Subject: Re: [vserver] Oddity with df
From: Phil Daws <uxbod@splatnix.net>
Date: Wed, 16 May 2012 13:41:53 +0100 (BST)

does anybody else see this with a CentOS 6 guest ?
-- 
Thanks, Phil

----- Original Message -----
> The sysinit script is a core CentOS one not our own. I wonder why
> hdv1 was not in proc mounts.
> 
> Sent from my Android phone using TouchDown (www.nitrodesk.com)
> 
> 
> -----Original Message-----
> From: Herbert Poetzl [herbert@13thfloor.at]
> Received: Tuesday, 15 May 2012, 9:28pm
> To: Phil Daws [uxbod@splatnix.net]
> CC: vserver@list.linux-vserver.org
> Subject: Re: [vserver] Oddity with df
> 
> 
> On Tue, May 15, 2012 at 08:53:41PM +0100, Phil Daws wrote:
> > Well under CentOS6 it looks like /etc/rc.sysinit wipes out
> > /etc/mtab and re-populates it with data from /proc/mounts. Any
> > thoughts on the best way to handle this so that:
> 
> data from /proc/mounts should be fine with a recent kernel
> 
> > /dev/hdv1 / ext2 rw 0 0
> 
> > would get pulled in ?
> 
> > I have tried creating a guest mtab under
> > /etc/vservers/<guest>/apps/init and by adding straight into
> > fstab but no joy :(
> 
> as you already stated/concluded, util-vserver populates
> /etc/mtab correctly, which is then removed by your sysinit
> script (which is probably useless anyway). at this point
> util-vserver is not involved in the guest startup anymore.
> of course, you can start a time delayed script or fix the
> rc.sysinit to do the right thing ...
> 
> best,
> Herbert
> 
> > --
> > Thanks, Phil
> 
> > ----- Original Message -----
> >> Hmmm, the plot thickens.  Out of curiosity I shut down the guest
> >> and
> >> removed the init file /etc/vservers/<guest>/apps/init/style which
> >> contained the text plain.  Now when I start the guest it complains
> >> about upstart issues:
> 
> >> initctl: Unable to connect to Upstart: Failed to connect to socket
> >> /com/ubuntu/upstart: Connection refused
> >> initctl: Unable to connect to Upstart: Failed to connect to socket
> >> /com/ubuntu/upstart: Connection refused
> 
> >> but when I check /etc/mtab the hdv is indeed there.  When you
> >> specify
> >> plain in the style file how does that alter the boot process as it
> >> appears to be wiping out what is in /etc/mtab ?
> >> --
> >> Thanks, Phil
> 
> >> ----- Original Message -----
> >>> Hi Herbert,
> 
> >>> Agreed. I shutdown the guest and then ran a simple script which
> >>> cat
> >>> /vservers/guest/etc/mtab into another file. I then started the
> >>> guest
> >>> and checked the log file for hdv1 and saw:
> 
> >>> /dev/hdv1 / ufs defaults 0 0
> >>> none /proc proc defaults 0 0
> >>> none /tmp tmpfs size=128m,mode=1777,nosuid,nodev,noexec 0 0
> >>> none /dev/pts devpts gid=5,mode=620 0 0
> 
> >>> then I bit further down it changed too:
> 
> >>> none / none rw 0 0
> 
> >>> Oh this is going to be a real pain to track down as I do not see
> >>> anything in /etc/init.d or the upstart configuration files :(
> >>> --
> >>> Thanks, Phil
> 
> >>> ----- Original Message -----
> >>>> On Tue, May 15, 2012 at 07:37:34PM +0100, Phil Daws wrote:
> >>>>> Hi Herbert,
> 
> >>>>> When you mentioned about the debug option I did look through it
> >>>>> and checked that file:
> 
> >>>>> [root@testvserver01 ~]# cat /lib/util-vserver/defaults/mtab
> >>>>> /dev/hdv1 / ufs defaults 0 0
> 
> >>>>> I also checked the other ones in the findObject call and they
> >>>>> did not exist. Strange eh ?
