Reserve conflict

[mrcuser]

Hi we have a lot of reserve conflicts in our errpt -a on our TSM servers the errors look like this:

LABEL: RESERVE_CONFLICT
IDENTIFIER: BF05CF18

Date/Time: Sat Feb 28 12:28:46 GMT 2009
Sequence Number: 2007
Machine Id: 005D754A4C00
Node Id: tsm5
Class: H
Type: INFO
Resource Name: rmt14
Resource Class: tape
Resource Type: 3592
Location: U0.1-P1-I3/Q1-W5005076302401518-L0
VPD:
Manufacturer................IBM
Machine Type and Model......03592J1A
Serial Number...............000007806530
Device Specific.(FW)........0C90
Loadable Microcode Level....A170029E

Description
AAA5

Probable Causes
TAPE DRIVE

Failure Causes
TAPE DRIVE

Recommended Actions
AAA6

Detail Data
ADDITIONAL INFORMATION
Unable to obtain reserving host information, errno EBUSY

I've looked at our paths and drives setup and have matched the correct serials numbers to the correct devices in aix and defined them as they should be but we're still getting a log media mount not possible errors in the activity log.
Is there anything to check additional to correct path and drive setup? Ive looked at the zoning and the

http://wwwns

all relate to the drives we've setup but still no joy. The drives are in a single 3494 library and are shared byt 4 tsm servers.

 

[ProbablyDown]

Hmm.. DO you have more than one library manager or just one TSM is the lib manager and the rest are clients?

 

[mrcuser]

We have 2 libraries which server 9 servers.
I did a reteach on the library last night as well as rebooting our lmc - this has now fixed the reserve conflicts, and the paths have all stayed online.Hopefully this has addressed the problem.

 

[casuser1]

I have had the same issue. What version of ATAPE are you on?

 

[mrcuser]

I will check tomorrow - the problem is intermittent it appeared to be around our zoning and pathing but thats all correct. Its a total pest.

 

[casuser1]

I have had it happen twice now. When I reboot the TSM server, it goes away. I upgraded ATAPE from 10.7.3 to 11.2.9 in March and that is when it started. I didn't really notice what did it the first time, but the second time it happened, there was a tape stuck in the gripper and a few hours later it started with the reservation conflict errors.

 

[mrcuser]

Hmmmmm - we're using 11.2.9.0. We seem to get them when either a path it attempted to be updated and that drive is in use or just randomly.I've searched the actlog for whats running at the time of the error in the errpt report but have ended up finding nothing.

 

[casuser1]

If you look at the readme for the newest version of Atape, you will see a lot of entries with "reservation" in it. I don't know if that will fix any problem we seem to be having, but I am thinking of upgrading to a newer version in the next month or two. I will try and keep track of any more errors we get and what I can try to connect it to.

 

[mrcuser]

Background on our setup:
1 TSM server - LMC controls 8 tsm server mount requests split over 2 libraries. Also we share physical hardware logically.
ie.
/dev/rmt1 is defined in TSM as RMT1_B and RMT1_C this causes additional problems for us as a drive can be in use but go to polling as the RMT_c drive will think its unavailable.

At first I though this was the issue with the reserve conflicts - ie. When I went to update the path back online I got reserve conflict as the drive was in use already. However this can happen overnight when SAN discovery is off and no one is updating the paths.

Strange problem that I dont see on a 3584 library that is direct fibre attached.

 

[casuser1]

My environment is 1 library manager and 3 library clients. The library manager controls all the tape mounts for the tsm servers. The library (a 3584)has 36 (soon to be 39) 3592 E05 tape drives that can be used by any of the TSM servers that is attached to a SAN. We also have 22 lan free clients that can use the tape drives. We have SAN discovery off and do alternate pathing for our tape drives. We also have a VTL that only 2 of our TSM servers use. I never had this problem until we upgraded to the new Atape. But we have upgraded to 5.5.2 server code, upgraded the drive & library code for the library. If you upgrade your ATAPE, let me know if it helps you. I am planning to upgrade ours in July or August. Here is the entry that makes me think that an upgrade will fix it. I may open a PMR too and see what support thinks.

11.3.4.0 Fix Read Element Status for slots exceeding max transfer size
Fix eraseimm reserving device if fails
Redesign tapeutil kill to also kill tmcp open
Fix erase immediate failover
Fix 3592 configuring with sys_encryption on as default
Change to issue multiple RES commands for read cartridge location

 

[mrcuser]

Sounds pretty similar to what we have setup. One thing to watch out for that did stop some of our problems was the lmcpd process running only on the library manager (in my case the LMC). I found the process was running on each TSM server so killed that and some problems where prevented. However I see frequent reserve conflicts throughout the day and night and pinning it has become so frustrating as IBM tell me to check all my pathing and SAN - which we have checked and its solid. May also raise a PMR but the time and what IBM expect me to collect is just so intensive that I just dont have the time!

 

[mbourne2006]

Watch out for changes in behaviour of SANDISCOVERY option as you upgrade. I was caught going from 5.5.2 to 5.5.3. The default value has changed ...

Have Fun !
Matthew Bourne
"Find a job you love and never do a day's work in your life.