Performance/ Restore is Queued for long time ?

[patriklinder100]

Hi
I am using NetBackup Datacenter 4.5 MP3 on Solaris 8
When I want to do a restore on a large number of files I the job gets queued for ages...

Small jobbs ( a few files gets started directly )
but when i want to restore a large number of files ( several thousands ) the job will not start.

Does any body have a tip/suggestion on performanse setup to solve this problem.

Regards
Patrik

 

[Chinita]

My master server is HPUX 11.0, 64bit, and I use Netbackup, Data Center 4.5, MP3. I had the same problem trying to do a TIR (true image restore) of 70GB from one to another Windows 2000 client. After queuing for 5 minutes the jobs ended with error 10 and/or error 5, share memory problem. I changed the following kernel parameters in the master server and reboot it. My master has 1GB of mem.

Some of these parameter may not applied to Solaris.

I increased:
maxdsize to 603979776
maxdsize_64bit to 1GB (shouldn't be higher than your memory)
maxtsize_64bit to 1GB (shouldn't be higher than your memory)

2) maxusers=400
3) Shared Memory
shminfo_shmmin=1
shminfo_shmmni=220
shminfo_shmseg=100
shminfo_shmmax=603979776
4) System Queues should be increased to:
seminfo_semume=64
seminfo_semmap=1026
seminfo_semopm=32
seminfo_semmni=1024
seminfo_semmns=1024
seminfo_semmnu=1024
seminfo_semmsl=60
5) Messages Queues should be increased to:
msginfo_msgmap=514
msginfo_msgmax=8192
msginfo_msgmnb=65536
msginfo_msgmni=256*
msginfo_msgssz=16
msginfo_msgtql=512*
msginfo_msgseg=8192

Also, I run vmstat -a and found that nofiles was 60, I increased it to 512 (run ulimit -n 512) this is set up only for your current session, if you log out it will go back to the previous value. I did it as root. Then I opened the gui as root and started the backup. It queued for about 10 minutes before started.