Hopefully your database server is also a media server.
The command we use on our remaining HP-UX 11.11 media server (running NBU 7.6.x but same was used back when we had 6.5.x) is:
/usr/openv/netbackup/bin/bpbackup -p <policy> -s <schedule> -i" >> dbbackup.log 2>&1
We define the policy in NetBackup using multiple streams but do not set any days in the schedules for “Start Window” tab. We set the frequency to allow it
to be run every 12 hours.
We do multiple streams to backup separate subdirectories for the database to make it run faster – one could specify separate data and index files in the streams
if all in one directory – we used to have separate filesystems so maintained separate directories when we went to a single filesystem to make this backup stream assignment more manageable.
Hot backup mode in Oracle causes a lot of logging while it is running so if left running too long might fill up your redolog or archive log dump spaces.
We do hot backup mode but only to do the split of data to other host so it only runs on our main database for a few minutes so we are doing a point in time backup which is probably what you did.
One thought – I haven’t worked with NetApp in years but I notice you said “snapshot” rather than “BCV” so I’m wondering if your NetApps might allow for full
copy vs snapshot. We did full copies when we had EMC Symmetrix then DMX and now do same on our Hitachi (they call it Shadow Image) and it is the full copy (i.e. a mirror of the original data) that we break off and mount on our backup server. We do it this
way because snapshots create pointers back to original data and we don’t want Production during the day impacted by the backup reads.
Flow is generally:
1)
Start background copy of production disks to other disks (or LUNs) using the disk array’s facility for this (BCV or Shadow Image in our case).
2)
Once the background disks (LUNs) are fully synced we put the database into hot backup mode.
3)
We split the backgrounds disks (LUNs) from the production disks.
4)
We take the database out of hot backup mode.
5)
We mount the background disks on the backup server.
6)
We do the backup (usually runs from early morning to midday) on the backup server.
7)
In the evening we unmount the background disks (LUNs) from the backup server and start again from step 1.
For the unmount and mount we’re also exporting LVM VGs (previously deported VxVM DGs) before the unmount then doing import before
the mount.
From: veritas-bu-bounces AT mailman.eng.auburn DOT edu [mailto:veritas-bu-bounces AT mailman.eng.auburn DOT edu]
On Behalf Of Sid Shapiro
Sent: Wednesday, July 15, 2015 6:36 PM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] older version issue.... netbackup 6.5.6
Hello,
I have a problem with netbackup 6.5.6 running on hp-ux (pa-risc) 11.11 (running oracle 7.3.4)
I am aware that almost everything I've jsut described is old and unsupported, and please don't suggest upgrading, unfortunately this is not possible.
Background (kind of long-winded, but I want to tell you as much as I can to forestall unrelated questions):
I have a database that is currently running on an EMC cx3-80 backend. It is being backed up via snapshots which are then mounted on another hp-ux 11.11 system running netbackup 6.5. The database files (snapshot) are then backed up using
a standard filesystem-type backup scheduled through the normal netbackup policies, etc. (oh - btw - the netbackup master server is 7.6.1)
This all works reasonably well.
We need to junk the cx3-80 in favor of a NetApp. The NetApp will not do the same snapshotting that the cx does, so generally we have moved most of our databases to RMAN backups, which is working pretty well.
However we can't run rman on the 7.3.4 database, and we can't go past 6.5.6 on hp-ux 11.
So the plan is to go to hot backup for this one database.
We have all sorts of hot backup scripts that we have run over the years and all work well, except in this case. I've burrowed down to the problem being that the bpbackup command, which is being called directly from the hot backup script,
is returning a
"EXIT STATUS 29: failed trying to exec a command".
We've chased this a bit. I've run "bp' to backup some files and that works fine. I've captured the arguments that bp uses to call bpbackup and tried to use the exact same command and arguments calling bpbackup directly. bp works, bpbackup
does not, even though they are running the exact same bpbackup command.
I've tried the bpbackup command on various other systems: aix 7.2, hp-ux 11.31 - all work correctly. It is only on hp-ux 11.11 and 11.0 that I'm seeing the problem.
symantec says error 29 indicates a permission and/or resource problem. Obviously that is not the case if bp calling bpbackup works and calling bpbackup directly does not. There is something else going on.
btw - the command I'm trying is "bpbackup -S herhxp03 /etc/passwd" - I don't think I can get much more basic that that.
I'm hoping that someone who has been around netbackup for a while might have some thoughts of either things to try or places to look or something!
Bio-Rad Corporate IT - Desk: (510) 741-6846 Mobile: (510) 224-4343
