Migrate to new box using ontape w/STDIO

We have a new AIX server and would like to use ontape with STDIO to migrate the IDS instance. I have used ontape w/STDIO to set up HDR. In this case I would like to use ontape -p via STDIO and only do a physical restore with no logical restore. Is this possible? I plan on stopping all access to the source server and then start the backup across ssh. There will be no reason to do the logical restore. Prior to starting the backup all users will be off and I will run onmode -c, onmode -l, onmode -c and flush the logical logs to the storage manager. On the new server the physical restore will finish and the instance will be in fast recovery. How to I get the instance online without doing the logical restore? AIX 5.2/5.3 IDS 10.00.FC4 Thanks Roy

Roy Mercer wrote: > We have a new AIX server and would like to use ontape with STDIO to > migrate the IDS instance. I have used ontape w/STDIO to set up > HDR. In this case I would like to use ontape -p via STDIO and only > do a physical restore with no logical restore. Is this possible? Yes > > I plan on stopping all access to the source server and then start the > backup across ssh. There will be no reason to do the logical > restore. Prior to starting the backup all users will be off and I > will run onmode -c, onmode -l, onmode -c and flush the logical logs to > the storage manager. If you only to physical restore - then you will be restoring up to the last checkpoint. I would suggest the following process... 1) create a true HDR environment with the new server as the secondary. 2) continue operating on the primary. The logs will be pushed from the primary to the secondary. 3) bring down the primary and make the existing server into the primary 4) run onmode -d standard on the new primary (and new system). > > On the new server the physical restore will finish and the instance > will be in fast recovery. How to I get the instance online without > doing the logical restore? See above... > > AIX 5.2/5.3 IDS 10.00.FC4 > > Thanks > Roy >

Madison, Good point. Here are the reasons for not doing this. We load tables everyday using HPL express mode. Load times in deluxe mode would take too long. Also, the fuzzy checkpoints are helping out in this case. This is a web facing server and the checkpoint times can not afford to be more than 1 second. I have thought about this as an option but would rather just do a clean cutover. I know I can make sure no more transactions go through the server while I am doing the backup. The only thing I need is what steps to take to get the new server out of fast recovery without doing a logical restore. Any tips? If not then I will proceed with the HDR and just make the secondary server the standard server. Roy On Jul 11, 10:06 am, Madison Pruet <mpru...@verizon.net> wrote: > Roy Mercer wrote: > > We have a new AIX server and would like to use ontape with STDIO to > > migrate the IDS instance. I have used ontape w/STDIO to set up > > HDR. In this case I would like to use ontape -p via STDIO and only > > do a physical restore with no logical restore. Is this possible? > > Yes > > > > > I plan on stopping all access to the source server and then start the > > backup across ssh. There will be no reason to do the logical > > restore. Prior to starting the backup all users will be off and I > > will run onmode -c, onmode -l, onmode -c and flush the logical logs to > > the storage manager. > > If you only to physical restore - then you will be restoring up to the > last checkpoint. I would suggest the following process... > > 1) create a true HDR environment with the new server as the secondary. > > 2) continue operating on the primary. The logs will be pushed from the > primary to the secondary. > > 3) bring down the primary and make the existing server into the primary > > 4) run onmode -d standard on the new primary (and new system). > > > > > On the new server the physical restore will finish and the instance > > will be in fast recovery. How to I get the instance online without > > doing the logical restore? > > See above... > > > > > > > AIX 5.2/5.3 IDS 10.00.FC4 > > > Thanks > > Roy- Hide quoted text - > > - Show quoted text -

