Re: [expert] A bit OT: Help needed - reaching the edge...

[Mark Weaver]

Pierre Fortin wrote:

On Wed, 01 Jan 2003 11:12:06 -0500 Mark Weaver
<[EMAIL PROTECTED]> wrote:

Pierre Fortin wrote:

On Wed, 01 Jan 2003 01:25:12 -0500 Mark Weaver
<[EMAIL PROTECTED]> wrote:

g wrote:

Mark Weaver wrote:

Thank you g! this looks like fun. I'm going to try this and see what
happens. I've been doing all the backup to scsi tape using tar

before>>

this.


you are most welcome. i hope you can get some mileage from scripts.

cpio works very well for tape archive. i started using cpio w/ basic
scripts in mid 70's for backing up 10 and 20 meg hdd's and has yet to
fail.
'testscript' is best done to insure you are getting what you want

from>>'find'.

with find and grep, cpio even does a fine batch/conditional file
transfer.


peace out.

tc,hago.

g

g,

I've got one more question. The other evening I was mucking about working some kinks out of my backup script and I think I "over-filled"

the tape. When it got full, of course, it started returning a lot of errors to the console, and now I can't get it to run at all.

When I try to run it from the command line I'm getting this returned

to >the console:

tar: /dev/st0: Cannot open: No such device or address
tar: Error is not recoverable: exiting now
tar: /dev/st0: Cannot open: No such device or address
tar: Error is not recoverable: exiting now

I got to diggin around in the log files and found this:

[root@mdw1982 root]# grep scsi /var/log/messages
Dec 29 14:57:26 mdw1982 kudzu: aliased scsi_hostadapter as aha1542
Dec 29 15:01:34 mdw1982 kernel: scsi0 : Adaptec 1542
Dec 29 15:01:36 mdw1982 kernel: Attached scsi tape st0 at scsi0,

channel>

0, id 1, lun 0
Dec 31 00:33:02 mdw1982 kernel: Sent BUS RESET to scsi host 0
Dec 31 00:33:12 mdw1982 kernel: scsi: device set offline - command

error>

recover failed: host 0 channel 0 id 1 lun 0
Dec 31 06:56:36 mdw1982 kernel: scsi0 : Adaptec 1542
Dec 31 06:56:36 mdw1982 kernel: scsi : 0 hosts left.
Dec 31 06:59:36 mdw1982 kernel: scsi0 : Adaptec 1542
Dec 31 06:59:38 mdw1982 kernel: Attached scsi tape st0 at scsi0,

channel>

0, id 1, lun 0
Dec 31 11:16:58 mdw1982 kernel: scsi: device set offline - command

error>

recover failed: host 0 channel 0 id 1 lun 0

I've tried reloading the module for the device, but to no avail. It appears from the log entries that it got disabled some how when the kernel received the errors from the hardware, and I haven't a clue as

to>

how to re-enable it. I've tried rebooting the machine, but that didn't

help either. I'm sorta stumped here. Any ideas?

thanks,

Mark

If the bus reset isn't taking, you may have to power off the device
for it to reset...  I've had some luck at just pulling the power
connector from hung scsi devices if they aren't in external boxes with
their own power switch; but its also easy to hang a Linux box via
scsi...

HTH,
Pierre

Hi Pierre,

That thought has been nagging at the back of my mind. I was waiting to see if anyone confirmed it and you have. It's a shame there isn't a way to reset them via software. I've noticed on the one Netware machine at work that when there's a hardware problem with on of the scsi interfaces

The bus reset *should* have done the job; but I don't recall the h/w
details of the reset...  there may be a flaw in the code that no one has
nailed down yet...  Back on 11/11/02, I was working with my scsi scanner
and managed to lock-up Linux tighter than a heated drumskin...  clearly an
indication of a permanent status that the code synchronously loops on
without the benefit of a watchdog timer... :^P

the only thing that helps is a powerdown and restart. Is this peculiar to scsi devices?

Mark

Nope...  remember...  various bus devices can be setup under one OS and
assumed setup when another OS is started (most machines have an option in
the BIOS to [not] reset them...  The general issue of 'reset' is fraught
with h/w-h/w and h/w-s/w interdependencies, resulting in lots of sighs...

Pierre

Understood...that leaves me with one really doable solution. That is to setup a machine who's sole function would be to do nothing but backups across the LAN. That way when there's a scsi/harware problem it's not going to affect the rest of the systems when the machine requires a restart.

Mark

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com