Subject: Re: [vserver] Up-to-date stable version?
From: Herbert Poetzl <herbert@13thfloor.at>
Date: Tue, 11 Aug 2009 13:52:48 +0200

On Tue, Aug 11, 2009 at 01:13:46AM +0000, kyle.bader@gmail.com wrote:
> We still use 2.6.22.19 at our shop (grsec patched). I've had to
> hack the e1000 driver to work with a newer igb nic until we can get
> 2.6.30x working (w grsec). Right now 2.6.30 panics too often in our
> environment (5-10 minutes after boot). I don't remember who's I spoke
> to in the irc chann but they said w/o gres it should be _more_ stable.
> A patch for the grec was "in the works"

grsec is a constant source of kernel instability, if any
of the Linux-VServer patches leads to a kernel crash, that
is considered a severe bug and will be fixed immediately

note that even the experimental prereleases should be
stable enough not to crash your kernel ...

(of course, with grsec combined, it's a completely new
game, and the Linux-VServer developers are not really
involved anymore, i.e. contact Harry, who currently does
the merges, or the grsec developers, unless you can show
that the issue exists without grsec)

best,
Herbert


> ------Original Message------
> From: John A. Sullivan III
> To: vserver@list.linux-vserver.org
> ReplyTo: vserver@list.linux-vserver.org
> Sent: Aug 10, 2009 2:02 PM
> Subject: Re: [vserver] Up-to-date stable version?
> 
> On Mon, 2009-08-10 at 22:50 +0200, Claus Herwig wrote:
> > Hello,
> > 
> > as I've to compile a new kernel for one of my vserver boxes, it's time
> > to come back to this much loved question:
> > 
> > Which (vanilla) kernel and which vserver patch version should I use for
> > a production system?
> > 
> > I know this question isn't entirely new, but even after scanning through
> > the last months of mailing list posts I'm missing a simple answer ;-)
> > 
> > The website states 2.6.22.19 + vs2.2.0.7, which is, well, ancient (1.5
> > years old i.E.)
> > 
> > A second look at the website says 2.6.30.1 + vs2.3.0.36.14, which is
> > experimental, what (at least for me) means that I shouldn't use it on a
> > production system.
> > 
> > Then I read in some other post, that it is of course useable in
> > production - which just leads to another question: Why is it called
> > experimental?
> <snip>
> Funny, I was looking at this today as I need to rebuild one of our
> vserver hosts.  We've been using kernel 2.6.28 and 29.  We've had a few
> issues - a nasty kernel bug in both that occasionally requires a cold
> boot (!) and constant segfaults in OSSEC.  We've needed to be on those
> kernels because of even worse problems with iSCSI in older kernels and
> we've learned to avoid the hang problem.  It still scares the living
> daylights out of me and we are eagerly awaiting 2.6.30 which apparently
> fixes this bug.  However, when I mouse over the link to 2.6.30.1 +
> vs2.3.0.36.14, it still says pre3 so I am assuming it is not yet ready
> for even daring production. Is that true? Thanks - John
> -- 
> John A. Sullivan III
> Open Source Development Corporation
> +1 207-985-7880
> jsullivan@opensourcedevel.com
> 
> http://www.spiritualoutreach.com
> Making Christianity intelligible to secular society
> 
> 
> 
> Sent via BlackBerry from T-Mobile