If your clients are Windows based, I would use psexec to remotely push an
msiexec command. Msiexec includes switches to suppress the reboot if required.
I don't have the software in front of me, but I now some versions included a
silentpatch.cmd file you could use for this purpose.
-Jonathan
All,
We are currently running 7.5 on Windows and looking to add deduplication
appliances as a backup target. Ideally, we would purchase the DPOO (OST/AIR)
license to keep track of the images as they write locally and then replicate to
DR, but that may not be a possibility based on budgetary
The barcode and mediaID are entirely separate entities. Even though your
barcodes are different, NetBackup automatically generates the same
mediaID for both media (GJW998L4 and LXW998L4 both look like W998L4 to
NetBackup). I suggest you create a mediaID generation rule for 8
character barcodes
Use bplist. The syntax is a pain, but it works.
Here's a windows syntax example:
bplist -A -C server -PI -I -s 01/01/2000 00:00:00 -e 04/28/2009 00:00:00
*53609A01_QPX5F_MJJ050005FA*
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
We did this with SDLT200 to LTO3. Connect both libraries to the same
media server, then bpduplicate the images. When you are done, promote
the LTO5 media to the primary copy and you are done.
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
We ran into numerous issues with NDMP and volumes with millions of
files. Have you considered SnapMirror to tape for the volumes with the
millions of files? You lose the capability to restore individual files
and incremental backups, but backup/restore speed is ridiculously fast
and rock solid. We
Which switch are you using with bpdbjobs? When I run bpdbjobs with the
-all_columns switch on my Windows 2008 R2 Master running 7.1 I get UNIX
time. Just a shot in the dark here, but have you checked your
localization settings in Windows? (Control Panel, Clock Language,
Region)?
-Jonathan
Just a shot in the dark here, but we have similar issues with Oracle
RMAN in DR. When RMAN queries NetBackup for images, it sends the client
name, as configured in NetBackup. If you are running backups from
SERVERA and then trying to do a restore to SERVERB, you need to change
SERVERB's client
You will need to license one or both of your NetApp controllers
depending on the model, which corresponds to tier of NetBackup
license. I believe our 2040s require a single tier-2 license, but our
3160s(?) require two tier-3 licenses. If I may, I suggest that you look
into snapmirror to tape and
We have 4 x HP LTO3 drives in two masters attached to old Quantum PX502
libraries and we use different drivers for both. On Windows 2003 we use
HP driver 1.0.4.0 and in Windows 2008 we use Microsoft driver
6.0.6001.18000
Here is a screenshot (Windows 2008) of the two drives:
I have a similar situation here and I tested flashbackup for Linux last
in 2010. I don't mean to dissuade you, but in my test a volume that was
running 30MB/sec with the standard client actually ran 28MB/sec with
flashbackup. If you get any better results I would love to hear about
it. I tested on
Just to clarify, this is our second Windows 2003 x86 to Windows 2008 x64
NBU 6.5.6 upgrade via cold catalog backup. We preformed our first last
December in the same manner with much success. There is a possibility
that the cold backup was bad, and we'd like to test a hot catalog backup
as well.
Try findstr /v on Windows for the second example. I’m not sure what “^$”
evaluates to, but this should help.
http://technet.microsoft.com/en-us/library/bb490907.aspx
-Jonathan
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On
Only speaking for myself here, but we're looking at vSphere 5 for VMs
with 8 vCPUS (cores.)
I believe someone from Symantec posted here a few weeks ago that vSphere
5 was supported with the same functionality as vSphere 4 but that new
functionality required 7.1.0.3. Do you know what this new
Just for clarification, what new features does vSphere 5 have in the
data protection realm? vSphere 4 introduced changed block tracking (CBT)
which revolutionized incremental backup. What does 5 have to offer? Not
that I'm bashing Veeam, I think they make a fine product. I'm just
wondering what
I think the enable changed block tracking may refer to upgrading your
VMs from virtual hardware version 4 to 7 to take advantage of CBT on
vSphere 4. This may be a concern for environments migrating from 3.x.
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
UEFI does create a small boot partition that NetBackup has issues
backing up. There was some discussion here some time ago about those
limitations. Run chkdsk d: on the citrix host and see what you get. If
the file system type is NTFS then you two are on different pages.
