IBM Support

HSCLA27C Partition Mobility Error

Technote (troubleshooting)


Problem(Abstract)

Live Partition Mobility fails with HSCLA27C error.

Symptom

HSCLA27C The operation to get the physical device location for adapter <adapter ID> on the virtual I/O server partition <VIOS LPAR> has failed.  The partition command is: migmgr -f get_adapter -t vscsi -s <adapter ID> -w 13857705810988171408 -W 13857705810988171409 -d 0 The partition standard error is: child process returned error

HSCLA27C The operation to get the physical device location for adapter  
<adapter ID> on the virtual I/O server partition <VIOS LPAR> has failed.
The partition command is: migmgr -f get_adapter -t vscsi -s <adapter ID> -w 13857705810988171414 -W 13857705810988171415 -d 0 The partition standard error is: child process returned error


Cause

This technote discusses the most common causes for the above error running VIOS 2.2.x


Environment

HMC managed LPM environment typically using virtual Fibre Channel adapters.

Diagnosing the problem

The HSCLA27C errors in and of itself can be rather genetic. If the migration/validation is being done via HMC GUI when it fails, click on the "Detailed Information" section in order to better understand the cause of failure


1. If the HMC level is older than v3.5.0, all virtual Adapters on source VIO server(s) profile properties tab must be created in a "desired" state (no). If set in a "required" state (yes), LPM will fail with HSCLA27C.

2. For active (live partition) mobility, ALL adapters assigned to the mobile LPAR MUST be logged in or LPM will fail with HSCLA27C. All vfchost adapters must be mapped and logged in or removed.

3. LPM requires that all fcs ports that are mapped to the mobile LPAR MUST have a SAN storage device presented to the fcs port. LPM requires that all virtual fcs adapters on the mobile LPAR MUST have at least one SAN disk zoned to the fcs port's primary AND secondary WWPN. If the mobile partition has any virtual fcs devices with no storage zoned, either zone one LUN or remove the virtual FC Client Adapter (on the client) and the respective virtual FC Server adapter (vfchost) from the VIOS.

4. The same FC adapter port (fcs# on the source VIOS) cannot be virtualized more than once to the same mobile partition.
Example of an incorrect configuration: fcs3 virtualized through vfchost8 and vfchost9, and both vfchost#s are mapped to the same mobile partition. This is not supported by LPM and may cause the validation to fail with the detailed information as shown below, in addition to the HSCLA27C error:

      HSCL400A There was a problem running the VIOS command. HSCLA29A The RMC command issued to partition <mobile LPAR>  failed.
      The partition command is:
      migmgr -f get_adapter -t vscsi -s <adapter ID>  -w
      13857705815095050398 -W 13857705815095050399 -c RPA -a ACTIVE_LPM -d 0
      The RMC return code is:
      0
      The OS command return code is:
      82
      The OS standard out is:
      82

      VIOS_DETAILED_ERROR
      domain_id for fscsi0 is: 149
      Stream id associated with port = 0
      Effective compatability = -1
      SCSI_ID=950600 WWPN=5000097208696521, SCSI_ID=950700
      WWPN=5000097208696519
      VFC txu= 100000, fcs txu= 100000
      Checking partition num of vfchost0
      ...
      Checking partition num of vfchost31
      Comparing SCSI targets for 1 VFC adapters                              
      Comparing SCSI targets for vfchost9 seen through fscsi3 to those of
      vfchost8
      Retrieved wwpn 0 for vfchost9 from ODM
      Do not have wwpn for vfchost9 so not safe to migrate
      Inactive needs to be preceeded by at least 1 successful live migration
      for wwpn propogation
      End Detailed Message.
    Recommendation
    Ensure each vfchost# used by the mobile partition is mapped to a different FC adapter port on the source VIOS. Example of a correct configuration: fcs3 is virtualized through vfchost8 and fcs4 is virtualized through vfchost9.

