This is almost the exact scenario I have. However, I need to ask these
questions:
How is the data moving? I use Fiber Channel to connect from the server to the
Data Domain and DD to SL8500. How the data moves will impact how many media
servers you need.
How fast do you need the data moved? If
If your tape drives are maxed out and jobs are queued too long, think about
putting storage in the middle.
Our infrastructure is fiber based not 10G, so we got two DD990 as an
intermediary backup target and now all jobs write there (we have over 200
drives defined) and once backup jobs
tape...
-Original Message-
From: Justin Piszcz [mailto:jpis...@lucidpixels.com]
Sent: Thursday, December 19, 2013 1:20 PM
To: David McMullin; veritas-bu@mailman.eng.auburn.edu
Subject: RE: [Veritas-bu] SAP backups / Online / Redo
Dave,
Curious how you determined/measured the available
I just upgraded from 7.0.1 to 7.5.0.3
I opened a thread on the Symantec Forum with a list of things I wish I had
known.
http://www.symantec.com/connect/forums/nb-7503-gotcha-list
The actual upgrade went very well.
1. I use SLP, I had to get an EEB before the SLP would start. (Symantec Bug
Not sure about documentation, but we do this for our Oracle RMAN backups on 7.1
as well as on 7.5 now.
Any user directed backup should work.
Main policy calls script using client name of the cluster
Script uses uname to determine which cluster member active, uses policy value
that matches
If you have backup schedules stepping on each other, you might try this...
I run the nbpemreq -predict_all command at midnight and have my operations
staff check off the backups as they run. Now you have a paper trail for audits
as well...
Run from cron at 00:00, use command nbpemreq
What is the output of this command:
/usr/openv/volmgr/bin/vmpool -list_scratch
Is your defined scratch pool what you expected?
I run a cron job that verifies my available scratch tapes and sets my pool
several times a day, ever since I found there is a bug in the java gui that can
set a pool
Check the /etc/hosts file, I had a similar problem and found it must have a
loopback address defined -
127.0.0.1 localhost loghost
Simple, but easily missed.
--
Message: 8
Date: Thu, 2 Feb 2012 10:30:38 +
From: Mr Crosby mr.cro...@gmail.com
Subject:
I have IBM LTO-5 drives, Production are at B6W2, Test are at A5R0 and A9Q5.
I have seen read/write errors, but mostly as a result of excessive head wear.
I would suggest keeping an eye on how often the drives request cleanings -
symptomatically if you see them starting to request cleanings more
Jack - what size buffer parameters are you using with your LTO-5?
We have had to replace virtually ALL our LTO-5 drives due to tape head wear -
what brand of tapes do you use?
Agree regarding performance, in fact the limiting factor is more often
non-drive, i.e. hba, or source system just
Be aware - the hoursago default is 24 hours - If you do not have it you get 24,
you may need to extend that to 48 or 72 depending on your longest running
backup. The date is based on backup START time, so if your backup started more
than 24 hours ago you MUST put extended time.
I set up
If you do windows backups, you need to also multiply # of drives with all
local drives - essentially how many 'child' processes can be started by your
parent job as well. Figure - how many jobs are running at once? Each job will
get the memory allocated to it.
Buffer settings?
It will depend
Wayne - You are looking at this from the wrong perspective.
You need to be concerned about RESTORING your data.
It does not matter how 'successful' your backups are if you cannot restore the
data.
IMHO you really need to get a DBA or 'someone' to sign off on your backup
procedure, as well as
Wayne - you are right - I totally misread the message thread, my comments were
addressed to the original questioner.
Your comments were an excellent summation of the options to backup the data.
My comments were basically meant to indicate that how you want to restore the
data, and where, will
You did not mention your OS, on unix I find these commands helpful:
List of incomplete SLP jobs:
bpimagelist -L -idonly -hoursago 72 -stl_incomplete | sort +3r +4r +5rn
(without the hoursago value, it defaults to only 24)
nbstlutil is the command to control the SLP jobs:
nbstlutil inactive
My co-worker Nate found this today:
You can find more details at the link below, but the summary is this: If you've
expanded an NTFS volume and the $BITMAP part of the MFT becomes fragmented in
the process, and you take a FlashBackup with the Exclude unused and deleted
blocks option Enabled
Depending on your environment, you can define it in your vm.conf
# cat /usr/openv/volmgr/vm.conf
ACS_UNKNOWN=HCART
ACS_LTO_1_5T=HCART
I have an acsls, so that is how it is defined. I chose hcart, since I was not
using it already.
Apparently there is no hard and fast rule yet about LTO5, so you
Please note - many software, in their wisdom, have directory and files named or
beginning with core, including pdde, jdk, tomcat and oracle.
find / -name core* -exec ls -lad {} \;
I do not exclude core, since my failure definition is not being able to
restore, not backing up too much stuff!