-Jonathan
-Original
What is the underlying disk structure (Raid, # spindles, type of disk) of the
lun that holds the Exchange databases? What is the disk latency before, during,
and after the backup? If your latency is high before the backup starts, then
it’s going to be terrible during the backup and lead to poor
at the same time. I limit the
jobs because e:, f: and g: may be on the same raid group, in which case they
share the same disk performance.
-Jonathan
From: WEAVER, Simon (external) [mailto:simon.wea...@astrium.eads.net]
Sent: Tuesday, October 11, 2011 3:36 AM
To: Martin, Jonathan; veritas-bu
You might want to call your local Suits-R-Us and buy a few hours with
an Oracle DBA. I can give you our script which is based on the goodies
script, but you really need a DBA. The database needs to be in archive
log mode, and you have to configure RMAN. We use a dedicated RMAN
database here, but I
We are currently 493 clients with a single NBU admin, and I estimate
that's about 30% of my time. That's with a liberal amount of scripting /
automation taking care of all the day-to-day tasks and a standard
environment across the board. That percentage will increase during
different periods,
I was working on our 7.1 upgrade project and was looking for a fision-io card
or other SSD solution for my catalog. The flat file aspect of the catalog is an
I/O hog, and I'd prefer an SSD solution to 8 or 10 spindles in my master. I
tend to agree with those that say it's a bad idea to put your
I generally exclude the SQL Databases and Logs. Those directories move
depending on your configuration and SQL Version.
I think the defaults for SQL 2008 are:
?:\Program Files\Microsoft SQL Server\MSSQL10.MSSQLSERVER\MSSQL\DATA
?:\Program Files\Microsoft SQL Server\MSSQL10.MSSQLSERVER\MSSQL\LOGS
If you mean the VM host (the physical server) then my experience is that
you have to use the fqdn or restores will not work. If you mean the VM
Guests (virtual machines hosted on the physical server) then I have
tested both display name and dns name, and both work. However, we have
decided to
What version of Oracle on what OS?
-Jonathan
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of WEAVER, Simon
(external)
Sent: Tuesday, July 19, 2011 9:16 AM
To: VERITAS-BU@mailman.eng.auburn.edu
Subject: [Veritas-bu] Oracle
Yes, select custom install and you can manually change the location from
C:\Program Files\Veritas to D:\Program Files\Veritas. Also be aware that I
recently installed to the C: (accidentally), removed the install via Add/Remove
Programs, Reinstalled to D:, and NetBackup's registry entries were
As far as I am aware, you cannot restore NDMP backups to non-NDMP
devices and writing NDMP backups to different NDMP devices is hit or
miss. I've run NDMP backups on several different NAS devices, NetApps
(various versions on ONTAP), Sun Unified Storage Platform, StorageTek,
OpenFiler and the best
We're looking at Microsoft's Data Protection manager here for this
purpose. Testing isn't complete, so no verdict is in. But the price of
online solutions is such that doing this in-house is the only reasonable
alternative.
-Jonathan
___
Veritas-bu
Shot in the dark here. Have you checked WINS? We have a *very* annoying
issue here where old WINS records take precedence over DNS via the WINS
lookup functionality in DNS. We normally have to tombstone the record
in WINS and clear the DNS cache to resolve.
-Jonathan
-Original Message-
Just a quick note here on business justification. It is a pretty easy
sell here that we need the ability to recover a fully functional VM in
less than an hour. We can't quite get this done with a traditional
backup and restore methodology, but we can do it with the Enterprise
Client. The other
As a brief follow-up just in case some other poor sod runs into this same
issue, SnapMirror to Tape was updated in ONTAP 8 and no longer accepts a
trailing slash. So in summary, /vol/volume/ works in ONTAP 7, but ONTAP 8
requires /vol/volume (no trailing slash).
