|
Re: [Networker] Networker 7.3.3 Restores Slow
2007-08-22 07:57:22
OK Thanks. To fill in some blanks.
The Networker Server has the SAN (FC) storage for the restore and the FC
attached Scalar i2000 (LTO3) tape library.
The Client for the backup/restore is a RedHat cluster member with the
Networker server.
I originally wanted to do a Browse the CFI for the restore, but when I
selected "add *" it would only process so much before it croaked because of
"out of memory" error.
So I started a directed recover using the SSID. After that was going slow I
then switched to browse the CFI again, but only selecting directories that
wouldn't use the same tape(s) as the SSID recover. This didn't seem to be as
slow as the SSID, but still slow.
I used to put the backups of this file system in our general FSPool, but the
last few fulls I made a Pool exclusively for this file system
(JTCArchivePool) so I wouldn't have to skip over data that I wasn't
interested it.
Using the following script:
#!/bin/csh
mminfo -q "client=$1,name=$2,"!"incomplete,"!"ssrecycle" -ot -r
volume,level,ssid,savetime,ssbrowse,ssexp,volretent,sumsize
I show for the full from a couple of weeks ago:
000041 full 4223104247 08/06/07 09/06/07 11/06/07 10/12/07 413 GB
000101 full 4223104247 08/06/07 09/06/07 11/06/07 11/06/07 416 GB
000149 full 4223104247 08/06/07 09/06/07 11/06/07 11/06/07 414 GB
000200 full 4223104247 08/06/07 09/06/07 11/06/07 11/06/07 82 GB
000221 full 4223104247 08/06/07 09/06/07 11/06/07 11/20/07 157 GB
000299 full 4223104247 08/06/07 09/06/07 11/06/07 11/06/07 416 GB
Then followed by incrementals every day since.
Looking at the logs, it only took alittle more than 25 hours to do this full
backup, but using Networker (started a week ago, and just now finishing it
up) but using LAN (cp -a source-dir target-dir/.).
Thanks again.
Randy
----- Original Message -----
From: "Faidherbe, Thierry" <Thierry.Faidherbe AT HP DOT COM>
To: <NETWORKER AT LISTSERV.TEMPLE DOT EDU>
Sent: Wednesday, August 22, 2007 5:51 AM
Subject: Re: [Networker] Networker 7.3.3 Restores Slow
A core info is missing : how did you do the recover ?
using ssid recover or browsing client file index ?
Another important think to take care of is tape multiplexing
during backup : How more multiplexed is the tape block, how
more tape blocks have to be processed for extracting data.
In other words, Backup Rule #1 : fast backup, slow recover.
You also have a look at which device mounted the volume for
recovery. Maybe did you recover across the lan ....
Th
Kind regards - Bien cordialement - Vriendelijke groeten,
Thierry FAIDHERBE
HP Services - Storage Division
Tru64 Unix and Legato Enterprise Backup Solutions Consultant
********* ********* HEWLETT - PACKARD SPRL/BVBA
******* h ******* Hermeslaan, 1a
****** h ****** B-1831 Diegem
***** hhhh pppp ***** BELGIUM
***** h h p p *****
***** h h pppp ***** Phone : +32 (0)2 / 729.85.42
****** p ****** Mobile : +32 (0)498/ 94.60.85
******* p ******* Fax : +32 (0)2 / 729.88.30
********* ********* Email/MSN : thierry.faidherbe(at)hp.com
Internet : http://www.hp.com/
I N V E N T
Enterprise number 0402.220.594 RPM/RPR Brussels
-----Original Message-----
From: EMC NetWorker discussion [mailto:NETWORKER AT LISTSERV.TEMPLE DOT EDU] On
Behalf Of Randy Doering
Sent: mercredi 22 août 2007 2:09
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Subject: [Networker] Networker 7.3.3 Restores Slow
Greetings, In the past I've used Networker backups to restore to a new
file
system for migration reasons. Helps to stay off of the current one plus
tested the backup (and restore) capabilities. I've done this for the past
couple of years and was pleased with the performance of the restore
operations (both directed recover and saveset recover).
About 2 weeks ago, I made a fresh full backup of a 2TB file system (after
splitting a 4.7TB into 2 chunks).
Last week I started the restore to some new space of this 2TB and it has
been taking "forever" to get the data off of the tapes.
When I did the backup, I was at 7.3.2 Junbo (399). A couple of weeks ago,
I
upgrade to 7.3.3.
Side Note: Last year when I started these type of migrations and tried the
Linux "cp -a source-dir target-dir/." I would get the "max arguement"
error
message which is why I went to the NW restore option. Now the "cp -a" is
working on these directories with 100K - 150K small files in them.
So as anyone done any massive restores with 7.3.3 and noticed any dramatic
performance hits?
Thanks,
Randy
To sign off this list, send email to listserv AT listserv.temple DOT edu and type
"signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
To sign off this list, send email to listserv AT listserv.temple DOT edu and type
"signoff networker" in the body of the email. Please write to
networker-request AT listserv.temple DOT edu if you have any problems with this
list. You can access the archives at
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
To sign off this list, send email to listserv AT listserv.temple DOT edu and type
"signoff networker" in the body of the email. Please write to networker-request
AT listserv.temple DOT edu if you have any problems with this list. You can access the
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER
|
|
|
