Unable to remove virtual scsi resource from running profile - Aix

This is a discussion on Unable to remove virtual scsi resource from running profile - Aix ; Hello All, Any help with this problem would be appreciated as its driving me nuts. As part of a new lpar build I've added a virtual scsi resource to a vio server's running profile via DLPAR operation. The DLPAR operation ...

+ Reply to Thread
Results 1 to 6 of 6

Thread: Unable to remove virtual scsi resource from running profile

  1. Unable to remove virtual scsi resource from running profile

    Hello All,

    Any help with this problem would be appreciated as its driving me
    nuts.

    As part of a new lpar build I've added a virtual scsi resource to a
    vio server's running profile via DLPAR operation. The DLPAR operation
    failed to complete successfully because of some RMC comms problem.

    A refresh of the RMC daemons on the VIO server has solved the comms
    problem, but now I am left with a virtual adapter resource on the
    running profile, but not on the server itself. Running cfgdev does not
    pick these resources up either, and I have found it impossible to
    deleted them from the running profile.

    When I attempt to remove them with a DLPAR operation from the WebSM,
    the 'delete' button doesn't do anything. I have tried to delete them
    from the HMC using the following command, but it fails with an error
    message that is as useful as a chocolate spade.

    :~> chhwres -r virtualio -m MNUXMIF -o r -p mnuf01v --rsubtype scsi -s
    126 -d --force

    HSCL294D DLPAR REMOVE Virtual I/O resources failed: HMC Virtual slot
    DLPAR operation failed. Here are the virtual slot IDs that failed and
    the reasons for failure:
    126 The RMC operation failed.

    >From IBM website:

    HSCL294D Dynamic logical partition REMOVE Virtual I/O resources
    failed: {0}.

    1. Correct the error and retry the operation.
    2. If the problem persists, contact your next level of support or
    your service provider.

    Does anyone know how I can remove the virtual devices from the vio
    servers running profiles without rebooting the vio server?

    Thanks and Regards,
    Leon


  2. Re: Unable to remove virtual scsi resource from running profile

    It may be a stupid question, but have you deleted the device (and its
    children) from VIO first?

    Also, I'm not 100% sure if I understood you correctly. Profile
    modification does not require communication with partition. DLPAR
    does, but it does not modify any of the partition's profiles.

    Any changes to a profile require partition shutdown. DLPAR does not.

    If DLPARing keeps failing, you may try to restart HMC and try again
    once it's up.
    Contact IBM support if the problem persists.


  3. Re: Unable to remove virtual scsi resource from running profile

    On Feb 8, 11:21 am, "brother_Matthius" wrote:
    > It may be a stupid question, but have you deleted the device (and its
    > children) from VIO first?
    >
    > Also, I'm not 100% sure if I understood you correctly. Profile
    > modification does not require communication with partition. DLPAR
    > does, but it does not modify any of the partition's profiles.
    >
    > Any changes to a profile require partition shutdown. DLPAR does not.
    >
    > If DLPARing keeps failing, you may try to restart HMC and try again
    > once it's up.
    > Contact IBM support if the problem persists.


    Sorry if I haven't made myself clear. Forget the partition profile, I
    am only talking about DLPAR and the running profile

    I haven't deleted the devices from the vio server because they have
    never been created. The original DLPAR operation failed due to RMC
    problems ie. they exist on the running profile but not on the server
    itself. Essentially, the vio server and the HMC are out of sync
    because of this failed DLPAR operation, and I can't seem to re-sync
    them because
    1: I can't add the resource to the vio server by running cfgdev
    2: I can't remove the virtual resources because the DLPAR operation
    fails.

    Just to clarify, DLPAR IS working for the vio server. I can add and
    remove other resources, including adding and removing virtual scsi. It
    is only the DLPAR operation to remove the out-of-sync virtual scsi
    that I previously tried to define.

    I suspect the DLPAR operation to remove the virtual adapters is
    failing BECAUSE they don't exist on the vio server. A bit of a
    Catch-22 situation. I can't delete them from the HMC because the don't
    exist on the vio server, but at the same time I can't add them to the
    vio either.

    Rebooting the HMC hasn't solved the problem.


  4. Re: Unable to remove virtual scsi resource from running profile

    > I suspect the DLPAR operation to remove the virtual adapters is
    > failing BECAUSE they don't exist on the vio server. A bit of a
    > Catch-22 situation. I can't delete them from the HMC because the don't
    > exist on the vio server, but at the same time I can't add them to the
    > vio either.
    >
    > Rebooting the HMC hasn't solved the problem.


    OK, thanks for clarification. :-)

    Looks like either P5 Hypervisor or SPs or HMC are playing dirty
    tricks.

    It's not an expert advice, but I'd try to restart both service
    processors.
    If it won't help, connect a second HMC with no configuration data
    about this particular server. Let it download config from server's
    NVRAM.
    If problem persists, then I'm afraid you will have to open a PMR at
    IBM and/or restart the partition.



  5. Re: Unable to remove virtual scsi resource from running profile

    On Feb 8, 2:28 pm, "brother_Matthius" wrote:
    > > I suspect the DLPAR operation to remove the virtual adapters is
    > > failing BECAUSE they don't exist on the vio server. A bit of a
    > > Catch-22 situation. I can't delete them from the HMC because the don't
    > > exist on the vio server, but at the same time I can't add them to the
    > > vio either.

    >
    > > Rebooting the HMC hasn't solved the problem.

    >
    > OK, thanks for clarification. :-)
    >
    > Looks like either P5 Hypervisor or SPs or HMC are playing dirty
    > tricks.
    >
    > It's not an expert advice, but I'd try to restart both service
    > processors.
    > If it won't help, connect a second HMC with no configuration data
    > about this particular server. Let it download config from server's
    > NVRAM.
    > If problem persists, then I'm afraid you will have to open a PMR at
    > IBM and/or restart the partition.


    I haven't reset the service processors as I don't beleive that will
    make any difference as they don't hold any partition information. I
    have tried the following, non of which has worked I might add.

    Performed a 'reload'. This copies the profile information held on the
    managed systems NVRAM onto the HMC. I have also removed the managed
    system and re-added it, forcing it to read the managed servers NVRAM.
    Alas, I still have the same problem.


  6. Re: Unable to remove virtual scsi resource from running profile

    On Feb 9, 2:45 am, "EddieB" wrote:
    > On Feb 8, 2:28 pm, "brother_Matthius" wrote:
    >
    >
    >
    > > > I suspect the DLPAR operation to remove the virtual adapters is
    > > > failing BECAUSE they don't exist on the vio server. A bit of a
    > > > Catch-22 situation. I can't delete them from the HMC because the don't
    > > > exist on the vio server, but at the same time I can't add them to the
    > > > vio either.

    >
    > > > Rebooting the HMC hasn't solved the problem.

    >
    > > OK, thanks for clarification. :-)

    >
    > > Looks like either P5 Hypervisor or SPs or HMC are playing dirty
    > > tricks.

    >
    > > It's not an expert advice, but I'd try to restart both service
    > > processors.
    > > If it won't help, connect a second HMC with no configuration data
    > > about this particular server. Let it download config from server's
    > > NVRAM.
    > > If problem persists, then I'm afraid you will have to open a PMR at
    > > IBM and/or restart the partition.

    >
    > I haven't reset the service processors as I don't beleive that will
    > make any difference as they don't hold any partition information. I
    > have tried the following, non of which has worked I might add.
    >
    > Performed a 'reload'. This copies the profile information held on the
    > managed systems NVRAM onto the HMC. I have also removed the managed
    > system and re-added it, forcing it to read the managed servers NVRAM.
    > Alas, I still have the same problem.



    try this:


    # rsthwres -m -r



+ Reply to Thread