-Jonathan
From: Martin
From: Scott Jacobson [mailto:sjaco...@novell.com]
Sent: Wednesday, June 01, 2011 11:24 PM
To: Scott Chapman; veritas-bu-boun...@mailman.eng.auburn.edu;
rusty.ma...@sungard.com
Cc: Simon (external)' 'WEAVER; Martin, Jonathan;
'VERITAS-BU@mailman.eng.auburn.edu'
Subject: Re: [Veritas-bu] Best way
If a standard client runs $500 and an enterprise client runs $2000, then
as long as you back up more than 4 virtual machines you are saving a
heap of money. I'm not sure why you need the SAN Media Server, I use a
standard media server that is already licensed and did not investigate
using a SAN
Thanks for the info. My sales team is coming in tomorrow, and apparently
we need to have a little chat. =P
-Jonathan
-Original Message-
From: Chapman, Scott [mailto:scott.chap...@icbc.com]
Sent: Wednesday, May 25, 2011 12:04 PM
To: Martin, Jonathan; 'WEAVER, Simon (external)';
'VERITAS
Just wanted to follow-up on this older thread I started a few weeks back
and say that nbd for VMware (non-shared storage) is working as intended
at impressive speeds. I've got approximately 50GB of Windows test VMs
and they are consistently pulling 60MB/sec. Granted this is an idle test
system and
I’m trying to run an SMTAPE backup of a NetApp FAS3160 to a Windows 2003 NBU
6.5.6 Media. I previously had this configuration running on a FAS2040 that
worked well, but the FAS3160 errors out with the following error:
5/12/2011 9:21:57 AM - Error ndmpagent(pid=1328) NDMP_LOG_ERROR 0 couldn't
In the windows gui (not sure about the java gui) select the 5 you do not
want to run and cut them out. Then re-run the policy and schedule.
Paste the selections back after the job starts.
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
a) No.
b) No.
c) No.
d) No.
You might be able to get file level backups working with 5, but it's not
supported and you would be running the 32bit client on a 64bit OS.
System State backups will definitely not function. They don't even
function in my 6.5.6 environments, much less 5.
-Jonathan
I was just reading through the NetBackup for VMware 7.1 Administrator's
Guide and it looks to me like there are no options to implement this
policy type other than at locations that have shared VM storage (by
presenting the luns to the master.) This seems very strange to me, given
that tons of
the NBD transfer type or NBDSSL transfer type.
-Jonathan
From: George Winter [mailto:george_win...@symantec.com]
Sent: Monday, May 02, 2011 1:14 PM
To: Martin, Jonathan; nbu
Subject: RE: [Veritas-bu] Netbackup for vSphere 4
Hi Jonathan.
You can implement the NetBackup for VMware solution
Just a shot in the dark, but have you enabled image compression? We have
servers with millions of tiny files that produce huge images files
(5GB+) and compressing those into a .tgz might chew up a fair amount of
space before the original image gets deleted.
-Jonathan
From:
Have you tried recovering an entire operating system with the 5.1 client
and 7.0 master? I'm currently planning our upgrade to 7.0, and most of
my 5.1 Windows 2000 I will convert to VMware ESX backups, but I've got a
few Windows 2000 non-vm, HP-UX, and Sun OS machines with 5.1 clients
that I will
Same issue here, 6.5.6. Master, 6.5.6 Windows 2008 x64 Client.
For example C:\Windows\Regedit.exe is missing in the OS backup of C:\
Also unavailable from the command line: bplist.exe -l C:\Windows\*.exe, but
some other executables are. (C:\Windows\FramePkg.exe for example.)
I can't speak to your specific linux kernel and I did not test with 6.5,
but I specifically keep an old 5.1 master around to pickup my legacy 3.x
and 4.5 NetBackup clients. In my testing, compatibility with 6.0 MP4 was
mixed. Some older clients would work, others failed completely. The
whole thing
Message-
From: Justin Piszcz [mailto:jpis...@lucidpixels.com]
Sent: Wednesday, March 02, 2011 10:29 AM
To: Martin, Jonathan
Cc: pranav batra; Veritas
Subject: Re: [Veritas-bu] Compatibility
Hi Jon,
Wow, legacy 3.x clients!!
How long have you been using NBU?
Justin.
On Wed, 2 Mar 2011, Martin
I'm not a fan of the per-TB licenses either, however my understanding of
the NetBackup Deduplication license is that you pay per TB up front
instead of on the back end, which means you don't have to pay for the DR
copy (or other copies.) I haven't priced this out yet, but it will be
interesting
Are you rebooting into last known good configuration? Windows keeps
two copies of the registry around just in case you reboot and the system
won't come back.
-Jonathan
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Saran
Brar
Archive_Logs
1
Setting it out past 7 days does show the Full is scheduled to run
(which
is once a week), but should the Incrementals which are scheduled for
after each session show up?
On 01/17/2011 03:07 PM, Martin, Jonathan wrote:
Have you tried nbpemreq
Have you tried nbpemreq -updatepolicies?
What happens if you run nbpemreq -predict -date mm/dd/ HH:MM:SS
using the time of the next full or incr to run?
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On
Hot catalog backups require their own volume pool. The drfile is just
what you suggested, the image / pointer to the catalog backup.
