Dave Markham wrote:
> Cheers. Yeah I guess i just wrote rack by habit :)
>
> I could create a separate policy in netbackup and do RAW file backup
> of the voting disk. Its another thing to admin schedule etc but it may
> work.
>
> For the moment i have just done the following in my
> bpstart_notify.<policy>
>
> change $OUTF
>
> #Edited by DaveM to create file per stream
> OUTF=/usr/openv/netbackup/bin/BPSTART_CALLED.$STREAM_NUMBER
>
> Then added the section at the end before the final exit.
>
>
> #Backup oracle voting disk
> VOTEDISK=/opt/app/oracle/oradev/crs/votedsk1
> STAGE=/usr/openv/netbackup/ORA_TEMP_AREA
>
> #Only do this once per backup
> #trying by STREAM_NUMBER=1
> if [ "$STREAM_NUMBER" = "1" ];then
> echo "Stream match at STREAM_NUMBER [$STREAM_NUMBER]. Running
> voting disk backup." >> $OUTF
> if [ ! -d "$STAGE" ];then
> mkdir $STAGE
> fi
>
> #Check voting disk is a sym link
> if [ -h "$VOTEDISK" ];then
> dd if=$VOTEDISK of=$STAGE/votingdisk bs=4k >/dev/null 2>&1
> #Exit 1 if above failed
> if [ $? -ne 0 ];then
> echo "dd process had a problem. Exiting" >>$OUTF
> exit 1
> fi
> else
> #Exit with error
> echo "Votingdisk [$VOTEDISK] wasnt a sym link. Exiting" >>$OUTF
> exit 1
> fi
> fi
>
> exit 0
>
>
>
> Jeff Lightner wrote:
>> RMAN can backup raw devices - we'll be doing that on our latest RAC
>> installation on Linux (the earlier one was using OCFS filesystem but
>> still had raw voting disk/OCR which is what prompted my question).
>> For that matter so can NBU itself can backup raw devices without RMAN.
>> We have some older systems that use raw devices for Oracle DB (no RAC)
>> and our backups for those are done by shutting down the database and
>> letting NBU do a raw device backup. We know it works because we've
>> used that backup to restore to another system when doing a data refresh.
>>
>> I'm just wondering about the voting disk and OCR for RAC since they are
>> part of CRS (a/k/a Oracle Clusterware). CRS is a higher level product
>> than the database (that is to say you install everything else AFTER you
>> have CRS working) so I was wondering if there was a backup/restore
>> methodology for CRS itself that included the voting disk and OCR.
>>
>> P.S. Probably was your spell checker "fixing" it for you but just in
>> case it wasn't - It is "RAC" not "rack". RAC = Real Application
>> Cluster.
>>
>> -----Original Message-----
>> From: Dave Markham [mailto:dave.markham AT fjserv DOT net] Sent: Friday,
>> February 01, 2008 5:24 AM
>> To: Jeff Lightner
>> Cc: Ed Wilts; veritas-bu AT mailman.eng.auburn DOT edu
>> Subject: Re: [Veritas-bu] Another bpstart, oracle and streams question
>>
>> Im no rack expert but the DBA's have said they need that file backing
>> up. As its a sym link to a raw slice im doing it via dd and the block
>> size they recommend.
>>
>> Im not sure if the rman job can actually do it as it may not be part
>> of the DB that Rman hooks into.
>>
>> Its only a 256mb file so i dont think it would harm to back it up if you
>>
>> are unsure.
>>
>> Are the STREAMS issue, im actually thinking what i want may be a bit
>> easier than running a hot oracle backup.
>>
>> I just need this job to run once so coulndt if do an if statement
>> something like below :-
>>
>> if [ $STREAM = 1 ]
>> then
>> do the checks and dd etc
>> fi
>>
>> This would then hold the first stream until the dd job had finished
>> and then back it up.
>>
>> Hmm one issue i guess is i need to make sure the stream im checking
>> for is the same stream as the file system the voting disk will be
>> backed up
>> to.
>>
>> Does anyone have confirmation of the STREAM variables passed to
>> bpstart_notify ??
>>
>> cheers
>>
>>
>> Jeff Lightner wrote:
>>
>>> Curious about this thread - so far as I know we're not doing
>>> anything special for backing up voting disk and OCR for Oracle RAC
>>> (unless the DBAs somehow set this up via RMAN to the policies we
>>> created in NBU). Is this something we should be doing?
