SCSI RESET ON SAN 1

[Masternet]

Hi: I have a SAN ambient where Networker for Linux 7.2 build 172 is installed with storage nodes.
When Legato make a backup, after a time, the tapes become Full state and NetWorker say "insert another tape to pool XXXX".
The tape is suspect status, so when I try recover data of this tape is impossible.

Scanner of tape show: "scanner: SYSTEM error: Tape label read for volume ? in pool ?, is not recognized by Networker."

I hear about SCSI Reset on tape Library Dell PowerVault 136T but am not shure if this can be a trubble for NetWorker writing operation over tapes.

thanks!

 

[605]

SCSI bust resets are responsible for such effects and you must suppress them. In general, they can happen on any hardware. One especially mentioned will most likely only point to specifc customer scenario but one again - it can happen to all hardware, in general.

 

[ag100]

Also, be sure no monitoring/reporting software is interacting with the tape devices in a disruptive manner. In some cases I've come across, some type of reporting/environmental monitoring software was probing the drive incorrectly, thus causing a rewind, which, when combined with specific circumstances (tape device is idle for a very short period of time) would allow NetWorker to start writing to the rewound tape, without verifying the label.

You can also use the "Verify Label on Unload" library option to warn you when this is happening.

good luck!

 

[605]

to ag100:

This is a good idea in general. However, i doubt that that it will work if a reset has initiated the unload process. Can you clarify this, please?

 

[ag100]

Hi - Sure, I'll be happy to clarify. I assume you're talking about the verify label on unload?? If not, please let me know.

If a SCSI reset (or anything) causes a tape drive to rewind, and following the rewind operation, NetWorker overwrites the tape label, a warning should be produced when the next label verification occurs (which, if set, should be when the next unload operation occurs).

The "Verify Label on Unload" option won't stop anything from happening, but if you setup traps, at least you'll know so you can correct the problem.

Does that make sense? If I missed something, please let me know.

Thanks,

-ag100

 

[Meatman666]

Are you running a Windows based NetWorker Server and Storage Nodes? If so, removable storage service gets in the way....disabled it.

GL


cheers,
Jim

 

[Masternet]

the os is Linux RedHat enterprise on all servers.

 

[bombcan]

Make sure your using the right device example on Solaris it is the #cbn device the "n" is for no rewind