-Jonathan
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Jim
Horalek
Sent: Friday, January
as catalog backup pools.
-Jonathan
From: Jim Horalek [mailto:j...@federaledge.com]
Sent: Friday, January 14, 2011 11:11 AM
To: Martin, Jonathan; 'VERITAS-BU@MAILMAN.ENG.AUBURN.EDU'
Subject: RE: [Veritas-bu] Diaster Recovery
Thanks j.onathan,
But should I have a separate
Is this during the phase I or phase II import? Have you even tried phase II
imports? You have to complete phase I imports of media, then phase II imports
of the images on those media before you can restore anything.
Up your logging to verbose on the master and check the bptm and bpdbm logs. I
1) Copy ...\NetBackup\db\images\client from master to dr-master. (Alternately
restore catalog or import media)
2) Put media in DR library and inventory. (If using copy method above, move
media into non-scratch volume pool.)
3) Change bp.conf on dr client to match source client name (if restore
Netbackup can store images from any policy type in the same volume pool, and
even on the same physical media. You don't need to divert your Windows and Unix
backups to different volume pools unless that is a business requirement. I use
volume pools in my environment to differentiate media
I've been working on an issue now for several weeks and it's got support
stumped (for the time being.)
Master: NBU 6.5.6 on Windows 2003 SP2
Clients: NBU 6.5.6. on Linux
Backups were working fine for months, then we started getting the
occasional error 233. Then one weekend we started
Is anyone else on the list testing Sun / Oracle's Unified Storage
Systems' ZFS over NDMP backup methodology? I've been struggling to
backup millions of tiny files with NFS mounts off a Sun 7210 device, but
after recently upgrading the firmware we've been testing ZFS over NDMP
with great success.
Just FYI, if you run dheapmon.exe from the task scheduler you can see
the output of all heaps without having to use the console.
-Jonathan
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Michael
Graff Andersen
Sent:
I'm trying to break up the backups of a certain large folder via
selection list and I want to make sure I get everything.
Client: Solaris 10 x86 running NBU 6.5.6
Master/Media: Windows 2003 running NBU 6.5.6
So far I've got:
/home/[a-nA-N]*
/home/[o-zO-Z]*
/home/[0-9]*
What
Directly off a recent Symantec quote:
SYMC NETBACKUP OPTION SHARED STORAGE OPTION 7.0 XPLAT 1 DRIVE STD LIC
This is a separate line item. Can't help you with the SAN Media server,
we don't use them here. Tier 1 is single CPU / Socket, Tier 2 is dual
CPU / Socket.
-Jonathan
-Original
We use the MS-SQL Client here. We don't really have dedicated SQL DBAs,
we have windows admins with sql training and Oracle DBAs who would
rather manage Oracle. The text files offer simple configuration, and
restores are a breeze with the GUI. I guess for the limited amount of
MS-SQL we do here it
We do this on one server here that often needs to write unique file
names to tape, then recover in batches. The syntax is whacky, and I
think if the filelist.txt file doesn't end with a blank line the entire
thing fails.
bprestore.exe -A -L C:\Path\RestoreLog.txt -R /C/Path/rename.txt -f
What are you high and low water mark settings? How much disk space is
available to the DSSU, how much data is on it, and how big were any jobs
trying to write to the DSSU at the time?
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
Are you assuming that NetBackup uses the high water marks for the entire
storage group? How is this group configured (round Robin, Least used
etc...) If you have a 2TB DSSU that is 95% full, and a 920GB job is
assigned to that DSSU (instead of another DSSU) then NetBackup is going
to clear enough
uEFI is a new standard for booting systems that replaces the traditional
BIOS boot. Our Dell servers toggle back and forth between BIOS and uEFI,
but once you've installed an OS in one mode or the other, you have to
wipe the disk clean to switch to the other boot methodology. I don't
know why
Does anyone running NetBackup DeDupe (Media server, not dedicated puredisk)
have any storage recommendations? We use 500GB SATA disks here as the basis for
our DSSU / D2D2T infrastructure and I need to know if we can convert them to
DeDupe storage as is, or if we need to invest in something
We break our large datasets into multiple streams and restart that way.
It's much easier for me to say to an app group all of the backups
finished except /lun7 and I'm rerunning that now than The backup is
80% complete and we're rerunning it from wherever it failed. For
backups that run 48 hours,
You cannot expire individual images on a media and re-use that space.
Once a certain image has been written, NetBackup will not overwrite that
position on the media until the entire tape has expired.
