                  >
                                  > When installing sp3 on one of our clusters, it only installed on one of
                                  the two nodes. At the end of the install we got a message "Pending File Operation" "A previous program installation created pending file operations on the computer where SQL 2005 setup is running." All that we had to do was reboot and re-run the sp3 install, but this more than doubled our outage duration--this outage would be unacceptable on a different system.
                                  >
                                  > If we'd known of this pending reboot requirement in advance, we could have
                                  failed over, rebooted the inactive node with very little impact, and then applied sp3 much more quickly. Or if the sp3 install had issued a warning early in the process, I could have aborted it and done same. But sp3 didn' t tell us about this until after the fact. Before we can continue these sp3 installs, we need to be reasonably confident that we won't be surprised by this again.
                                  >
                                  > We now know how to look for pending file rename operations in
                                  "PendingFileRenameOperations" under HKLM\SYSTEM\CurrentControlSet\Control\Session Manager, and this check is part of our sp install process going forward.
                                  >
                                  > The question is: Is that it? Or do we need to check other things too? Does
                                  anyone know what is checked when determining a box must be rebooted before sp3 can be installed?
                                  > http://www.hsn.com/
                                  >

                                   

                                   

                                Your message has been successfully submitted and would be delivered to recipients shortly.