IT00584: VIRTUAL MACHINE GUEST DOES NOT POWER ON AFTER INSTANT ACCESS RESTORE WHEN DATA IS STORED ON A VIRTUAL TAPE LIBRARY

IBM Tivoli Storage Manager (TSM) client 7.1.0.x interim fix downloads
IBM Tivoli Storage Manager (TSM) Client 7.1.1 Downloads and READMEs

APAR status

• Closed as program error.

Error description

• When VMware backup data is stored in a Virtual Tape Library
  (VTL), a Full VM Instant Access (IA) restore operation of the
  Virtual Machine (VM) guest does not work correctly. Tivoli
  Storage Manager (TSM) reports that the operation was successful
  and that the virtual disks are ready for access. However, when
  you try to power on the guest it will not boot.
  
  This problem occurs due to contention for the same tape volume
  between the data mover (dsmagent.exe) and the Recovery Agent
  mount command (TDPVMwareShell."mount"). This contention occurs
  because objects needed to build the virtual machine, and the
  actual data from the machine are on the same volume.  The
  following is a brief overview of the events:
   - The data mover mounts the volume and restores data.
   - The data mover session initiates the mount operation and does
     not end. This keeps the volume mounted to that session.
   - The mount operation session sits in MediaW (media wait) while
     the data mover session is in IdleW (idle wait).
   - After 90 seconds the mount operation stops waiting for the
     tape and ends its session.
   - The data mover ends its session with the TSM server and
     reports that the VM disks are ready for Instant Access.
  
  At the time the sessions are in their respective wait states,
  the operation is effectively in a deadlock situation: the data
  mover session will not end until mount completes, and mount
  cannot complete until the data mover releases the tape by ending
  its session.
  
  Note: This problem cannot occur when the data is stored in
  random access or sequential access disk/file storage pools.
  
  Versions Affected: Tivoli Storage Manager Backup-Archive Client
  version 7.1 on Windows
  
  Customer/L2 Diagnostics: A 'service' level trace of the data
  mover and Recovery Agent mount should be gathered to confirm
  this issue.  Additionally, review the 'query session
  format=detailed' and 'show asvol' outputs on the Tivoli Storage
  Manager Server to verify if there are sessions in MediaW
  requesting volumes which are mounted under sessions for the same
  node name that are in IdleW.  For example, the following shows
  session 310 has a volume mounted that session 312 is waiting
  for:
             Sess Number: 310
            Comm. Method: Tcp/Ip
              Sess State: IdleW
                Platform: TDP VMware
             Client Name: PROXY1_DC2 (PROXY1_DC2_DM)
     Media Access Status: Current output volumes: VOL1234
  
             Sess Number: 312
            Comm. Method: Tcp/Ip
              Sess State: MediaW
                Platform: TDP VMware
             Client Name: PROXY1_DC2 (PROXY1_DC2_DM)
     Media Access Status: Waiting for mount of volume: VOL1234
  
  Initial Impact: Medium
  
  Additional Keywords: tsmve power on console hang tdp4ve
  

Local fix

• Before initiating the IA restore, verify the current IDLETIMEOUT
  value with 'query opt idletimeout'.  Then reduce the
  'IDLETIMEOUT' to 1min using 'setopt idletimeout 1'. Once the
  restore has completed, and the machine can be powered on,
  increase the timeout to the prior value with the setopt command.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: Users of the TSM for VE Instant Access or    *
  *                 Instant Restore function where the backend   *
  *                 data store of the TSM Server is a Virtual    *
  *                 Tape Library (VTL).                          *
  ****************************************************************
  * PROBLEM DESCRIPTION: See ERROR DESCRIPTION                   *
  ****************************************************************
  * RECOMMENDATION: Apply fixing level when available. This      *
  *                 problem is currently projected to be fixed   *
  *                 in level 7.1.1. Note that until the fixing   *
  *                 level is available, this information is      *
  *                 subject to change at the discretion of IBM.  *
  ****************************************************************
  TSM had two sessions open to the server.  The first session
  held the 'tape' mount for the VTL.  This blocked the second
  session from getting access to the 'tape' and blocked the
  function.
  

Problem conclusion

• The client was modified to close the first session so the
  mount can complete for the second session.
  

Temporary fix

• A fix for this problem is currently projected to be available
  in level 7.1.0.3. Note that until the fixing level is available
  this information is subject to change at the discretion of IBM.
  

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  TSM CLIENT

• Fixed component ID

  5698ISMCL

Applicable component levels

• R71W PSY

     UP