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
Launch the MS-SQL Wizard from the server and look for backups in the
last 24 hours (or whatever time frame you are interested in.) Each
database with images will be listed, as well as whether the backups are
full or transaction logs. You can also get this information from the
activity monitor. The
Incrementals will not schedule if a full from the same policy is still running.
You will have to wait, kill the last full stream, or move the incremental to
another policy to have it auto run.
-Jonathan
From: veritas-bu-boun...@mailman.eng.auburn.edu
I've tested NDMP on 6 differetnt arrays and it has never moved millions
of small files well. We maxed out backup performance on our NetApp FAS
2xxx with 2 streams at approx 20MB/sec total. We're hoping to test
SMTape, which purportedly does a bit level dump of the entire array. I
haven't had a
I've just recently had a meeting with my Symantec rep, and we were led to
believe that license cost and replication were NetBackup advantages. For
example, I was under the impression that if you backed up 10TB of data weekly,
but kept 100 deduped copies on 40TB of storage, that you only needed
Do the backups run if you kick off the missing hosts manually?
This might be a corrupt policy. Have you tried recreating the policy?
-Jonathan
___
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
Shot in the dark. Is this Windows 2008? Did you remember to
shift-right-click on the setup.exe and run as administrator? I seem to
remember getting some weird Windows Installer errors when not running
as administrator.
-Jonathan
From: veritas-bu-boun...@mailman.eng.auburn.edu
When you put tapes into the library without a barcode, run vmphyinv to
automatically mount each tape and inventory based on its media header.
That will fix your robot inventory issue, and make those media available
for restore. You might want to remove the media with barcodes from the
library
If memory serves this was a new feature in 6.5 (Maybe 6.0). 5.1 does not
behave in this manner.
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Nate
Sanders
Sent: Thursday, August 05, 2010 10:58
Do you have a full catalog backup from before the image expired? You
could restore it to a temporary server, then run bpimagelist -media -L
-backupid $image to find the media. That's a lot of work, but it beats
scanning 100 media to find 1.
-Jonathan
-Original Message-
From:
How about dropping the No.Restrictions touch file into db\altnames\ for
a quick fix?
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of Iverson,
Jerald
Sent: Friday, July 30, 2010 6:28 PM
To:
Here's a start.
E) 2 - MPX is disabled by default so only one stream will write to tape.
But, the job has a parent job which is also active, so 2.
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of
We get poor performance (4MB/sec) performance running NFS mounts on our
FAS 2040, but we've found we can run many simultaneous streams and get
that into the 20+MB/sec range. We've got one Sun device we run 16
simultaneous NFS streams on that pushes 30MB/sec. Have you tried hitting
multiple mounts
: Re: [Veritas-bu] NDMP backups from Netapp using Netbackup
On Tue, Jul 27, 2010 at 3:29 PM, Martin, Jonathan
jmart...@intersil.com wrote:
We get poor performance (4MB/sec) performance running NFS mounts on our
FAS 2040, but we've found we can run many simultaneous streams and get
I have a test 6.5.6. master running on CentOS. Simmilar to Jeff's
scenario, I had to drop a real redhat version file into
/usr/openv/netbackup before some of the patches would install, but it
runs now without issue. This is not production, but it does work just
fine.
As far as CentOS in the
I use the following command every 30 minutes on several Windows 5.1, 6.0
and 6.5 Masters without issue. If the robot happens to be offline the
task fails without issue. I've never had an issue with inventories and
normal operations.
vmupdate -rt tld -rn 0 -use_barcode_rules
-Jonathan
, I do it to
save space on the DR end.
-Jonathan
From: WEAVER, Simon (external) [mailto:simon.wea...@astrium.eads.net]
Sent: Thursday, July 08, 2010 11:12 AM
To: Martin, Jonathan; pranav batra; Veritas
Subject: RE: [Veritas-bu] Catalog Copy ????/
Hi Jonathan
I thought Symantec did
I tried to find a tool that retained compression on a per-file basis to
perform a catalog migration but I could not find one. I considered
custom writing my own in .net but never got around to it.
