> The whole redhat 2.4.20 kernel is crap. I've had a number of boxes lockup
> on it. Initially it was the firewall, but then at home I've had other
> boxes lock up under load with it. 2.4.20-19.x got a little better but was
> still totally screwed up. Going to a stock 2.4.20 fixed all the issues
> I've been having.
I have my own patched set of 2.4.20-19.9.SGI_XFS_1.2.0 kernels which
perform very well on several servers from RedHat 7.2 to RedHat 9. One
RedHat 9 box is a dual Xeon 2.8G, 4G ram and lot's of disks, running two
quite big, heavily loaded SAP-DB instances on XFS, on top of LVM. Others
are running Cyrus-IMAPd mailservers, Samba, whatever. If the kernels were
crap, I knew it for sure. Where did you get your kernels from?
> The root of this one is IMHO RH 2.4.20 is actually 2.4.21-pre3 plus a mass
> of patches.
> --On Wednesday, August 06, 2003 21:15 +0200 Simon Matter
> <simon.matter@xxxxxxxxxxxxxxxx> wrote:
>>> Yah for production go with a kernel as close to stock as posible.
>>> anything by redhat in the 2.4.20 series, and you should be fine. In
>>> other words go get a fresh treee, patch in ONLY the XFS stuff. Don't
>>> put in low latency/preeempt patches, they still screw everything up.
>>> And just stay away from RedHate kernels.
>> I don't agree. I'm using XFS enabled RedHat for a long time now on
>> servers with great success. Are the problem you're talking about always
>> NFS related? That's the only thing I'm not using alot with the XFS