Madison, In this case I would like to only restore to the last checkpoint. I am going to make sure everyone is off and manually switch the logs and start a checkpoint. As this point I will be where I would like the new server to take over. I did just try this and after the physical restore was done, I ran onmode -m.The online log indicated that the logical restore would be skipped. I thought in earlier releases this could not be done. In this version it let me do it. online log output 10:08:47 Physical Restore of rootdbs, physdbs01, llogdbs01, xxxdbs01,xxxdbs0 2, xxxdbs03, xxxdbs04, xxxdbs06, xxxdbs07, xxxdbs08, xxxxldbs02, xxxxbs10 , xxxxxxdbs01, xxxxxirdbs01, xxxxiridx01, xxxxldbs01, xxxxbs12k Completed. 10:08:47 Checkpoint Completed: duration was 0 seconds. 10:08:47 Checkpoint loguniq 1011007, logpos 0x12018, timestamp:0xe301cb71 10:32:19 No logical log restore will be performed. 10:32:19 Preparing Physical Log for Fast Recovery ... 10:32:19 Clearing the physical and logical logs has started 10:33:06 Cleared 3194 MB of the physical and logical logs in 46seconds 10:33:06 Physical Recovery Started at Page (2:31020). 10:33:06 Physical Recovery Complete: 0 Pages Examined, 0 Pages Restored. 10:33:07 Logical Recovery Started. 10:33:07 10 recovery worker threads will be started. Stopping the webservers and changing the sqlhosts file will guarantee that no changes will be made to the database during the transfer. Thanks Roy On Jul 11, 10:30 am, Roy Mercer <roy.mer...@gmail.com> wrote: > Madison, > > Good point. > Here are the reasons for not doing this. > We load tables everyday using HPL express mode. Load times in deluxe > mode would take too long. > Also, the fuzzy checkpoints are helping out in this case. This is a > web facing server and the checkpoint times can not afford to be more > than 1 second. > > I have thought about this as an option but would rather just do a > clean cutover. > I know I can make sure no more transactions go through the server > while I am doing the backup. > The only thing I need is what steps to take to get the new server out > of fast recovery without doing a logical restore. > > Any tips? If not then I will proceed with the HDR and just make the > secondary server the standard server. > > Roy > > On Jul 11, 10:06 am, Madison Pruet <mpru...@verizon.net> wrote: > > > > > Roy Mercer wrote: > > > We have a new AIX server and would like to use ontape with STDIO to > > > migrate the IDS instance. I have used ontape w/STDIO to set up > > > HDR. In this case I would like to use ontape -p via STDIO and only > > > do a physical restore with no logical restore. Is this possible? > > > Yes > > > > I plan on stopping all access to the source server and then start the > > > backup across ssh. There will be no reason to do the logical > > > restore. Prior to starting the backup all users will be off and I > > > will run onmode -c, onmode -l, onmode -c and flush the logical logs to > > > the storage manager. > > > If you only to physical restore - then you will be restoring up to the > > last checkpoint. I would suggest the following process... > > > 1) create a true HDR environment with the new server as the secondary. > > > 2) continue operating on the primary. The logs will be pushed from the > > primary to the secondary. > > > 3) bring down the primary and make the existing server into the primary > > > 4) run onmode -d standard on the new primary (and new system). > > > > On the new server the physical restore will finish and the instance > > > will be in fast recovery. How to I get the instance online without > > > doing the logical restore? > > > See above... > > > > AIX 5.2/5.3 IDS 10.00.FC4 > > > > Thanks > > > Roy- Hide quoted text - > > > - Show quoted text -- Hide quoted text - > > - Show quoted text -