>>>
>>>
>>>
>> ------------------------------------------------------------------------
>>
>>> *From:* veritas-bu-bounces AT mailman.eng.auburn DOT edu
>>> [mailto:veritas-bu-bounces AT mailman.eng.auburn DOT edu] *On Behalf Of *Ed
>>>
>> Wilts
>>
>>> *Sent:* Thursday, January 31, 2008 10:50 AM
>>> *To:* dave.markham AT fjserv DOT net
>>> *Cc:* veritas-bu AT mailman.eng.auburn DOT edu
>>> *Subject:* Re: [Veritas-bu] Another bpstart, oracle and streams
>>>
>> question
>>
>>> The short answer is: don't even try if you have multiple disks. You
>>> *may* be able to get away with it now using the parent/child job
>>> functionality that got added in 6.0 but we haven't tested this, having
>>>
>>
>>
>>> given up on a long time ago after a multi-year battle.
>>>
>>> Sure, you can get it work most of the time, and many people do.
>>> However, there are many, many edge cases where this will break down,
>>> and if you do go down this route, you'll find some them eventually.
>>> You'll find cases where the bpstart didn't run or the bpend didn't
>>> run, or they'll run twice.
>>>
>>> We definitely had a really, really bright guy working on the
>>> bpstart/bpend scripts and he could not solve the problem so that it
>>> was 100% reliable. We simply wanted to put Oracle into hotbackup mode,
>>>
>>
>>
>>> do the backups, and take it out of hotbackup mode. Like you said,
>>> you'll be writing some logic around it. However, you'll be writing and
>>>
>>
>>
>>> writing and writing, tweaking it regularly as you run into all of
>>> the edge conditions and you'll cuss and swear as you stumble over
>>> them. Bottom line: it can't be done reliably.
>>>
>>> Use a different method of coordinating this.
>>>
>>> .../Ed
>>>
>>> On Jan 31, 2008 8:57 AM, Dave Markham <dave.markham AT fjserv DOT net
>>> <mailto:dave.markham AT fjserv DOT net>> wrote:
>>>
>>> Guys im running Netbackup 6.0mp4 and a number of oracle SSO media
>>> servers backing themselves up. Solaris9/10
>>>
>>> This has been mentioned before many times im sure and i found a thread
>>> on it called 'Yet another bpstart/bpend questions' but i dont have a
>>> copy of it to reply to.
>>>
>>> Basically as im running a UNIX backup (forget oracle for the moment)
>>>
>> and
>>
>>> we have multistreams set and all_local_drives directive, a
>>> bpstart_notify script we have which does this (there are extra checks
>>> etc also):-
>>>
>>> dd if=/path/to/votedsk1 of=$STAGING/votingdisk bs=4k >/dev/null 2>&1
>>>
>>> ..is obviously running on each stream which is kicked off.
>>>
>>> I just need it to run once.
>>>
>>> There was talk about the bpstart script taking a STREAMS parameter but
>>> it doesnt say so within the /opt/openv/netbackup/bin/goodies version
>>>
>> so
>>
>>> can anyone confirm there are STREAMS variables passed to it?
>>>
>>> I was thinking ( perhaps like many people ) to create some lock file
>>>
>> and
>>
>>> then remove it. This leads to all sorts of issues regarding streams
>>> restarting, certain ones finishing before others etc and so i'd need
>>>
>> to
>>
>>> create some good logic around it.
>>>
>>> What are people doing in the wild?
>>>
>>> I just need the above command to run once per backup.
>>>
>>>
>>>
>>> --
>>> Ed Wilts, Mounds View, MN, USA
>>> mailto:ewilts AT ewilts DOT org <mailto:ewilts AT ewilts DOT org>
>>>
>>>
>> ----------------------------------
>> CONFIDENTIALITY NOTICE: This e-mail may contain privileged or
>> confidential information and is for the sole use of the intended
>> recipient(s). If you are not the intended recipient, any disclosure,
>> copying, distribution, or use of the contents of this information is
>> prohibited and may be unlawful. If you have received this electronic
>> transmission in error, please reply immediately to the sender that
>> you have received the message in error, and delete it. Thank you.
>> ----------------------------------
>>
>>
>
>
_______________________________________________
Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
|
