From: Herbert Poetzl (herbert_at_13thfloor.at)
Date: Thu 07 Nov 2002 - 16:07:47 GMT
On Thu, Nov 07, 2002 at 06:45:16AM +0000, Nuno Silva wrote:
> Hi,
>
> 2.4.19 died on me twice(!!!!!) and 2.4.18 has a very nice "history".
> My advice is: "upgrade" to 2.4.18 :)
> ..or you can try the 2.4.20preX branch with X > 10.
2.4.18 and 2.4.19 both have bad karma, just search for
kernel, 2.4.18, 2.4.19, ext3, filesystem, and corruption ...
On Thu, Nov 07, 2002 at 04:30:33PM +0100, Jon Bendtsen wrote:
> Lars Braeuer wrote:
>
> [cut]
>
> > another thing to consider: could it be related to the software raid?
> > since at least
> > two of you complaining about the stabiltiy of the vservers are using
> > software raids?
>
> 3, because i'm using software raid as well...
I would not draw too many conclusions from that
as well, because if you take a closer look, you'll
probably find a dozen of things these systems have
in common ...
- using IDE drives?
- using ext2/ext3?
- patched with ctx-NN ;)
- SMP /non SMP?
also check for features introduced/included/selected
in the kernels ...
the minimum you would need to make some minimum
cross relation would be
- all kernel configurations
- the distributions/configurations
- I/O modules active (MD, LVM, IDE, SCSI, ...)
- memory/harddisk configurations
Nevertheless I am sure the information is there,
and if it can be combined/evaluated maybe one
or the other cause/bug could be found ...
(although I doubt, that this will become reality,
because nobody would give out access to all his
configuration/logging information ...)
best,
Herbert
PS: SMP machines perform _MUCH_ better as physical
vserver hosts than non-SMP machines. I had otherwise
identical machines in production, and the SMP machine
was more than 10 times faster, regarding the response
time to connections/logins/etc ...