Backup Issue Separate Tape Drives

[Rahim Johnson]

We are having issues with our backups and are trying to determine if this is a network issue or something else.

The problem has occurred twice in the last week around the same time.

We have two MSL 6000 tape drives. We are backing up our tru64 box to one and our HP-UX servers to another. We are getting the following errors when the problem occurs.

*servernames and filesystem mountpoints have been renamed

The is the message from the Tru64backup:
[Warning] From: BMA@xxxxx "HP:Ultrium 2-SCSI_1" Time: 11/15/2008 12:21:40 AM
The device "HP:Ultrium 2-SCSI_1" could not be opened("Device could not be accessed")

[Major] From: BMA@xxxxx"HP:Ultrium 2-SCSI_1" Time: 11/15/2008 12:27:32 AM
[90:54] /dev/ntape/tape2
Cannot open device (Operation timeout.)

Here is the warning from our other backup.

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:50:49
[90:51] /dev/rmt/3mn
Cannot write to device ([5] I/O error)

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:52:08
[90:159] Apparently SCSI Unit Attention.

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:52:08
[90:162] Skipping write of filemark. (Apparently SCSI Unit Attention).

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:52:08
[90:162] Skipping write of filemark. (Apparently SCSI Unit Attention).

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:52:08
[90:135] Cannot eject medium. (Trailing filemark(s) could not be written.)

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:52:08
[90:64] By: UMA@server.domain@/dev/picker
Cannot unload exchanger medium (Details unknown.)

Report Messages Other Than Normal [server.domain]:

[Critical] From: VBDA@server.domain "/path/path" Time: 11/15/08 00:50:49
Received ABORT request from SM => aborting.

[Critical] From: VBDA@server.domain "/path" Time: 11/15/08 00:50:49
Received ABORT request from SM => aborting.

[Critical] From: VBDA@server.domain "/path/path" Time: 11/15/08 00:50:49
Connection to Media Agent broken => aborting.

[Critical] From: VBDA@server.domain "/path" Time: 11/15/08 00:50:49
Connection to Media Agent broken => aborting.

Report Messages Other Than Normal [server.domain]:

[Critical] From: VBDA@server.domain "/path" Time: 11/15/08 00:50:49
Received ABORT request from SM => aborting.

[Critical] From: VBDA@server.domain "/path" Time: 11/15/08 00:50:49
Connection to Media Agent broken => aborting.

Report Messages Other Than Normal [servername.domain]:

[Critical] From: VBDA@servername.domain "/path" Time: 11/15/08 00:50:49
Received ABORT request from SM => aborting.

[Critical] From: VBDA@servername.domain "/path" Time: 11/15/08 00:50:49
Received ABORT request from SM => aborting.

[Critical] From: VBDA@servername.domain "/path" Time: 11/15/08 00:50:49
Received ABORT request from SM => aborting.

[Critical] From: VBDA@servername.domain "/path" Time: 11/15/08 00:50:49
Connection to Media Agent broken => aborting.

[Critical] From: VBDA@servername.domain "/path" Time: 11/15/08 00:50:49
Connection to Media Agent broken => aborting.

[Critical] From: VBDA@servername.domain "/path" Time: 11/15/08 00:50:49
Connection to Media Agent broken => aborting

We are connected to a SAN via FC

Thanks in Advance

[Johnson Punniyalingam]

insf -e

ioscan -fnC tape

check ur "FC" status,

Thanks,
Johnson

[Rahim Johnson]

I will look into what you have suggested.


Just to clarify this information, these are two separate libraries (they have two drives each)

One Library handles 1.6TB Tapes the other 800GB tapes.

The servers are on the same subnet however.

So we have two separate libraries backing up two separate servers, and they seem to have this problem at the same time. As I mentioned before, I suspect a network problem, but I want to make sure before I suggest that to our network people.

Thanks

[Rahim Johnson]

One more note, this does not happen every night.

I happened first on 11/12 and then again on 11/15 both times between 12:00am and 1:00 am

[Torsten.]

Because of the cannot write, cannot unload, cannot eject messages I would suspect the library hardware first.

Have a look at the lib's frontpanel and/or web interface first - better to use HP LTT from

http://www.hp.com/support/tapetools

and pull a support ticket.

Have a look into the hp-ux servers syslog for any related error messages too.

[Rahim Johnson]

Would this happen on two separate libraries?

One of which was just installed three months ago.

The other has been running for about two years. I am not doubting that it could happen but I certainly hope not.

I will look into that as well.

[Torsten.]

Do you get this from both libraries or only one?

[Rahim Johnson]

Both backups.

The first entry is not as detailed, (that backup is almost complete by this time)

This is Libary 1 (the older one backing up tru64)
[Warning] From: BMA@xxxxx "HP:Ultrium 2-SCSI_1" Time: 11/15/2008 12:21:40 AM
The device "HP:Ultrium 2-SCSI_1" could not be opened("Device could not be accessed")

[Major] From: BMA@xxxxx"HP:Ultrium 2-SCSI_1" Time: 11/15/2008 12:27:32 AM
[90:54] /dev/ntape/tape2
Cannot open device (Operation timeout.)

This is Libary 2 (the newer one backing up our UX servers)

Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:50:49
[90:51] /dev/rmt/3mn
Cannot write to device ([5] I/O error)

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:52:08
[90:159] Apparently SCSI Unit Attention.

[Major] From: BMA@server.domain "Drive 1" Time: 11/15/08 00:52:08
[90:162] Skipping write of filemark. (Apparently SCSI Unit Attention).


I probably need a course in how to ask for help properly, even after I read my post, it was not clear. Thanks for your patience.

[Torsten.]

I'm not familiar with tru64, sorry.

If you see the errors on hp-ux, are the tapes still CLAIMED in ioscan?

[Rahim Johnson]

All devices are still claimed.

The errors are actually coming from Dataprotector for both servers.
(another piece of info that I failed to include)