Roy Mercer wrote: > Madison, > > In this case I would like to only restore to the last checkpoint. I > am going to make sure everyone is off and manually switch the logs and > start a checkpoint. As this point I will be where I would like the > new server to take over. > > I did just try this and after the physical restore was done, I ran > onmode -m.> The online log indicated that the logical restore would be skipped. > I thought in earlier releases this could not be done. In this > version it let me do it. > > online log output > 10:08:47 Physical Restore of rootdbs, physdbs01, llogdbs01, xxxdbs01,> xxxdbs0 > 2, xxxdbs03, xxxdbs04, xxxdbs06, xxxdbs07, xxxdbs08, xxxxldbs02, > xxxxbs10 > , xxxxxxdbs01, xxxxxirdbs01, xxxxiridx01, xxxxldbs01, xxxxbs12k > Completed. > 10:08:47 Checkpoint Completed: duration was 0 seconds. > 10:08:47 Checkpoint loguniq 1011007, logpos 0x12018, timestamp:> 0xe301cb71 > > 10:32:19 No logical log restore will be performed. > 10:32:19 Preparing Physical Log for Fast Recovery ... > 10:32:19 Clearing the physical and logical logs has started > 10:33:06 Cleared 3194 MB of the physical and logical logs in 46> seconds > 10:33:06 Physical Recovery Started at Page (2:31020). > 10:33:06 Physical Recovery Complete: 0 Pages Examined, 0 Pages > Restored. > 10:33:07 Logical Recovery Started. > 10:33:07 10 recovery worker threads will be started.> > Stopping the webservers and changing the sqlhosts file will guarantee > that no changes will be made to the database during the transfer. > > Thanks > Roy > > > On Jul 11, 10:30 am, Roy Mercer <roy.mer...@gmail.com> wrote: >> Madison, >> >> Good point. >> Here are the reasons for not doing this. >> We load tables everyday using HPL express mode. Load times in deluxe >> mode would take too long. >> Also, the fuzzy checkpoints are helping out in this case. This is a >> web facing server and the checkpoint times can not afford to be more >> than 1 second. >> >> I have thought about this as an option but would rather just do a >> clean cutover. >> I know I can make sure no more transactions go through the server >> while I am doing the backup. >> The only thing I need is what steps to take to get the new server out >> of fast recovery without doing a logical restore. >> >> Any tips? If not then I will proceed with the HDR and just make the >> secondary server the standard server. >> >> Roy >> >> On Jul 11, 10:06 am, Madison Pruet <mpru...@verizon.net> wrote: >> >> >> >>> Roy Mercer wrote: >>>> We have a new AIX server and would like to use ontape with STDIO to >>>> migrate the IDS instance. I have used ontape w/STDIO to set up >>>> HDR. In this case I would like to use ontape -p via STDIO and only >>>> do a physical restore with no logical restore. Is this possible? >>> Yes >>>> I plan on stopping all access to the source server and then start the >>>> backup across ssh. There will be no reason to do the logical >>>> restore. Prior to starting the backup all users will be off and I >>>> will run onmode -c, onmode -l, onmode -c and flush the logical logs to >>>> the storage manager. >>> If you only to physical restore - then you will be restoring up to the >>> last checkpoint. I would suggest the following process... >>> 1) create a true HDR environment with the new server as the secondary. >>> 2) continue operating on the primary. The logs will be pushed from the >>> primary to the secondary. >>> 3) bring down the primary and make the existing server into the primary >>> 4) run onmode -d standard on the new primary (and new system). >>>> On the new server the physical restore will finish and the instance >>>> will be in fast recovery. How to I get the instance online without >>>> doing the logical restore? >>> See above... >>>> AIX 5.2/5.3 IDS 10.00.FC4 >>>> Thanks >>>> Roy- Hide quoted text - >>> - Show quoted text -- Hide quoted text - >> - Show quoted text - > > Put the server in quiescent mode... This will assure you that no activity will be done... Regards. -- Fernando Nunes Portugal http://informix-technology.blogspot.com My email works... but I don't check it frequently...

