Summary: ASE with Raid 5 Configuration

From: Padiyath Kumar <Kumar.Padiyath_at_psi.ch>
Date: Sun, 24 Sep 2000 10:39:31 +0200

  I got only one answer from Mr.Alan from Compaq and many many thanks. Here
is his answer:

          "I don't know what, if any, coordination has be performed
        with ASE on configuration changes. It probably needs to
        know what units are shared among the systems, so there
        may be something that needs to be done. If it keeps such
        units open, changing the size out from under could at
        best be confusing and at worst may cause serious problems.
        The ASE documentation will probably describe what needs
        to be done in general terms. It doesn't know the arrays
        presented by an HSZ family controller from a bare SCSI
        disk of the same size. Array reconfiguration is done
        with SWCC, hszterm or on the array controller console.

        First, see if the disk is labeled:

                # disklabel /dev/rrz27c

        If not, or if all the partitions are unused, it may be
        safe to do the controller changes without worrying about
        what ASE wants. But, do check the ASE documentation.

        Once the unit is free of ASE's influence, the steps you
        want are:

        1. Verify there's no data on the device. If there is
            make a backup just in case.

        2. Delete the unit.

                DELETE D300

        3. Delete the associated storage set. You'll have to
            get the name from the "SHOW STORAGESET" command.
            It should be the one that doesn't have a unit
            allocated with it.

                DELETE "storage-set-name"

        4. Delete the member devices.

                DELETE "device-name"
                DELETE "device-name"
                DELETE "device-name"
                etc...

        5. Remove the old devices and put the new ones in.

        6. You can add the new disks by hand using "ADD DISK",
            but I typically trust "RUN CONFIG" to handle that
            for me.

        7. Create the desired Storage set:

                ADD STRIPE "stripe-set-name" ...

            or:

                ADD RAID "raid-set-name" ...

        8. Initialize the storage set.

                INIT "storage-set-name"

            You may want to consult the configuration documentation
            for the controller to see if there's much value in picking
            a chunk size other than the default. Depending on the
            intended I/O load, benchmarks of various sizes may be
            wise.

        9. Create a unit associated with the storage set:

                ADD UNIT D300 "storage-set-name" ...

        10. On the host side, use scu(8) to rescan the particular
            SCSI bus and verify that the new unit is the right
            size.

        11. Label it with the disklabel command:

                # disklabel -wr /dev/rrz27c whatever

        After that it becomes an ASE problem again.

        Unless you make a mistake, the only unit affect by the
        reconfiguration will be the one deleted.

        re: Question #2.

        You should be doing regular backups of all the file systems.
        RAID-1 and RAID-5 protect against device failures, not user
        mistakes. Neither RAID level will protect against catastrophic
        failures. That's what backups are for.

        From an AdvFS point of view, if you want to swap devices
        the simplest way is to add the new device with the addvol
        command and then remove the old device with rmvol. AdvFS
        will migrate the data. Again, I don't know what coordination
        needs to be done with ASE."
Received on Sun Sep 24 2000 - 08:41:35 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:41 NZDT