I
Regarding disk usage:
Run this command sequence:
# To list values for logging
for i in `vxlogcfg -l -p 51216| grep -v ist`
do
echo $i /tmp/current.vxlog.config
vxlogcfg -l -p 51216 -o $i /tmp/current.vxlog.config
done
Review what is in the /tmp/current.vxlog.config file
Here is how we resolved this - it is a feature of linux...
# cat /etc/host.conf
#
# /etc/host.conf - resolver configuration file
#
# Please read the manual page host.conf(5) for more information.
#
#
# The following option is only used by binaries linked against
# libc4 or libc5. This line
My standard is to make a policy that never runs named ARCHIVED and put the
clients there.
If clients are not in a policy they do not show up in various lists, NB should
scan the images directory for them but does not...
It is also easier when installing to put them in a policy that does not run
Regarding your request and the responses, some thoughts:
If you HAVE to HAVE the DATA...
1. immediately FREEZE ALL SCRATCH TAPES AND/OR MOVE THEM to an unused pool - so
you don't overwrite the tapes. You will need to leave them there until you find
the tape. Restoring the catalog from a week
Please note this potential issue with OpsCenter!
From the manual:
If you have NOM you need to upgrade to Ops Center, but if you do that and your
master server is still on 6.5.x you will need to install and config an agent to
monitor the 6.5.x server. But if you upgrade right away you don't
of sometimes being unable to determine what
tapes are needed until we try to restore and it asks for an unexpected tape...
-Original Message-
From: Lightner, Jeff [mailto:jlight...@water.com]
Sent: Wednesday, May 19, 2010 4:27 PM
To: David McMullin; veritas-bu@mailman.eng.auburn.edu
I am working with my DBA and Symantec, opened a ticket and found this issue
from the RMAN side:
Here is my note to them:
Please review case # 291-053-107.
Our Oracle DBA are extremely upset that the tools available to them to
troubleshoot in a DR situation are so poor.
They have no reliable way
Perhaps it is an oracle thing?
I had our DBA remove the check logical parameter and our throughput improved
10X
--
Message: 11
Date: Tue, 9 Mar 2010 13:11:49 -0600
From: Heathe Yeakley hkyeak...@gmail.com
Subject: [Veritas-bu] Tru64 NetBackup Performance
To:
I have performed this rename on Solaris, but used the process mentioned below
as a 'cheat'
Production master name is foo, at ip address 1.1.1.1
Copy is bar, at 1.1.1.2
On bar - edit /etc/hosts and add entry for foo at 1.1.1.2
Set DNS to use files first then DNS
On media servers, do the same!
Justin -
You should look at your storage units and drives. I set mine up in vertical
lines, do you have highly used storage units all in the top LSM?
I did a Google search and found this link:
https://twiki.cern.ch/twiki/bin/view/FIOgroup/FreeCellsBalancer
I will just shut up and let you
NEW in 6.5.4 the RMAN parent job seems to maintain ownership of the child job
tapes, so that your jobs reuse the same tapes and don't unload them between
jobs.
This process bypasses priority, so even if another job - like a restore or
duplication is waiting for that tape, the child backup
Harry - use the ACSLS to take care of this. Here is more than you ever wanted
to know!
I share an SL8500 between NB and an AS/400, it works fine.
I use Access Control, and watch_vols to assign tapes, I never see the AS/400
tapes in my inventory.
(go to SUN web site and find pdf ACSLS73ADM.pdf
Running 6.5.3.1 on Solaris Master/Media servers.
We backup to a VTL, and duplicate to tape using SLP - we have had no issues
restoring from either VTL or tape.
RMAN asks NetBackup for the image, and as long as NetBackup knows where it is,
(i.e. where is primary copy) it restores fine.
Message:
- So what I understand you are doing is using the SLP to get around the
boundaries of tape pools being assigned to a given media, this way all the data
gets backed up by a pool of all of the tapes as one group?
- Yes, we backup to disk into one or two pools, with a short retention (SLP
This is exactly what I found, and SLP (storage lifecycle policies) are my
solution.
You set up the SLP to backup to disk pool A as step one, and have a second step
that copies that from disk to tape pool B with your normal retention.
Essentially you get a one tape drive vault process, and no
One thing I found that is helpful on our ACSLS is this command:
cmd_proc_shell - allows you to run acsls commands in a shell script.
So for example, this will dismount all drives from one LSM ( I have separate
scripts so I can be particular )
# cat dismount.all.0.ksh
#!/usr/bin/ksh
#
# uses
From my java admin console doing a restore I have the option of previewing the
restore. This lists the tapes needed and I can check to ensure they are in my
robot.
I just checked the BAR console, and there is a button on the left side that
allows you to preview media.
You may need to select
The vault selection screen allows you to select the date or time range to vault
- if set to only vault yesterday backups, it will not vault backups from prior
days.
Date: Thu, 18 Jun 2009 15:08:54 -0400
From: JBrownell netbackup-fo...@backupcentral.com
Subject: [Veritas-bu] Temporarily
In response to request for unified logging thoughts:
I have set up a separate unix file system for all my logs - both unified logs
and netbackup ones.
Unified logs use a -p and -o designation, -p should be 51216 for NetBackup, and
to list your -o use this command:
vxlogcfg -l -p 51216
So -
Regarding moving master server OS:
1. Install the new server OS and NetBackup
2. Stop NetBackup on old environment
3. Cold Catalog Backup
4. Rename old master and new master - so new master has old name
5. Start NetBackup on new master
6. Catalog Restore from Cold Catalog Backup
7. TEST TEST
Verdict is in - multiple sources advise against the process I mentioned.
Guess the moral of the story is to stay in a *nix environment...
From: Ed Wilts [mailto:ewi...@ewilts.org]
Sent: Wednesday, May 06, 2009 2:34 PM
To: David McMullin
Cc: veritas-bu@mailman.eng.auburn.edu
Subject: Re
and reconnecting in our early configuration are very happy with them.
Since we configured them to one OS we have had no issues.
David McMullin
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas
should play nicely with the
existing unix servers, and it might be better to have windows connections to
backup the VMWare - is this a false assumption?
David McMullin
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http
The behavior you describe is what you will see with the SQL and oracle agents -
A parent jobs starts and spawns child jobs. The number of these child jobs and
how many run at once - you can run them serially or parallel - is controlled by
variables in the script NetBackup calls.
I agree with
drive serial numbers.
You must configure these devices manually. * is this what you are seeing?
David McMullin
--
Date: Wed, 11 Mar 2009 13:30:58 +
From: Clooney, David david.cloo...@bankofamerica.com
Subject: [Veritas
(on unix)This command is your friend:
{install_path}/volmgr/bin/vmpool -list_scratch
If there are no configured scratch pools, this command will fix it:
{install_path}/volmgr/bin/vmpool -set_scratch scratch_pool_name
FYI - There is another known issue with windows java console, where
!
David McMullin
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
the midnight full backup
ran.
My operators now know to look for log running jobs of this policy, but it would
be nicer if SQL and NetBackup worked better together.
David McMullin
-Original Message-
From: bob944 [mailto:bob...@attglobal.net]
Sent: Saturday, February 21, 2009 1:22 PM
I have MSSQL backups running to Windows servers.
They call $ALL databases.
Every now and then, (is it a windows 'feature' - I am exposing my unix bias),
they just ... stop responding to NetBackup.
The parent job starts and calls some number of child jobs, the child jobs
complete and the
is that this is not their problem. Any ideas?
David McMullin
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
more
robust and have a timeout set for the bsa_put() and/or the VxBSAInit() function
call.
David McMullin
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
for a select time,
your backups may have already expired...
Good luck
David McMullin
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
AFAIK - Here is the key - You need to insure whatever action you take is in
line with EXISTING retention policy and is NOT being done in light of some
legal action that is pending.
You MUST have a retention policy.
If your retention policy says you keep it for X days, then scratch the tapes,
for each database
=
Can you list the rman_bkp_status.sh script?
Thanks!
David McMullin
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
Regarding the error you are getting - you might want to open a trouble ticket
with Symantec.
For the error:
Caught CORBA SystemException from SSmgrOperations: system exception, ID
'IDL:omg.org/CORBA/NO_PERMISSION:1.0'
OMG minor code (0), described as '*unknown description*', completed =
Do you have multiple NIC on the systems?
We had an issue where the default outgoing path was not what NetBackup expected
and had issues.
We modified the /etc/host.conf to include these lines:
multi on
mdns off
That did the trick...
David McMullin
When running Storage Life Cycle Policies, jobs queue up for duplication.
When the duplication jobs run, it first allocates a tape drive, then searches
for a tape. If either the source or destination tape it needs is already in
use, the duplication job sits there queued, waiting for the tape.
\
[EMAIL PROTECTED] [EMAIL PROTECTED]
fi
fi
David McMullin
-Original Message-
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED]
Sent: Wednesday, November 26, 2008 4:48 AM
To: veritas-bu@mailman.eng.auburn.edu
Subject: Veritas-bu Digest, Vol 31
I have SLP active and working great on HP-UX master/media servers at 6.5.2A.
The issue I am seeking assistance with is twofold:
Understanding the SLP recycle process and how to manage it.
If I issue this command:
'bpimagelist -L -idonly -stl_incomplete'
It lists jobs not completed yet. (default
57 matches
Mail list logo
