Backup Central

Solaris 9/Networker7.1; NetApp running 6.5.2R1, Qualstar 8464 with 4 LTO drives.

The Solaris 9/Networker7.1 server controls the jukebox robotics and two
drives. Two drives in the jukebox are connected over SCSI to the NetApp.
Backups for the NetApp are over NDMP.

Last week one of the drives on the NetApp failed. The drive was removed and
the entry in the jukebox for that drive was set to disabled. The entry for
Networker for that drive was also set to disabled (taken out of service mode).

Tried to do backups from the NetApp using the remaining drive but just got
messages about terminating on signal 11 each time the remianing drive was
accessed.

I decided to delete the device entry in Networker for the remote drive that
was removed. This stopped the signal 11 errors. Had to do this with jbedit
because the GUI gave me endless complaints. But than I got the following
messages when trying to load a tape in the remaining drive:
Illegal Request, ATL :
transfer empty - command aborted

BHTi : no destination drive

EXABYTE : destination drive not installed

QUALSTAR: destination drive not installed

SPECTRA : destination drive not available

The NetApp sees the remaining drive and I can probe for tape drive status and
I can load and unload tapes via the jukebox front panel.


Decided to delete the second drive from Networker and put it back with jbedit.
It asked me for the element ID and gave me a deafult of [1] which I accepted.
This put back the drive but requests to mount tapes in this drive went the
first local drive on the jukebox and things hung.


Decided to remove the jukebox resource and reinstall using jbconfig.
After discovering the the auto-detected SCSI library and two local drives
information is requested for the next drive which I configure as a remote
drive and am prompted to configure the node on which the device is being
configured as a Dedicated Storage Node (DSN) to which I respond [no] no.
Jbconfig exits with a message about a missing enabler licences. If I respond
yes I get the same message.


Now I am confused. My original installation was a migration update from 6.1.3
and I had reponded no to Dedicated Storage Node since on my Solaris server
backups other clients as well. But now I wonder which node is being referred
to: the netapp or the networker host. Either way I cannot get jbconfig to work.

.... lots of cursing ...

I recoverd nsrdb and returned to my original configuration before I started to
mess around. I disabled the two netapp attached drives and the netapp group.
At least I can back up my non netapp clients.


Anyone have experience with a situation like this and can throw some light.




Joel









| Joel Krajden | Rm: LB-915, Tel: 514 848-2424 3052 |
| | Fax: 514 848-2830 |
| Senior Systems Analyst | Email: joelk < at > cs.concordia.ca |
| Dept. of Computer Science | http://www.cs.concordia.ca/~staffcs/joelk |
| Concordia University | Remember it's a circus and the clowns |
| Montreal, Canada | are supposed to make you laugh, not cry. |

Note: To sign off this list, send a "signoff networker" command via email

On Mon, 20 Sep 2004, Joel Krajden wrote:

Solaris 9/Networker7.1; NetApp running 6.5.2R1, Qualstar 8464 with 4 LTO drives.

The Solaris 9/Networker7.1 server controls the jukebox robotics and two
drives. Two drives in the jukebox are connected over SCSI to the NetApp.
Backups for the NetApp are over NDMP.

Last week one of the drives on the NetApp failed. The drive was removed and
the entry in the jukebox for that drive was set to disabled. The entry for
Networker for that drive was also set to disabled (taken out of service mode).

Tried to do backups from the NetApp using the remaining drive but just got
messages about terminating on signal 11 each time the remianing drive was
accessed.

When you removed that tape drive, did you put a SCSI terminator on the
SCSI bus to make up for the missing tape drive? If not, that might
be the cause of your tape drive problem.

Note: To sign off this list, send a "signoff networker" command via email

Stan,

I wish things were that simple. The remaining tape device is terminated.
And all is well from the filers point of view.

My suspcion is that networker over NDMP still thinks there is a drive there
when there isn't.

Joel

Stan Horwitz wrote:
On Mon, 20 Sep 2004, Joel Krajden wrote:


Solaris 9/Networker7.1; NetApp running 6.5.2R1, Qualstar 8464 with 4 LTO drives.

The Solaris 9/Networker7.1 server controls the jukebox robotics and two
drives. Two drives in the jukebox are connected over SCSI to the NetApp.
Backups for the NetApp are over NDMP.

