Re: [Networker] Unload failure / ADIC / DLT8000 / FCR250 / SAN / Brocade

No

The problem is that NetWorker attempts the unload but the drive does nothing 
and the unload fails and is retried a number of times but it never happens.  I 
can make it happen by using the mt offline command (mt -f /dev/rmt/0cbn 
offline) and the tape immediately starts to unload(as indicated by the left 
lites going out and the right activity lite going to flash.

Some of the confusion is in the terminology used by Legato vs the mt command.
(unload, eject, offline, mount, unmount, etc.)

The bottom line is that whatever NetWorker does the drive fails to start to 
unload at which time I issue the offline command via mt and it immediately 
starts to unload.

So NetWorker must be doing something different (or the parts involved between 
NetWorker and the tape drive are doing something that causes the unload not to 
start).
The parts are:
        NetWorker (nsrjb command?)
        HBA driver (Hardware is LP8000, firmware is xx, config is yy)
        SAN switches
        FCR250 (Crossroads fibre-channel to scsi router - 2 FC ports from SAN, 
2 SCSI busses behind each FC port
                Config and mapping to/from FC and SCSI
        DLT8000 firmware (robotics firmware)
If I understand things correctly the robotics of the ADIC Scalar 1000 library 
are not at all involved in the process of unloading a tape from a tape-drive. 
By "unloading" here I mean simply rewinding all the tape out of the drive to go 
offline waiting for the robotics to "open the door" thereby ejecting the tape 
from the drive ready for the picker to grab it and put it into a slot.  A SCSI 
command is simply sent to the tape-drive and it starts to unload the tape from 
the tape-drive in prep for the robotics to "open the door" and grab the tape 
and put it in the approp slot.
The problem has nothing to do with the robotics at all - it simply has to do 
with the proper SCSI command not getting to the drive telling it to rewind all 
the tape out of the drive ready for the "door to be opened".

I guess I'm just going to have to capture the SCSI traffic through the 
FCR250(s) and then do the same for when I manually issue the "mt -f 
/dev/rmt/0cbn offline" command and see what the difference is...
What's the best tool to do this - besided dumping traces from the FCR's ?

thanks - John


        

-----Original Message-----
From: John Herlihy [mailto:johnh AT xsidata.com DOT au]
Sent: Wednesday, April 21, 2004 7:58 PM
To: Legato NetWorker discussion; Ballinger, John M
Subject: RE: [Networker] Unload failure / ADIC / DLT8000 / FCR250 / SAN
/ Brocade


How are you trying to offline the drives? Are you just trying to unload
through Networker, or are you trying to reset the jukebox (ie nsrjb -H)
through Networker? Tape drives do not get ejected during a reset, but
Networker will still sit there trying to move the tapes from the drive
back to the slot and you have to do your "mt ...... offline" trick to
eject the drives.

> -----Original Message-----
> From: Ballinger, John M [mailto:john.ballinger AT PNL DOT GOV] 
> Sent: Tuesday, 20 April 2004 10:16
> To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
> Subject: [Networker] Unload failure / ADIC / DLT8000 / FCR250 
> / SAN / Brocade
> 
> 
> We are experiencing a situation where NetWorker 7.0 on 
> Solaris fails to offline a device on our Scalar 1000 tape robot.
> Yet at the same time if as soon as we see this happening we 
> run an "mt -f /dev/rmt/0cbn offline" command the drive 
> immediately starts to unload as it should have earlier.
> 
> It does this with all of the drives.
> Sometimes it works but most often fails
> 
> Several complicating factors:
>         The library is on a SAN (All Brocade switches)
>         The tape library is an ADIC Scalar 1000 with 7 
> DLT8000 scsi tape drives
>         The FC to SCSI routing is done with an FCR 250 (Crossroads)
> 
> Why isn't NetWorker getting the unload command to the drive 
> (thru the NetWorker server HBA thru the SAN thru the FCR to the drive.
> 
> So you can see there are multiple places for the command to 
> get lost etc.
> 
> Anyone have any ideas ?
> 
> thanks - John
> 
>       
> 
> --
> Note: To sign off this list, send a "signoff networker" 
> command via email
> to listserv AT listmail.temple DOT edu or visit the list's Web site at
> http://listmail.temple.edu/archives/networker.html where you can
> also view and post messages to the list.
> =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
> 

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=