> 
> >>>> not at all, just means that my original assumption is correct
> >>>> and something _inside_ your guest removes the file on startup
> 
> >>>> HTH,
> >>>> Herbert
> 
> >>>>> --
> >>>>> Thanks, Phil
> 
> >>>>> ----- Original Message -----
> >>>>>> On Tue, May 15, 2012 at 06:48:47PM +0100, Phil Daws wrote:
> >>>>>>> Hi Herbert,
> 
> >>>>>>> Here is the debug file.
> 
> >>>>>> so, what does your /lib/util-vserver/defaults/mtab contain?
> 
> >>>>>> best,
> >>>>>> Herbert
> 
> >>>>>>> --
> >>>>>>> Thanks, Phil
> 
> >>>>>>> ----- Original Message -----
> >>>>>>>> On Tue, May 15, 2012 at 06:03:34PM +0100, Phil Daws wrote:
> >>>>>>>>> Hi Herbert,
> 
> >>>>>>>>> Not that I am aware of. Using the scientific image I built
> >>>>>>>>> a
> >>>>>>>>> CentOS6 one and changed the repo URLs. That is all which
> >>>>>>>>> has
> >>>>>>>>> been done.
> 
> >>>>>>>> well, the guest's mtab is written by util-vserver, so the
> >>>>>>>> version would be interesting, also mtab can be overridden
> >>>>>>>> by guest config entries (default and guest specific) which
> >>>>>>>> might be the case here as well, finally a --debug guest
> >>>>>>>> start might shed some light on this issue ...
> 
> >>>>>>>> best,
> >>>>>>>> Herbert
> 
> >>>>>>>>> --
> >>>>>>>>> Thanks, Phil
> 
> >>>>>>>>> ----- Original Message -----
> >>>>>>>>>> On Tue, May 15, 2012 at 05:51:13PM +0100, Phil Daws wrote:
> >>>>>>>>>>> Hmmm, trawling around it would appear I should see
> >>>>>>>>>>> something
> >>>>>>>>>>> like /dev/hdv1 / ext2 rw 0 0 in /etc/mtab. If I check and
> >>>>>>>>>>> older
> >>>>>>>>>>> vserver system I do see that but not on the latest
> >>>>>>>>>>> release.
> >>>>>>>>>>> Has something changed in util-vserver where that does not
> >>>>>>>>>>> automatically get added now ?
> 
> >>>>>>>>>> IIRC, newer systems switched to /etc/mtab being a link
> >>>>>>>>>> to /proc/mounts (which should work fine) so maybe your
> >>>>>>>>>> setup removes/clears out /etc/mtab inside the guest?
> 
> >>>>>>>>>>> --
> >>>>>>>>>>> Thanks, Phil
> 
> >>>>>>>>>>> ----- Original Message -----
> 
> >>>>>>>>>>>> Hello,
> 
> >>>>>>>>>>>> within our test lab I am trying out 3.2.15-vs2.3.2.8 and
> >>>>>>>>>>>> a
> >>>>>>>>>>>> few CentOS6 clients. Now that the upstart issues have
> >>>>>>>>>>>> been
> >>>>>>>>>>>> resolved all are looking pretty good. One oddity I have
> >>>>>>>>>>>> noticed is that when I perform a 'df' from within a
> >>>>>>>>>>>> guest
> >>>>>>>>>>>> the
> >>>>>>>>>>>> response is:
> 
> >>>>>>>>>>>> df: no file systems processed
> 
> >>>>>>>>>>>> is there a new capability flag I am missing ?
> 
> >>>>>>>>>> not that I know of ...
> 
> >>>>>>>>>> best,
> >>>>>>>>>> Herbert
> 
> >>>>>>>>>>>> --
> >>>>>>>>>>>> Thanks, Phil
> 
> 
>