Bus/Target vs. Taget/Bus

[Christian Bieletzki]

I have an important question to the taget/bus or bus/target setting.

I currently have the situation that our NSR is put first and then I see the tape drives

0 AF Lun= 0 CTLR UP COMPAQ NS ROUTER E1200
1 PSCSI 1 Target= 5 Lun= 0 CHGR UP COMPAQ MSL5000 Series 0423
2 PSCSI 1 Target= 1 Lun= 0 TAPE UP HP Ultrium 1-SCSI E38W
3 PSCSI 1 Target= 2 Lun= 0 TAPE UP HP Ultrium 1-SCSI E38W

Running this configuration shows a total of 4 tape drives in device manager and the configuration is unstable. Especially if there is a backup problem we need to reboot all SAN attached servers. Will a change in configuration by using the "Bus/Target" order maybe fix the reboot issues?

Regards
Christian

[Steven Clementi]

Christian:

Which backup software are you using?

For some backup solutions to pwork properly, your mappings must be done via "target/bus".

When mapping is done via "bus/target", the autofill places the NSR last in the list and some software has a problem talking to the library.

Chaning the oder may fix the problem, but there might be other issues effecting your backup solution as well.

Is your backup server(s) in separate zones from other devices? I.e. 1 zone per backup server which includes the host port and the nsr port? Or do you have 1 zone with multiple hosts? or combined with storage ports?


Steven

[Christian Bieletzki]

Hi Steven,

We use Windows 2003 and we have Veritas Backup Exec 9.1. That is causing loads of trouble once we need to power cycle the library. The maping is as follows:

MAPPED HOSTS

GPSMX04 Host Id undefined FC Port 0 Indexed
GPSBackup Host Id undefined FC Port 0 Indexed
GPSFP01 Host Id undefined FC Port 0 Indexed
GPSSD01 Host Id undefined FC Port 0 Indexed
C92FAC0B (FC Port Name (Low)) Host Id undefined FC Port 0 Indexed
C92B19A7 (FC Port Name (Low)) Host Id undefined FC Port 0 Indexed

ACTIVE MAPS
Active FC Maps

Indexed FC Port 0 Map

LUN Protocol Bus Device Specific Address Type Status Device Description
0 AF 0 Lun= 0 CTLR UP COMPAQ NS ROUTER E1200
1 SCSI 1 Target= 5 Lun= 0 CHGR UP COMPAQ MSL5000 Series 0423
2 SCSI 1 Target= 1 Lun= 0 TAPE UP HP Ultrium 1-SCSI E38W
3 SCSI 1 Target= 2 Lun= 0 TAPE UP HP Ultrium 1-SCSI E38W

I don't know if you have all information from there. Let me know if you need more.

We plan to reconfigure our library with is currently serving 4 drives as follows.

Bus1 on on NSR1 has only the Robotics.
Bus2 on the NSR1 has drive 1+2

Bus1 on NSR2 has only one drive 3
Bus2 on NSR2 has only one drive 4

Christian

[Steven Clementi]

You have 6 servers accessing the library.

What is your zoning configuration? Can you post a capture of it? Depending on how your switch is zoned, some of the servers might be effected by problems occuring on 1 particular server.

telnet into the SAN Switch where your NSR is attached and run these commands...

switchshow
zoneshow
version

Or just cfgshow.

Capture the output and post here if you can.

Steven

[Christian Bieletzki]

I have attached the information

[Steven Clementi]

Have you always had this problem?


The first potential problem I see is that you have 1 zone with all your servers and your nsr's.


Potentially, one particular server can be causing some problems on the "scsi bus" causing all the servers to have issues.

HP best practices for zoning usually say to create separate zones for each individual server that you have attaching to another SAn Resource (Tape or Storage). In your case, there should be 7 different zones for each initiator (your host). "Single Initiator Zoning".

This could be only part of the problem, but lets take it one step at a time.

Are all the servers 1 OS? or multiple types of OS? All windows? All Linux? a Mix?


Steven

[Christian Bieletzki]

all servers have the same OS which is Windows 2003 server.

Yes, we have that problem since a long time now. If it runs, then it is ok for three or four weeks - sometimes longer but then it fails.

If it fails, the there are no error messages for the tape hardware or the library itself however, I see the CPQKPGSA driver with SCSI timeouts in the server's event log.

We have one master Backup Server, However the data stream travels between each server itself and the library as all of our attached servers have the SAN SSO Option.

[Steven Clementi]

Seeing many timeouts and CPQKQPSA errors could mean that you have a bad card or intemintant hba, maybe even a faulty nsr port or even a faulty transceiver on your SAN Switch

If it is an hba, it could be effecting all the servers since they are technically all on the same visible bus (the "Backup_All" zone). If it is an nsr issue, the zoning would not matter too much since all the servers wouldbe able to see the nsr anyway.

Either way, you would have to start making some changes to see if the issues gets resolved. It will obviously take a while to determine if the issue is resolved. It may also require some interaction with HP.

If it comes to having HP take a long, one of the first things they will suggest is changing your zoning. If you up to it, We can help you throug it if necessary.

Steven

[Marino Meloni_1]

>>>I currently have the situation that our NSR is put first and then I see the tape drives
0 AF Lun= 0 CTLR UP COMPAQ NS ROUTER E1200
1 PSCSI 1 Target= 5 Lun= 0 CHGR UP COMPAQ MSL5000 Series 0423
2 PSCSI 1 Target= 1 Lun= 0 TAPE UP HP Ultrium 1-SCSI E38W
3 PSCSI 1 Target= 2 Lun= 0 TAPE UP HP Ultrium 1-SCSI E38W
Running this configuration shows a total of 4 tape drives in device manager and the configuration is unstable

-I can just see two drives mapped on your nsr, do you have two nsr?
-Your configuration is fine, with Windows environment

>>>We plan to reconfigure our library with is currently serving 4 drives as follows.
Bus1 on on NSR1 has only the Robotics.
Bus2 on the NSR1 has drive 1+2
Bus1 on NSR2 has only one drive 3
Bus2 on NSR2 has only one drive 4

Do you have already two NSR> so first config is fine, alternatively you can remove the SF (NSR E1200) from your map but this will change nothing if you do not use direct backup (mean from disk system to dibrary also called serverless backup)

What I would recommand is to update NSR firmware, and library aswell

[Joseph Salzer_3]

One thing to look at also, is that if I remember correctly, Backup Exec expects to see the changer first.