[Users] How to kill dead container (init is dead)

[Andrey Mirkin]

Hello,

Can you please tell what vzdump tool is doing. It is not standard tool from 
our packages.
If this tool perform checkpointing of your containers, then some error can 
occur during checkpointing and container can be left in frozen state.
Actually info that you have provided to us points that container is in frozen 
state (all processes in D state (it is stopped state) and they do not react 
on signals).
Can you please post here sources of vzdump utility or provide a link where we 
can find it.

Regards,
Andrey

On Thursday 31 July 2008 10:19 Pongracz Istvan wrote:
> Hi,
>
> Thank you for your response!
> Unfortunately I had te reboot last night.
>
> That container had the following processes:
> init(D)
>
>   |---------- sshd(D)
>   |---------- svnserve(D)------svnserve(D)
>   |---------- syslog-ng(D)
>   |---------- cron(D)
>
> This container was unused, I did not use it for months.
>
> Normally I make a daily routine from cron of HN:
> - glsa-check: vzctl exec $i glsa-check -t all 2>&1
> - rkhunter from HN: /usr/bin/rkhunter --nocolors --cronjob  -c
> --report-warnings-only --summary --configfile $i
> - and daily backup:
>    vzdump --exclude 10014 --exclude 107  --exclude 10222 --exclude 10016
> --exclude 10019 --exclude 10020 --exclude 10021 --exclude 10030
> --suspend --dumpdir $BASE$SUB  --compress --mailto
> pongracz.istvan at osbusiness.hu -all
>
> In fact, I have no idea, what happened. The system log of this container
> contains nothing, even the HN has nothing unusual.
>
> The cron (or syslog-ng) stopped working, when the vzdump backup started
> (last message in the system log).
> Nagios detected that, svn stopped after ~8 hours from the daily backup.
> When I tried to enter or connect via sshd the afternoon, the container
> was dead.
>
> Sorry, I know, this was not really useful.
>
> Thank you for your time.
>
> Cheers,
> István
>
> 2008. 07. 31, csütörtök keltezéssel 05.03-kor Andrey Mirkin ezt írta:
> > Hello,
> >
> > Can you please post here 'ps axf' from your system.
> > Also please provide what steps were performed before container get stuck
> > in such a state.
> >
> > Regards,
> > Andrey
> >
> > On Thursday 31 July 2008 00:16 Pongracz Istvan wrote:
> > > Hi,
> > >
> > > I use openvz kernel 2.6.18-028stab051 for long months on my gentoo
> > > system.
> > > The uptime now is 105 days.
> > >
> > > It seems, one of my containers completely dead:
> > > all processes are dead, including the init process.
> > >
> > > I tried to kill them by issuing kill -9, but it is not working.
> > > vzctl also cannot stop the container.
> > >
> > > I tried to send other signal to these processes and the cron started.
> > > The only process, which is run, but not really useful:
> > > it is eating cpu.
> > >
> > > The last message to the system log happened before the daily vzdump
> > > started.
> > > Since then, there is nothing in the syslog of the container.
> > >
> > >
> > > So, my 1st priority question, is there any other trick to restart a
> > > container or I have to reboot?
> > >
> > > Cheers,
> > > István