On Saturday 28 February 2004 05:36, Keith Roberts wrote:

On Fri, 27 Feb 2004, John Andersen wrote:

...

To: suse-security@suse.com From: John Andersen <jsa@pen.homeip.net> Subject: [suse-security] Kernal patch

That patch was auto-applied by YOU on one of my production servers (a dual PII with a software raid on a dual IDE card and which boots from a SCSI drive (aic7xxx).

It was NOT safe. It would NOT boot, it kept getting dma timeouts and lost interrupts on one of the ide drives (which were part of the raid 1 array, not the boot drive).

I had terrible problems with Linux Software RAID1 using the md kernel driver, under 8.1.

I had 2 WD800JB drives as a RAID1 array on hdb & hdc.

hda was another WD800JB with the SuSE 8.1 system on it.

hdd was the CD-ROM drive.

I used to get funny dma timeout messages in /var/log/boot.msg.

Seek error ...

This only happened on the WD800JB connected to hdc.

The other two WD drives did not get this.

Sorry to hear about this Keith. In my case, my software raid 1 has been rock solid stable since SuSE 7.2, thru two upgrades. My only trick has been quality drives, and installing the system on other disks, just useing the raid for company business data. I've often contemplated going to one of those ide raid cards so that Linux sees it as a single device, but have held off because 1) I have the processing power to handle it as the CPU(s) are never busy, and 2) IINBDFI (if its not broke don't fix it), and 3) as my CUP power grows with new platforms the raid performance grows as well where-as the hardware raid remains at the speed it was born with. This same raid has been running for 4 years. But again, the topic of the thread was a bad kernel update not bad hardware. Moving back to the prior kernel has FIXED everything. -- _____________________________________ John Andersen