This is a multipart message in MIME format.
--=_alternative 005D2EF480257130_=
Content-Type: text/plain; charset="us-ascii"
Hi Dave,
Assuming you have to follow this method its not a problem so long as:
1) you don't mind having root level passwordless access from the
master/media server to the clients host (via rsh or ssh public/private
keys).
2) you don't mind having restore over the network, or you don't mind
having to manually move the backup disk between media server and client.
The backup will be of the media server where u mount the backup disk, so
that is where you need to code your notify scripts to do the necessary
disk mount/unmounting....
cheers Andy.
Ok, been tasked with giving my input on a backup solution for a new
project. Someone had designed it to start backing up things using
scripts from cron calling other scripts and then netbackup bpbackup
commands and all sorts of random junk.
Anyway assume we are talking Solaris 8 or 9 with fiber attached units.
The question i have which was from the original design ( and i guess
still a requirement ) is how do people go around backing up database
volumes. I myself have had dbas ( running oracle ) to do hot_archives
and then backup that. The hot_archives are called from bpstart_notify so
are captured within netbackup if they take too long or if they fail. I
can also do some checking if directories exist etc before running the
oracle backup. Also have the added benefit of not writing to tape until
exit call from hot_archive which means db backup has finished.
Ok, thats one way, the other way i do on another project is use SSO
option on the database servers which have dedicated HBA's through fiber
to the tape units and shared drives. I just point at a location and
assume everything in there is ready to be backed up. This project is
slightly different and the DBA's are responsible for the backup.
The way this new project is wanting to go is by mounting a storage
volume ( disk ) on the db server and writing a backup to it. then
unmounting it and then mounting it on the backup server to be written to
tape directly.
What i want to know is how ( if we have to go this route ) can i control
this within netbackup and not start backups unless mount and unmount
requests have been done?
Do i have bpstart scripts on the db servers, which do the mounts, then
the oracle backups and then the unmounts. If this fails can i send
something back through bpstart so the backup doesnt run and fails with a
certain netbackup code?
The benefit of the mount, unmount is so the resource is not used on the
db servers to do backups, but im not sure how i am going to control the
process within netbackup to capture any errors. So has anyone done
anything similar which they can advise with??
Thanks
Dave
--=_alternative 005D2EF480257130_=
Content-Type: text/html; charset="us-ascii"
<br><font size=2 face="sans-serif">Hi Dave,</font>
<br>
<br><font size=2 face="sans-serif"> Assuming you have to follow this
method its not a problem so long as:</font>
<br>
<br><font size=2 face="sans-serif">1) you don't mind having root level
passwordless access from the master/media server to the clients host (via rsh
or ssh public/private keys).</font>
<br>
<br><font size=2 face="sans-serif">2) you don't mind having restore over the
network, or you don't mind having to manually move the backup disk between
media server and client.</font>
<br>
<br><font size=2 face="sans-serif">The backup will be of the media server where
u mount the backup disk, so that is where you need to code your notify scripts
to do the necessary disk mount/unmounting.... </font>
<br>
<br><font size=2 face="sans-serif">
cheers Andy.</font>
<br>
<br>
<br>
<br>
<br><font size=3 face="Courier New">Ok, been tasked with giving my input on a
backup solution for a new<br>
project. Someone had designed it to start backing up things using<br>
scripts from cron calling other scripts and then netbackup bpbackup<br>
commands and all sorts of random junk.<br>
<br>
Anyway assume we are talking Solaris 8 or 9 with fiber attached units.<br>
<br>
The question i have which was from the original design ( and i guess<br>
still a requirement ) is how do people go around backing up database<br>
volumes. I myself have had dbas ( running oracle ) to do hot_archives<br>
and then backup that. The hot_archives are called from bpstart_notify so<br>
are captured within netbackup if they take too long or if they fail. I<br>
can also do some checking if directories exist etc before running the<br>
oracle backup. Also have the added benefit of not writing to tape until<br>
exit call from hot_archive which means db backup has finished.<br>
<br>
Ok, thats one way, the other way i do on another project is use SSO<br>
option on the database servers which have dedicated HBA's through fiber<br>
to the tape units and shared drives. I just point at a location and<br>
assume everything in there is ready to be backed up. This project is<br>
slightly different and the DBA's are responsible for the backup.<br>
<br>
The way this new project is wanting to go is by mounting a storage<br>
volume ( disk ) on the db server and writing a backup to it. then<br>
unmounting it and then mounting it on the backup server to be written to<br>
tape directly.<br>
<br>
What i want to know is how ( if we have to go this route ) can i control<br>
this within netbackup and not start backups unless mount and unmount<br>
requests have been done?<br>
<br>
Do i have bpstart scripts on the db servers, which do the mounts, then<br>
the oracle backups and then the unmounts. If this fails can i send<br>
something back through bpstart so the backup doesnt run and fails with a<br>
certain netbackup code?<br>
<br>
The benefit of the mount, unmount is so the resource is not used on the<br>
db servers to do backups, but im not sure how i am going to control the<br>
process within netbackup to capture any errors. So has anyone done<br>
anything similar which they can advise with??<br>
<br>
Thanks<br>
Dave<br>
</font>
--=_alternative 005D2EF480257130_=--
|
