Usage of SLVM

• On Solaris 10 hosts we are beginning to use SLVM (DiskSuite) for system disks.

  • see also: http://unixway.com/vm/disksuite/mirroros.html

• During jumpstart (see pluto:/jumpstart/, i.e. disk72.prof) the meta disk set will be configured.
• the feature systemfs adapts /etc/system to be available to boot with one disk. (force slvm to start if any valid statedb replicas are available)

• in case of problems an email to uco@mail.ifh.de will be generated to replace the faulty disk

  • Further messages about disk problems on this system will NOT be sent until the file /etc/no_raid_mails is deleted.

  • Fix the problem, the file /etc/no_raid_mails should be automatically removed

    • we have had different cases
      • after a crash during boot we got the message and it takes only seconds, then the problem was fixed

      • the next time we got: invoke metasync <metaset>, after running this command the command metastat -a runs without problems

      • scsi errors, see Troubleshooting

Protokoll Plattentausch root-mirror

# c0t0d0s2 soll getauscht werden

touch /etc/no_raid_mails

# vor dem Platteziehen 
metastat -a | -p
metadetach d40 d41
metadetach d30 d31
metadetach d10 d11
metadetach d0 d1
metadetach d20 d21

# Platte ziehen

prtvtoc /dev/rdsk/c0t1d0s2 | fmthard -s - /dev/rdsk/c0t0d0s2

metadb -a -c 3 -f c0t0d0s7

metattach d40 d41
metattach d30 d31
metattach d10 d11
metattach d0 d1
metattach d20 d21

installboot /usr/platform/`uname -i`/lib/fs/ufs/bootblk /dev/rdsk/c0t0d0s0


metadb

metastat -a| grep ync

Detach a submirror from a mirror for patching or live upgrading

• check the mirror

     metastat -p
    d20 -m d21 d22 1
    d21 1 1 c0t0d0s4
    d22 1 1 c0t1d0s4
    d10 -m d11 d12 1
    d11 1 1 c0t0d0s1
    d12 1 1 c0t1d0s1
    d0 -m d1 d2 1
    d1 1 1 c0t0d0s0
    d2 1 1 c0t1d0s0
    d30 -m d31 d32 1
    d31 1 1 c0t0d0s5
    d32 1 1 c0t1d0s5

• detach the submirrors of the second disk

    metadetach d20 d22
    metadetach d10 d12
    metadetach d0 d2
    metadetach d30 d32

• patch or what ever you want and then boot and look if everything works
  • if not
  • set the pxe flag, run the workflow (feature dhcp)
  • boot into Miniroot via net/dhcp
  • change the submirrors to the second disk
• if everything is ok attach the submirros again

    metattach d20 d22
    metattach d10 d12
    metattach d0 d2
    metattach d30 d32

• check if resync is working

      metastat |grep sync
        State: Resyncing    
      Resync in progress: 17 % done
      State: Resyncing    
        State: Resyncing    
      Resync in progress: 22 % done
      State: Resyncing    
        State: Resyncing    
      Resync in progress: 13 % done
      State: Resyncing    

• check the status of the metadevice databases

    metadb -i

• boot this machine not before resyncing finished

Troubleshooting

• open a call auto-service@mcs.de

  • require a new disk

  • attach a new generated explore output, file generated by /var/site/explorer/runexplorer -d on the host, location: /var/site/explorer/

• Replacing or Reactivating the corresponding device example for one mirror set:

     Metadevices are not Okay:
  
  d40: Mirror
      Submirror 0: d41
        State: Needs maintenance 
      Submirror 1: d42
        State: Okay         
      Pass: 1
      Read option: roundrobin (default)
      Write option: parallel (default)
      Size: 82496832 blocks (39 GB)
  
  d41: Submirror of d40
      State: Needs maintenance 
      Invoke: metareplace d40 c0t0d0s5 <new device>
      Size: 82496832 blocks (39 GB)
      Stripe 0:
          Device     Start Block  Dbase        State Reloc Hot Spare
          c0t0d0s5          0     No     Maintenance   Yes 

  To do: Replace the disk c0t0d0 and run the following commands:

         metareplace -e d40  c0t0d0s5
         metareplace -e d30 c0t0d0s4
         metareplace -e d20 c0t0d0s3
         metareplace -e d10 c0t0d0s1
         metareplace -e d0 c0t0d0s0
     
         check the resync process with
         metastat | grep Resync

• if there are some splices in sync, run metadetach for the correspondig ones, see above

  • on x86 systems run cfgadm -c unconfigure ... before removal

    • on Sparc not necessary

  • on x86 systems run cfgadm -c configure ... for the new disk after insertion

    • on Sparc not necessary
  • run with the new disk to ensure that the partition tables of both disks are identical:

   # disk c0t0d0 running
   # disk c0t1d0 new
   prtvtoc /dev/rdsk/c0t0d0s2 | fmthard -s - /dev/rdsk/c0t1d0s2

  • check the state databases with metadb -i

• Problems with state databases

  •       Database replicas are not active:
            flags           first blk       block count
          Wm  p  l          16              8192            /dev/dsk/c0t0d0s7
          W   p  l          8208            8192            /dev/dsk/c0t0d0s7
          W   p  l          16400           8192            /dev/dsk/c0t0d0s7
         a    p  luo        16              8192            /dev/dsk/c0t1d0s7
         a    p  luo        8208            8192            /dev/dsk/c0t1d0s7
         a    p  luo        16400           8192            /dev/dsk/c0t1d0s7

    To do:

          metadb -i           # displays the status of the metadevice state database
          metadb -d c0t0d0s7  # deletes all replicas that are located on the specified slice.
          metadb -a -c 3  c0t0d0s7 # creates new ones
          metadb -i

    after a reboot what is not necessary the output is better, the master flag will be set again

        metadb -i
            flags           first blk       block count
         a m  p  luo        16              8192            /dev/dsk/c0t0d0s7
         a    p  luo        8208            8192            /dev/dsk/c0t0d0s7
         a    p  luo        16400           8192            /dev/dsk/c0t0d0s7
         a    p  luo        16              8192            /dev/dsk/c0t1d0s7
         a    p  luo        8208            8192            /dev/dsk/c0t1d0s7
         a    p  luo        16400           8192            /dev/dsk/c0t1d0s7