Re: [Vserver] Wiki : HowTo graph vserver usage with cacti

From: Herbert Poetzl <herbert_at_13thfloor.at>
Date: Fri 16 Dec 2005 - 18:08:19 GMT
Message-ID: <20051216180819.GB6028@MAIL.13thfloor.at>

On Fri, Dec 16, 2005 at 03:56:56PM +0100, Grzegorz Nosek wrote:
> Hello,
>
> 2005/12/16, GarconDuMonde <gdm@fifthhorseman.net>:
> > me also! i really don't know that much about kernel stuff at all,
> > and it's not highest on my list of priorities to learn at the
> > moment, but i am keen to "monitor the hell out of the vservers"
> > (what a great phrase!), ultimately on a number of different physical
> > hosts. currently, the only monitoring tool i really have experience
> > with is munin, but cacti has been highly recommended to me, so
> > i'm very happy to learn it and then try to help get it as good as
> > possible for vserver usage.
>
> We're actually using an in-house monitoring tool (similar to munin or
> cacti but better suited to our needs) but the overall idea is very
> similar (get data via snmp and graph it).

in general, if there is real interest, and folks (at
least 3 parties) are volunteering to test and add the
required userspace tools/interfaces, please contact me
with a wish list (i.e. what kind of information you
would like to monitor) and we can probably get an
implementation done ... (of course, funding such
features migh be an alternative too :)

> > err, did you patch the kernel to collect the usage stats? as i said,
> > i know zero about kernel stuff, really, but would have thought that
> > hte more patches to the kernel, the more insecure; and that it would
> > be better to monitor from some other method.
>
> Yep, but I don't like the patch. Due to various implementations of
> filesystems (tested on ext2, ext3 and reiserfs) I couldn't use
> current->xid (the context of the current process), as the journaled
> filesystems perform the write with a kernel thread (kjournald and

please send me the patch (maybe again?) or point me
to the url where I can have a look at it ...

> pdflush IIRC). I got a bit lost in the VFS-block devices interface so
> I took the brute force of adding a xid field to struct page. The
> performance loss is/was unmeasurable but still I don't like it :)
>
> WRT your point that patching the kernel makes it insecure, IMHO it's
> somewhat true but it all depends on the patch quality. Would you say
> Openwall makes your box insecure? And AFAIK there's currently no way
> to distinguish disk I/O between vservers to collect stats from
> userspace.

latest devel releases support per context CFQ queues,
so that might get a little easier there :)

best,
Herbert

> Best regards,
> Grzegorz Nosek
> _______________________________________________
> Vserver mailing list
> Vserver@list.linux-vserver.org
> http://list.linux-vserver.org/mailman/listinfo/vserver
_______________________________________________
Vserver mailing list
Vserver@list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver
Received on Fri Dec 16 18:08:39 2005

[Next/Previous Months] [Main vserver Project Homepage] [Howto Subscribe/Unsubscribe] [Paul Sladen's vserver stuff]
Generated on Fri 16 Dec 2005 - 18:08:43 GMT by hypermail 2.1.8