10 Replies Latest reply: Apr 23, 2013 1:11 AM by Artem Danielov RSS

    forcing DG mount during ASM startup

    Artem Danielov
      Have been scratching my head for a couple of days. May be someone can help?

      Test setup:
      RHEL 6.2
      GI : 11.2.0.1
      RAC: 11.2.0.1
      3 nodes
      One ASM diskgroup with Normal redundancy and three failure groups, containing voting disks,OCR, and data


      ASM startup (and as a result the whole cluster startup) fails if one of the failure groups is missing. That's understandable.

      But how can I force mount of the diskgroup during ASM/cluster startup, so it ignores the missing failure group?
        • 1. Re: forcing DG mount during ASM startup
          ACR80
          I suppose you simply can't.

          As per [http://docs.oracle.com/cd/E11882_01/server.112/e18951/asmdiskgrps.htm#CHDDBHCH]

          "A normal redundancy disk group must contain at least two failure groups. A high redundancy disk group must contain at least three failure groups. However, Oracle recommends using several failure groups. A small number of failure groups, or failure groups of uneven capacity, can create allocation problems that prevent full use of all of the available storage."

          You have to fix the DG and than startup.

          Regs,
          Acr
          • 2. Re: forcing DG mount during ASM startup
            Mihael
            One ASM diskgroup with Normal redundancy and three failure groups, containing voting disks,OCR, and data
            ASM startup (and as a result the whole cluster startup) fails if one of the failure groups is missing. That's understandable.
            If you have only three failure groups for normal redundancy, then disk (or disks) in one of failure group become unprotected. If this disk is lost, ASM can't recover data.
            • 3. Re: forcing DG mount during ASM startup
              Catch-22
              Can you explain why? I would agree if ASM was mirroring disks, similar to RAID, but it is not.

              ASM uses the free space of disk failure groups for redundancy. With normal redundancy, and 2 failure groups or more, any one failure group should be able to fail since any data, regardless of failure group will at least have one more duplicate in another failure group.
              • 4. Re: forcing DG mount during ASM startup
                Artem Danielov
                I understand that the DG is unprotected if in Normal Redundancy mode one FG is failed. However, when ASM is already running with the DG mounted, it will not stop just because a FG fails. So, it would also make sense to be able to start ASM and mount the DG in the unprotected (degraded) mode when the DG cannot be quickly fixed.

                Anyone has any other thoughts how to force ASM to start with one failed FG?
                • 5. Re: forcing DG mount during ASM startup
                  Hemant K Chitale
                  I don't think that you can FORCE during the startup.

                  You'd have to issue a separate
                  ALTER DISKGROUP dgname MOUNT FORCE;
                  command.


                  Hemant K Chitale
                  • 6. Re: forcing DG mount during ASM startup
                    Artem Danielov
                    Let's assume it's not possible to start ASM when there is a failed FG in Normal Redundancy mode.
                    Then, how does it work in Exadata?

                    I'm assuming that OCR and Voting disks are typically mirrored across cells in Exadata. Is this correct?
                    If so, what happens if one cell is dead? Can I still start/restart any RAC node that used the failed cell? Or do I have to fix the cell first?
                    • 7. Re: forcing DG mount during ASM startup
                      Mihael
                      Yes, you are correct.
                      By the way, can you explain why ASM does not have such feature as hot spare that enables automatic repair of failed disks.
                      • 8. Re: forcing DG mount during ASM startup
                        Mihael
                        I didn't found how mount options can be set in OCR. But I have tested that if diskgroup with failed disk is manually mounted with 'force' option, then after reboot the diskgroup will be mounted automatically.
                        • 9. Re: forcing DG mount during ASM startup
                          Catch-22
                          ASM is not a software RAID and does not duplicate disks. If you have only two disks with normal redundancy the effect is probably similar to RAID, but ASM works on the concept of free disk space in failure groups and file extents.
                          • 10. Re: forcing DG mount during ASM startup
                            Artem Danielov
                            So, what's the consensus? If one cell in Exadata is dead then it is impossible to start or restart any node? Is this true?
                            Don't have an Exadata to try on, unfortunately. Trying to understand how to make it work in a non-Exadata ASM environment.