Messages in this thread | | | Date | Thu, 13 May 1999 13:56:35 -0400 (EDT) | From | Jeff Morris <> | Subject | Re: what is the "best" RAID 5 scsi controller? |
| |
On Mon, 10 May 1999 Leonard N. Zubkoff wrote:
> From: Paul Slootman <paul@wau.mis.ah.nl> > Date: Mon, 10 May 1999 16:09:58 +0200 > > The thing I'm wondering about is support for the specific RAID-5 > features from within the Linux environment. E.g., how is failure of a > drive reported (if at all)? I'd like a message in the syslog, at the > least... A utility for checking the status / configuring the RAID > subsystem would also be a must; I'd hate having to reboot into DOS or > even --horrors-- NT to do that. Even using the onboard BIOS firmware > isn't something I'd want. > > All the RAID controllers I've seen have pretty NT programs for > configuring and monitoring the RAID subsystem, but not for anything > else (except perhaps SCO Unix...)
Paul,
We're using the Mylex AcellaRAID 250 controller here in two Linux servers, with Leonard's driver. For our configuration we're running three drives in each, in removable drive caddies (the relatively inexpensive ones from CRU), as a RAID5 volume, with no hot standby.
I've been VERY happy with this setup so far. The hardware and driver both have worked flawlessly, and the /proc and syslog support is very good. In addition to errors getting logged, it's trivially easy to write a script to monitor the /proc status of the array and have it e-mail you, send a message to your terminal, or what have you in the event of a failure or change in status. We've actually integrated it in to Perl based, in-house network monitoring system (which took all of about 5 minutes), so in addition to our administrators getting an e-mail alert in the event of a drive failure, whoever is on-call at the time actually gets an alphanumeric page as well, all within less than a minute of the drive failure. To set up the volume initially, we did need to use the card's BIOS utility, but that's not a big deal; once set up, you never need to touch it again. To get status, conduct rebuilds or consistency checks, or do other routine tasks, just use the /proc interface.
We've tested the setup by powering down and replacing drives on a live and active system repeatedly, and again, never a hitch. As the volume is rebuilt, progress reports are even logged to syslog.
I've also written a quick and dirty "raid" script to interface with the /proc interface, so you can simply type things like "raid rebuild 0:1" on the command line, or just "raid" to see the current status. It's nothing fancy, but if you or anyone else is interested in it, you're welcome to use it.
We looked at quite a few ways of going to RAID 5 under Linux, and the AccelaRAID seemed like the winner in all respects. At only about $450 (a bit less than that for the lower powered AcellaRAID 150), it's an awesome deal.
And thanks Leonard for an excellent driver! :-)
Jeff S. Morris President and Chief Technical Officer, The Xensei Corporation
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |