Veritas-bu

[Veritas-bu] Veritas Netbackup sg problem

2000-10-19 00:52:37
Subject: [Veritas-bu] Veritas Netbackup sg problem
From: Micheal.Choi AT chase DOT com Micheal.Choi AT chase DOT com
Date: Thu, 19 Oct 2000 12:52:37 +0800
hi all,
     I am currently using a STK L180 with JNI's fibre channel card on Solaris
2.6.
We planned for a Share Storage Option for Netbackup and have already implemented
the Master Server. It works well.

     On implementation of the Slave server which uses the same set of tape
drivers
with the master server with the same hardware configuration, problems happened
on
/dev/sg attachment of the Fibre Channel tape drives when the driver is loaded:

sg56 attach: - soft state zalloc failed
sg57 attach: - soft state zalloc failed
sg58 attach: - soft state zalloc failed
sg59 attach: - soft state zalloc failed
sg60 attach: - soft state zalloc failed
sg61 attach: - soft state zalloc failed

and the sg driver of the fibre channel tapes drivers are NOT attached
and therefore Netbackup cannot function properly.

The scsi tape driver /dev/st, however, is attached and functioned properly.

The main differences between the master and the slave servers is that:
1)   The slave server is currently using Networker, with the lus driver loaded.
We will
     decommission it after the Netbackup is functional.
2)   The slave server is also attached to a Sun A1000, in which both lus and sg
     created huge amount of devices to it.
3)   The slave servers is also attached to another small DLT library with 2 SCSI
DLTs
     These two drives, however, function without any problem, with the sg
drivers
     attached normally.

When we ask Veritas, they blamed the problem on the firmware of JNI card (2.4.1)
and
request us to downgrade it to 2.3. We only read from the documents that it is
related to
Sun E10K. We also consulted EMC and EMC, however, mentioned they do not have any
experience with that firmware level at all.

We have another to-be slave sever which has more or less the same configuration
with the
previous slave server. When we implement netbackup slave to that the problem is
repeated.

Since the slave servers are very important to our production environment, we
cannot afford
any change to it unless we have a certain amount of certaintythat  it would fix
the problem.
I wondered if anyone has encountered such problems before and can share with us?

thx in advance,
Michael Choi
Chase Manhattan Bank Hong Kong






<Prev in Thread] Current Thread [Next in Thread>