From: Herbert P÷tzl (herbert_at_13thfloor.at)
Date: Tue 26 Aug 2003 - 16:29:11 BST
On Wed, Aug 27, 2003 at 12:07:31AM +1000, Lucian Daniel Kafka wrote:
> At 03:05 PM 26/08/2003 +0200, you wrote:
> >On Tue, Aug 26, 2003 at 09:06:40PM +1000, Lucian Daniel Kafka wrote:
> >> Hi.
> >> I have looked around but couldn't find any reference to a clean way of
> >> deleting a vserver (and all files & setup associated with it). Any hints?
> >> Also, are the latest 2.4.22-ctx17a and vserver23 considered
> >> production-level stable?
> >please define 'production-level stable' ...
> Herbert, I guess I am after some assurance that if we start using
> vservers in our hosting environment, the servers won't crash,
> lock, or do anything unexpected.
can't never ever give you this kind of assurance ;)
everything over 10-20 lines of code, can't be proven
logically correct, so this is wishful thinking ...
windows crashes at least twice a day, so vserver with
100 - 150 days uptime are pretty stable, what do you think?
> I have tried the 21-ctx17 + vserver22 on a test server and it
> worked ok.
worked okay, meaning you had no crash in 5 hours 8-)
> I would like now to convert our hosting servers to vserver-enabled
> platforms to make use of the nice inherent features.
sounds good to me ...
> Should I do it, or the code is not mature enough to risk it?
I would do it ... vserver patches are pretty stable now,
unless you want to use bleeding edge features/enhancements ...
> Also, I have trailed today the 22-ctx17a + vserver23 and it didn't
> really worked.
guess you experienced some of the changes and/or features
introduced recently ... but they are mere cosmetic warnings
and/or kernel misbehaviour not vserver problems ...
> When I ran newserver it didn't work, and came back with errors
hmm, I guess newserver isn't used very often/widespread,
because, usually you create the vservers from templates (YMMV)
> saying that there is no chrootsafe kernel support, and then some
> operations denied.
the chrootsafe() message is a compatibility fallback, because
the author decided to publish tools (0.23) which use/provide
some features, not (yet) supported by the patches ...
and if you've seen ulimit() messages, those are the result
of recent changes in kernel behaviour, but there is a fix
2.4.22-rcX with my ctx17a patches seemed very stable to me ...
I guess 2.4.22 can be considered stable with ctx17a ...
> Kind regards,
> Lucian Kafka