Why not simply compress the entire DR volume? You'll take a slight
performance hit come DR
Are you hitting multiple VMs simultaneously on the same datastore? Is
the speed better when you only run one backup at a time? We've
identified serious performance issues related to requesting to many
random I/Os from a raid group as the same time. Backing up multiple VMs
and rebooting multiple
Just a shot in the dark here, but is it possible the old drives are
hidden because of the path changes? You can check using this support
article. http://support.microsoft.com/kb/315539 Also, have you run the
TUR fix for your fiber tape drives?
http://support.microsoft.com/kb/842411
-Jonathan
What do you mean by DSSU flush? Images that have not been copied to
tape will not purge unless you explicitly expire them. You can use
bpimagelist -L -backupid $image against each image on the DSSU to see
the number of copies. There is a known issue with 5.1 where partial
images are left behind,
Run an ls on the directory where the images are. You should see
something like:
|Image Identifier|C#|F#|Backup Time|.img
Server_1234567890_C1_F1_1234567890.img
Server_1234567890_C1_F2_1234567890.img
Server_1234567890_C1_F3_1234567890.img
Server_1234567890_C1_F4_1234567890.img
% of the disk space was orphaned files. Thank you kind
sir!
On 07/01/2010 12:45 PM, Martin, Jonathan wrote:
Run an ls on the directory where the images are. You should see
something like:
|Image Identifier|C#|F#|Backup Time|.img
Server_1234567890_C1_F1_1234567890.img
-Original Message-
From: Nate Sanders [mailto:sande...@dmotorworks.com]
Sent: Thursday, July 01, 2010 5:48 PM
To: VERITAS-BU@MAILMAN.ENG.AUBURN.EDU
Cc: Martin, Jonathan
Subject: Re: [Veritas-bu] Disk Staging Flush
Has the job just gotten so large that it cannot write to disk staging
(700GB
-Original Message-
From: WALLEBROEK Bart [mailto:bart.wallebr...@swift.com]
Sent: Monday, June 28, 2010 4:58 AM
To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu
Subject: RE: [Veritas-bu] Destaging going slow
Martin,
To be very honest, I have no idea what the detailed specs are behind
First of all, suck it Neil Conner. I'm about to disagree with Ed
(again) and there is nothing you and your fish eating friends can do
about it.
Gotten away with it? I'm not stealing bread from the supermarket, I've
made a calculated decision. I run 18 MD1000s in this configuration
globally
that are getting the same speeds when comparing
destaging speed to a back up taken directly to tape from the disk used
for staging ? Or do you just have to live with the performance hit when
destaging ?
Martin, Jonathan wrote:
I'm with Mark Phillips on this one. We use the direct attached
and weaknesses
of any proposed storage solution before purchasing and implementing?
-Jonathan
From: Lightner, Jeff [mailto:jlight...@water.com]
Sent: Tuesday, June 29, 2010 3:54 PM
To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu
Subject: RE: [Veritas-bu] Destaging going slow
Gotten
...@swift.com]
Sent: Thursday, June 24, 2010 11:19 PM
To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu
Subject: RE: [Veritas-bu] Destaging going slow
Martin,
Even when we fully format the DSSU disk and we then run 1 backup job to this
disk and directly afterwards we duplicate it to tape we get
. In the
future, I'll consider your internet sensibilities before telling you to go to
hell.
-Jonathan
-Original Message-
From: Conner, Neil [mailto:n...@mbari.org]
Sent: Friday, June 25, 2010 12:02 PM
To: Martin, Jonathan; WALLEBROEK Bart; Veritas List
Subject: Re: [Veritas-bu] Destaging
...@swift.com]
Sent: Thursday, June 24, 2010 2:46 AM
To: Martin, Jonathan; veritas-bu@mailman.eng.auburn.edu
Subject: RE: [Veritas-bu] Destaging going slow
Jonathan,
Actually we already tried out different sets of block sizes on the DSSU. 32
seems to be the fastest we can find. We do not use compression
Format your DSSUs with a 64k block size.
Use 64K block size on raid group.
Disable compression (if you can't use 64k block size.)
Lower the number of concurrent writes to the DSSU.
Occasionally clear the DSSU to relieve fragmentation.
Clear the DSSU, write a single backup to the disk, then stream
Have you checked the system log for NTFS corruption? I would run a
chkdsk /f S: just to be sure. Careful though, it will knock the drive
offline while the chkdsk is running.
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
What do your client list and backup selections look like? Why did you
select cross mount points, are you using NTFS mount points?
-Jonathan
-Original Message-
From: veritas-bu-boun...@mailman.eng.auburn.edu
[mailto:veritas-bu-boun...@mailman.eng.auburn.edu] On Behalf Of zimmy00
Sent:
1 - 100 of 691 matches
Mail list logo