Re: ATA Raid

From: Matthias Saou (thias_at_spam.spam.spam.spam.spam.spam.spam.egg.and.spam.freshrpms.net)
Date: 02/04/04

  • Next message: Charles Howse: "Empty email from emiliano.brunetti@fastwebnet.it"
    To: fedora-list@redhat.com
    Date: Wed, 4 Feb 2004 15:55:47 +0100
    
    

    Bill Gradwohl wrote :

    > Has anyone tested an ATA Raid 1 setup that utilizes the hptraid.o,
    > pdcraid.o, or silraid.o modules?
    >
    > By testing I mean force a failure and see what happens.

    Yep, I have. Failures were not intentional, though.

    > I spun up a copy of Fedora on a TYAN dual P3 mobo that has a Promise
    > 20267 chip, applied all the recommended patches via up2date, and insmod'd
    > ataraid and pdcraid, formatted and then mounted a partition for testing.
    >
    > I was surprised to find that the Promise controller allowed the O/S to
    > find hde and hdf even though I used both identical drives to create a
    > single RAID1 array via the Promise BIOS. My Google research suggests that
    > all these mobo installed RAID controllers don't provide real hardware
    > RAID, but offer a quasi (and I think unreliable) hardware/software
    > solution. Is that true?

    Possibly. I've installed RHL 7.3 on a bunch of Intel 1U servers which have
    those Promise FastTrack ATA RAID controllers integrated, and had nothing
    but trouble.
    After system crashes (those servers are incredible unreliable for reasons
    I've been unable to track down), the arrays on some servers became
    completely hosed, and all attempts to get things back to normal failed. I
    went as far as going into single user mode and dd'ing directly one drive to
    the other using /dev/hdX directly. I knew it was something to avoid, but
    that was my last resort and still didn't get things working again.

    BTW, I'm still running 2.4.18-x on those servers as I've experienced severe
    and systematic data corruption with later 2.4.20-x errata kernels for RHL
    7.3. I've still got a bugzilla entry open for that IIRC, and don't know if
    later 2.4 kernels from FC1 or 2.6 are any better since I've never had the
    chance to test any on that hardware.

    My advice is clearly to stay away from all that ATA RAID cr*p when there's
    "Promise" written on it. Either just use the second disk for regular or
    occasional backups, or use software raid, which is much more reliable in my
    experience and very easy to set up and monitor, especially since mdadm came
    along.

    Matthias

    -- 
    Clean custom Red Hat Linux rpm packages : http://freshrpms.net/
    Fedora Core release 1 (Yarrow) - Linux kernel 2.6.1-1.61
    Load : 0.14 0.17 0.11
    -- 
    fedora-list mailing list
    fedora-list@redhat.com
    To unsubscribe: http://www.redhat.com/mailman/listinfo/fedora-list
    

  • Next message: Charles Howse: "Empty email from emiliano.brunetti@fastwebnet.it"

    Relevant Pages

    • Re: Both sides of Mirrored Drives showing as separate drives
      ... As to your last comment regarding the hardware RAID, ... data if you break/recreate the RAID array..... ... > Basically, these servers have the OS on Mirrored/RAID1 disks, and BOTH ... Looking at the Disk Management MMC, it correctly shows the mirror, ...
      (microsoft.public.windows.server.general)
    • RE: BugCheck and Supposed Reboot
      ... Windows Updates are NOT installed. ... To answer your RAID question - No IBM hardware, ... >> software changes to those Servers and they have been in production for over ...
      (microsoft.public.windows.server.sbs)
    • Re: hardware raid vs. software raid
      ... hardware raid, I suppose I would have to wait for replacement parts. ... I take periodic snapshots of servers when I get a chance to restart them ... To UNSUBSCRIBE, email to debian-user-REQUEST@xxxxxxxxxxxxxxxx ...
      (Debian-User)
    • Re: Advice: Hardware vs. Software RAID5
      ... I've ordered four 350G SATA drives and I plan to put them into some ... driver that will let me do RAID 5 in the bios. ... survive various forms of hardware failure because the hardware is cheap and ... your data cannot survive the failure of the controller. ...
      (Debian-User)
    • Re: Switch from SATA-RAID to gmirror?
      ... HP-UX boxes running on K-series hardware. ... the case of system failure the recovery procedures would work. ... Fortunately, we've just bought redundant servers for everything. ... Currently we're in a situation where a guy's whole business depends on a single-disk webserver running Postgres, which desperately needs upgrading for performance tuning, but which we just can't do. ...
      (freebsd-questions)