5. Check the VIO server and mobile LPAR for any full file systems:
    On VIOS:
    $ df -g

    On AIX:
    # df -g

6. Check vendor specific disks for NPIV enablement. Example;
    For Hitachi HDLM Tunables the HDLM ODM setting required is:

    # /usr/D*/bin/dlmodmset -o
    Lun Reset                   : on
    Online(E) IO Block          : off
    NPIV Option                 : off
    KAPL10800-I The dlmodmset utility completed normally.

    # /usr/D*/bin/dlmodmset -v on
    KAPL10805-I The setup of the HDLM execution environment ODM will be changed. NPIV Option = on. Is this OK? [y/n]:y
    KAPL10800-I The dlmodmset utility completed normally.
    #
    # /usr/D*/bin/dlmodmset -o
    Lun Reset                   : on
    Online(E) IO Block          : off
    NPIV Option                 : on
    KAPL10800-I The dlmodmset utility completed normally.

    This should be set before the devices are configured.

7. Updating VIOS via updateios without the -install flag or via alt_root_vg command with -bundle update_all option will only update filesets that are already installed. New filesets that are not forced in via prerequisites will not be installed. This can lead to LPM operation to fail with error similar to the one below:

    HSCLA27C The operation to get the physical device location for adapter <adapter ID>  on the virtual I/O server partition <VIOS LPAR name>  has failed.
    The partition command is: migmgr -f get_adapter -t vscsi -s
    <adapter ID>  -d 0 The partition standard error is: child process returned error

    At higher debug levels or if you run the failing command at the VIOS command line you will see an error trying to load libvlog.a

    Running method '/usr/lib/methods/mig_vscsi'
    exec(): 0509-036 Cannot load program migmgr because of the following errors:
    0509-150 Dependent module libvlog.a(shr.o) could not be loaded.
    0509-022 Cannot load module libvlog.a(shr.o).
    0509-026 System error: A file or directory in the path name does not exist.


    Next is a list of filesets known to not be installed when tested locally. The list may very depending on ioslevel the VIOS was updated to:

    ios.database.rte 6.1.7.0 # VIOS DB
    ios.svm.rte 6.1.7.0 # Secure Virtual Machine
    ios.vlog.rte 6.1.7.0 # Virtual Log Device Software
    ios.vnet 6.1.7.0 # Virtual Network Management
    pool.basic.rte 6.1.7.0 # Virtual Server Storage Subsy...
    pool.msg.en_US.basic.rte 6.1.7.0 # Virtual Server Storage Subsy...
    printers.hplj-4250.rte 6.1.7.0 # Hewlett-Packard LaserJet 4250
    printers.hplj-8060.rte 6.1.7.0 # Hewlett-Packard Color LaserJ..


    Recommendation
    Review /home/padmin/install.log and '# lppchk -v' output for these missing filesets. If your VIOS is missing any of the above, redo the update with proper options.
    $ updateios -commit
    $ updateios  -dev <fileset source> -install -accept
    A reboot may be necessary to get everything started correctly

8. If all above fail, this is a zoning issue in which the second WWPN was not manually zoned correctly or failed to login. Both of the WWPNs from the virtual fiber channel adapter in the client LPAR's profile properties MUST be zoned through the same VSAN / NODE / FABRIC for LPM. Engage your SAN admin to verify the storage is zoned to the second WWPN for each virtual FC adapter configured on the mobile LPAR.
HMC command, chnportlogin, can be used to to help SAN admin troubleshoot WWPN logins of the secondary WWPNs.

Resolving the problem

Make any recommended changes for any of the possible problems in the list above.

Related information

NPIV: HSCLA27C HSCL400A HSCLA29A with OS RC 83 during L
vSCSI: HSCLA27C HSCL400A HSCLA29A with OS RC 80 during
Suspend Fails with Message HSCLA27C

Document information

More support for: Virtual I/O Server

Software version: 2.2.0, 2.2.1, 2.2.2, 2.2.3, 2.2.4, 2.2.5

Operating system(s): AIX, Other

Software edition: Enterprise

Reference #: T1018971

Modified date: 26 April 2017


Translate this page: