GPFS problems

From: Jan-Frode Myklebust (janfrode@PARALLAB.UIB.NO)
Date: Mon Apr 07 2003 - 19:01:59 EDT

• Next message: Patrick B. O'Brien: "mkcd -L -d /dev/cd0 error"
• Previous message: Michelle DeVault: "password on SMS menu?"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Hi,

we just had a diskfailure in a un-replicated GPFS/hacmp filesystem, and of
course lost the whole filesystem. After the failure I deleted the
filesystem with:

        # mmdelfs parallab -p

and tried to recreate it on the remaining disks. Unfortunately this
gave me problems.

# mmcrfs /home/parallab parallab -F /home/ibm/disk.desc -A no -Q yes -C gpfs1
GPFS: 6027-470 Disk gpfslv40 may still belong to an active file system.
GPFS: 6027-470 Disk gpfslv50 may still belong to an active file system.
GPFS: 6027-470 Disk gpfslv45 may still belong to an active file system.
etc..

OK, I assumed something was locked in the existing volumes, and since
I don't know what I decided to just export them, and recreate them.

        # masternode:
        mkvg -n -f -s 64 -c -y gpfshomevg$i hdisk$i
        varyonvg gpfshomevg$i
        mklv -b n -w n -y gpfshomelv$i gpfshomevg$i 543
        varyoffvg gpfshomevg$i

        # other nodes:
        importvg -y gpfshomevg$i $HDISK
        varyonvg gpfshomevg$i
        chvg -a n gpfshomevg$i
        varyoffvg gpfshomevg$i

        # all nodes:
        /usr/es/sbin/cluster/utilities/fence_clear_all -h $HDISK
        varyonvg -c gpfshomevg$i

        # put disk descriptions into /home/ibm/homeparallab.desc
        mmcrfs /home/parallab parallab -F /home/ibm/homeparallab.desc -C gpfs1 -A no -Q yes -M 2 -R 2
        GPFS: 6027-470 Disk gpfshomelv95 may still belong to an active file system.
        GPFS: 6027-470 Disk gpfshomelv99 may still belong to an active file system.
        GPFS: 6027-470 Disk gpfshomelv97 may still belong to an active file system.
        GPFS: 6027-470 Disk gpfshomelv74 may still belong to an active file system.
        etc..

and still same error. The mmfs.latest.log only says:

        Tue Apr 8 00:25:53 2003: Command: tsctl showCfgValue maxblocksize
        Tue Apr 8 00:25:54 2003: Command: tscrfs /dev/plab -F /var/mmfs/tmp/tsddFile.mmcrfs.45848 -c 0 -I 16384 -i 512 -M 2 -n 32 -R 2 -s roundRobin -w 0
        Tue Apr 8 00:25:57 2003: Command err 22: tscrfs /dev/plab -F /var/mmfs/tmp/tsddFile.mmcrfs.45848 -c 0 -I 16384 -i 512 -M 2 -n 32 -R 2 -s roundRobin -w 0

which isn't exactly enligthening to me...

Does anybody have any idea to what might be the problem here? I have
contacted IBM-support, but still not gotten trough to someone
competeny, so maybe this list might be quicker?

  -jf

• Next message: Patrick B. O'Brien: "mkcd -L -d /dev/cd0 error"
• Previous message: Michelle DeVault: "password on SMS menu?"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:16:43 EDT