Eric,
There were many VM issues all the way up to the current kernels IMHO.
However, the most recent stable kernel, 2.4.18, seems to work quite
well. I've got a couple of servers at work running 2.4.18 with no
problems. One of them with over 100 days of continuous uptime running on
XFS that's used as a samba/netatalk fileserver for 25+ clients, DNS,
DHCP and squid proxy server. So far, it's been solid.
The Linux VM was completely redone in 2.4.10 and took a few kernel
releases after that to really settle down. 2.4.18 seems to be the first
really solid 2.4 series kernel for me. Hope that helps,
-Walt
Eric Mei wrote:
> Thanks a lot!
>
> As you know, if we try to backport vm fixes to 2.4.5, which kernel have
> most probably addressed this issue? 2.4.7? There is no evidence about it
> from kenrel ChangeLog.
>
> Sorry to occupy XFS's bandwidth again :-)
>
>
> Eric Sandeen wrote:
>
>>On Sat, 2002-06-29 at 03:40, Eric Mei wrote:
>>
>>
>>>Hi Team,
>>>
>>>This is a long report. I know some of you might be busy on kernel
>>>summit, we just can's believe XFS has such a serious bug.
>>
>>
>>s/has/had/
>>
>>
>>
>>>For some reason, we must stick on 2.4.5-xfs-1.0.1.
>>
>>
>>You're essentially saying that you must stick with old, (apparently)
>>buggy code. We simply don't have the bandwidth to support xfs & kernel
>>code from 1 year ago.
>>
>>Your best bet would be to look through the mailing list archives to see
>>if anyone has had a similar problem, and see if a mod was checked in to
>>fix it; then try to backport that fix.
>>
>>http://oss.sgi.com/projects/xfs/mail_archive/200108/msg00307.html
>>
>>This thread makes it sound like it may well be an underlying kernel
>>problem with 2.4.5-era kernels.
>>
>>-Eric
>
>
|