Fernando Nunes wrote: > Roy Mercer wrote: >> Madison, >> >> In this case I would like to only restore to the last checkpoint. I >> am going to make sure everyone is off and manually switch the logs and >> start a checkpoint. As this point I will be where I would like the >> new server to take over. >> >> I did just try this and after the physical restore was done, I ran >> onmode -m.>> The online log indicated that the logical restore would be skipped. >> I thought in earlier releases this could not be done. In this >> version it let me do it. >> >> online log output >> 10:08:47 Physical Restore of rootdbs, physdbs01, llogdbs01, xxxdbs01,>> xxxdbs0 >> 2, xxxdbs03, xxxdbs04, xxxdbs06, xxxdbs07, xxxdbs08, xxxxldbs02, >> xxxxbs10 >> , xxxxxxdbs01, xxxxxirdbs01, xxxxiridx01, xxxxldbs01, xxxxbs12k >> Completed. >> 10:08:47 Checkpoint Completed: duration was 0 seconds. >> 10:08:47 Checkpoint loguniq 1011007, logpos 0x12018, timestamp:>> 0xe301cb71 >> >> 10:32:19 No logical log restore will be performed. >> 10:32:19 Preparing Physical Log for Fast Recovery ... >> 10:32:19 Clearing the physical and logical logs has started >> 10:33:06 Cleared 3194 MB of the physical and logical logs in 46>> seconds >> 10:33:06 Physical Recovery Started at Page (2:31020). >> 10:33:06 Physical Recovery Complete: 0 Pages Examined, 0 Pages >> Restored. >> 10:33:07 Logical Recovery Started. >> 10:33:07 10 recovery worker threads will be started.>> >> Stopping the webservers and changing the sqlhosts file will guarantee >> that no changes will be made to the database during the transfer. >> >> Thanks >> Roy >> >> >> On Jul 11, 10:30 am, Roy Mercer <roy.mer...@gmail.com> wrote: >>> Madison, >>> >>> Good point. >>> Here are the reasons for not doing this. >>> We load tables everyday using HPL express mode. Load times in deluxe >>> mode would take too long. >>> Also, the fuzzy checkpoints are helping out in this case. This is a >>> web facing server and the checkpoint times can not afford to be more >>> than 1 second. >>> >>> I have thought about this as an option but would rather just do a >>> clean cutover. >>> I know I can make sure no more transactions go through the server >>> while I am doing the backup. >>> The only thing I need is what steps to take to get the new server out >>> of fast recovery without doing a logical restore. >>> >>> Any tips? If not then I will proceed with the HDR and just make the >>> secondary server the standard server. >>> >>> Roy >>> >>> On Jul 11, 10:06 am, Madison Pruet <mpru...@verizon.net> wrote: >>> >>> >>> >>>> Roy Mercer wrote: >>>>> We have a new AIX server and would like to use ontape with STDIO to >>>>> migrate the IDS instance. I have used ontape w/STDIO to set up >>>>> HDR. In this case I would like to use ontape -p via STDIO and only >>>>> do a physical restore with no logical restore. Is this possible? >>>> Yes >>>>> I plan on stopping all access to the source server and then start the >>>>> backup across ssh. There will be no reason to do the logical >>>>> restore. Prior to starting the backup all users will be off and I >>>>> will run onmode -c, onmode -l, onmode -c and flush the logical logs to >>>>> the storage manager. >>>> If you only to physical restore - then you will be restoring up to the >>>> last checkpoint. I would suggest the following process... >>>> 1) create a true HDR environment with the new server as the secondary. >>>> 2) continue operating on the primary. The logs will be pushed from the >>>> primary to the secondary. >>>> 3) bring down the primary and make the existing server into the primary >>>> 4) run onmode -d standard on the new primary (and new system). >>>>> On the new server the physical restore will finish and the instance >>>>> will be in fast recovery. How to I get the instance online without >>>>> doing the logical restore? >>>> See above... >>>>> AIX 5.2/5.3 IDS 10.00.FC4 >>>>> Thanks >>>>> Roy- Hide quoted text - >>>> - Show quoted text -- Hide quoted text - >>> - Show quoted text - >> >> > > Put the server in quiescent mode... This will assure you that no > activity will be done... > > Regards. > onmode -c block Use dd and remsh to copy the chunks onbar -r -e onmode -m :D