Last week one of the drives on the NetApp failed. The drive was removed and
the entry in the jukebox for that drive was set to disabled. The entry for
Networker for that drive was also set to disabled (taken out of service mode).

Tried to do backups from the NetApp using the remaining drive but just got
messages about terminating on signal 11 each time the remianing drive was
accessed.


When you removed that tape drive, did you put a SCSI terminator on the
SCSI bus to make up for the missing tape drive? If not, that might
be the cause of your tape drive problem.


| Joel Krajden | Rm: LB-915, Tel: 514 848-2424 3052 |
| | Fax: 514 848-2830 |
| Senior Systems Analyst | Email: joelk < at > cs.concordia.ca |
| Dept. of Computer Science | http://www.cs.concordia.ca/~staffcs/joelk |
| Concordia University | Remember it's a circus and the clowns |
| Montreal, Canada | are supposed to make you laugh, not cry. |

Note: To sign off this list, send a "signoff networker" command via email

-----Original Message-----
From: Joel Krajden [mailto:joelk < at > CS.CONCORDIA.CA]
Sent: Monday, September 20, 2004 19:33
To: NETWORKER < at > LISTMAIL.TEMPLE.EDU
Subject: [Networker] NDMP backup fails after one drive is removed from
jukebox

I decided to delete the device entry in Networker for the
remote drive that
was removed. This stopped the signal 11 errors. Had to do
this with jbedit
because the GUI gave me endless complaints.
Complaints about... what. Aren't these relevant?

I guess I'd do the same as you did. Remove the drive in the GUI,
after removing it from any pool that used it.

Decided to remove the jukebox resource and reinstall using jbconfig.
After discovering the the auto-detected SCSI library and two local
drives...
Two local drives? One was removed wasn't it?
Inquire sees only one drive?

Seems to be some sort of SCSI-problem if it sees ghost drives.

Cor Kuin.

Note: To sign off this list, send a "signoff networker" command via email

Kuin, CNM wrote:
-----Original Message-----
From: Joel Krajden [mailto:joelk < at > CS.CONCORDIA.CA]
Sent: Monday, September 20, 2004 19:33
To: NETWORKER < at > LISTMAIL.TEMPLE.EDU
Subject: [Networker] NDMP backup fails after one drive is removed from
jukebox

I decided to delete the device entry in Networker for the
remote drive that
was removed. This stopped the signal 11 errors. Had to do
this with jbedit
because the GUI gave me endless complaints.

Complaints about... what. Aren't these relevant?

Yes and no. Before you can delete the device you need to delete it from the
pools. No problem I had done that. Next you need to delete the devices from
the jukebox resource. When you try that networker complains that the number of
devices does not match not match the number of drives. So I reduced the number
of drives and devices from 4 to 3. Try again to delete and networker complains
about the number of loaded slots. Which slots are they talking about? Tapes in
drive slots? All 4 tape drives are empty. What does this have to do with the
attempt at device deletion and just what am I supposed to do. Searched the
manuals and man pages and still could not figure out what was required.
Defeated by bad software/documentation I gave up.



I guess I'd do the same as you did. Remove the drive in the GUI,
after removing it from any pool that used it.


Decided to remove the jukebox resource and reinstall using jbconfig.
After discovering the the auto-detected SCSI library and two local

drives...
Two local drives? One was removed wasn't it?
Inquire sees only one drive?

I have 4 drives in the jukebox. Two directly attached to the networker server
and two attached to the filer.

Inquire sees the MC and the two local drives.

Inquire -N prompts for the NDMP tape server and password but returns nothing
except the string "this may take some time".



Seems to be some sort of SCSI-problem if it sees ghost drives.

Cor Kuin.

--
Note: To sign off this list, send a "signoff networker" command via email


| Joel Krajden | Rm: LB-915, Tel: 514 848-2424 3052 |
| | Fax: 514 848-2830 |
| Senior Systems Analyst | Email: joelk < at > cs.concordia.ca |
| Dept. of Computer Science | http://www.cs.concordia.ca/~staffcs/joelk |
| Concordia University | Remember it's a circus and the clowns |
| Montreal, Canada | are supposed to make you laugh, not cry. |

Note: To sign off this list, send a "signoff networker" command via email