We run V8.5D of QAD Mfgpro with Progress version 8.3B(22).
We use TRW multi-db version of Mfgpro. Currently in order to backup our databases to a known position we have to shut them all down, split off a mirror and backup up the mirror copy, whilst restarting the databases on the remaining original disks. We re-sync the mirrors after the backup has been completed.WE HAVE 22 SEPARATE COUNTRY DATABASES AND OUR PROGRAMS CAN UPDATE MULTIPLE DATABASES IN ANY ONE TRANSACTION.
This necessitates a 25 minute plus shut down of the system, which we would like to avoid.
We have on occasion accomplished an on-line backup by using an "atomic" mirror split and backing up from the mirror whilst the data bases remain in use .(Clearly the mirror copy is not really being used but represents the last position of everything before the mirror was split with transactions in flight as it were)
We have also successfully recovered the databases from one of these "flying" backups by carrying out the following steps:-
Recover the files from backup.
Remove the *.lk lock files.
Truncate the Before Image files for all databases.
Restart the databases .
Does any one have any suggestions as to whether the "flying" backup technique is safe and reliable ?
Could proquiet and probkup on-line accomplish the same result?
Any other bright ideas would be welcome.
(I think a VXFS snap dump only works on single LV's and as we have multiple LV's it would be impossible to get a synchronous position.)
We use TRW multi-db version of Mfgpro. Currently in order to backup our databases to a known position we have to shut them all down, split off a mirror and backup up the mirror copy, whilst restarting the databases on the remaining original disks. We re-sync the mirrors after the backup has been completed.WE HAVE 22 SEPARATE COUNTRY DATABASES AND OUR PROGRAMS CAN UPDATE MULTIPLE DATABASES IN ANY ONE TRANSACTION.
This necessitates a 25 minute plus shut down of the system, which we would like to avoid.
We have on occasion accomplished an on-line backup by using an "atomic" mirror split and backing up from the mirror whilst the data bases remain in use .(Clearly the mirror copy is not really being used but represents the last position of everything before the mirror was split with transactions in flight as it were)
We have also successfully recovered the databases from one of these "flying" backups by carrying out the following steps:-
Recover the files from backup.
Remove the *.lk lock files.
Truncate the Before Image files for all databases.
Restart the databases .
Does any one have any suggestions as to whether the "flying" backup technique is safe and reliable ?
Could proquiet and probkup on-line accomplish the same result?
Any other bright ideas would be welcome.
(I think a VXFS snap dump only works on single LV's and as we have multiple LV's it would be impossible to get a synchronous position.)