Subject: Re: [vserver] 4.9 Kernels Hit Psand Repos
From: Daniel Urist <durist@ucar.edu>
Date: Thu, 8 Feb 2018 15:42:36 -0700
Thu, 8 Feb 2018 15:42:36 -0700
I don't seem to be able to start vservers. I get the following error with a
newly built stretch vserver:

# vserver -v vsrvdev1 start
> The following problem(s) were encountered while verifying vshelper
> functionality:
> * The vshelper state-directory '/var/run/vshelper' does not exist; since
> it is created by 'make install', this indicates a serious problem with
> your util-vserver installation
>
> To fix this, you can:
> * disable vshelper entirely by executing
>   | touch "/etc/vservers/.defaults/apps/vshelper/disabled"
> * disable only this message by executing
>   | touch "/etc/vservers/.defaults/apps/vshelper/warning-disabled"
> * configure the util-vserver vshelper script, e.g. by adding
>   | kernel.vshelper = /usr/lib/x86_64-linux-gnu/util-vserver/vshelper
>   to /etc/sysctl.conf and rebooting the machine, or by executing
>   | echo "/usr/lib/x86_64-linux-gnu/util-vserver/vshelper"
> >/proc/sys/kernel/vshelper


If I disable vshelper, I get the following when I try to start the vserver:

# vserver -v vsrvdev1 start
> /proc/uptime can not be accessed. Usually, this is caused by
> procfs-security. Please read the FAQ for more details
> http://linux-vserver.org/Proc-Security
>
> Failed to start vserver 'vsrvdev1'


My kernel version is 4.9.78-vs2.3.9.6-beng-reptoline

I have all the util-vserver systemd packages installed:

# dpkg -l | grep util-vserver
> ii  libvserver0
>  0.30.216-pre3126-stretch0.1-1  amd64        dynamic libraries for
> util-vserver
> ii  util-vserver-systemd                          0.30.216-pre3126-1.3
>        amd64        utilities for managing Linux-VServer guests
> ii  util-vserver-systemd-build                    0.30.216-pre3126-1.3
>        amd64        tools which can be used to build vservers
> ii  util-vserver-systemd-build-dbgsym             0.30.216-pre3126-1.3
>        amd64        Debug symbols for util-vserver-systemd-build
> ii  util-vserver-systemd-core                     0.30.216-pre3126-1.3
>        amd64        core utilities of util-vserver-systemd
> ii  util-vserver-systemd-core-dbgsym              0.30.216-pre3126-1.3
>        amd64        Debug symbols for util-vserver-systemd-core
> ii  util-vserver-systemd-dbgsym                   0.30.216-pre3126-1.3
>        amd64        Debug symbols for util-vserver-systemd
> ii  util-vserver-systemd-legacy                   0.30.216-pre3126-1.3
>        amd64        legacy scripts for util-vserver-systemd
> ii  util-vserver-systemd-legacy-dbgsym            0.30.216-pre3126-1.3
>        amd64        Debug symbols for util-vserver-systemd-legacy
> ii  util-vserver-systemd-python                   0.30.216-pre3126-1.3
>        amd64        python-bindings for util-vserver-systemd
> ii  util-vserver-systemd-python-dbgsym            0.30.216-pre3126-1.3
>        amd64        Debug symbols for util-vserver-systemd-python
> ii  util-vserver-systemd-sysv                     0.30.216-pre3126-1.3
>        amd64        initscripts for util-vserver-systemd




On Mon, Feb 5, 2018 at 6:54 AM, Tor Rune Skoglund <trs@swi.no> wrote:

> Hi,
>
> did not notice any such issue --- on the 4.9.76 at least. Will be
> installing the 4.9.78 now on at least two test servers today, and will
> report to the ML if we see any issues like the one you mentioned.
>
> BR,
> Tor Rune Skoglund
>
> Den 05. feb. 2018 09:47, skrev Andrew Ruthven:
> > Hey,
> >
> > Has anyone noticed big I/O issues with this kernel?
> >
> > I ran it for a few days, and found that processes (even on the main
> > system) were often sitting in a D state, seemingly blocked on I/O. But
> > there didn't seem to be much I/O happening. Even ssh'ing in to any
> > vservers or physical host was incredibly slow.
> >
> > Cheers,
> > Andrew
> >
> > On Tue, 2018-01-30 at 22:50 +0000, Ben Green wrote:
> >> Hi all,
> >>
> >> I've just finished and uploaded Linux kernel 4.9.78 with patch
> >> vs2.3.9.6. They are installable directly from the psand.net
> >> repository, or via the metapackage:
> >>
> >> linux-image-vserver-4.9-beng
> >>
> >> For i386 and amd64 architectures, and there's a version I've
> >> compiled
> >> with gcc-8 which can be installed directly, or via the metapackage:
> >>
> >> linux-image-vserver-4.9-beng-reptoline
> >>
> >> That one is amd64 only of course. These images are only provided in
> >> the Debian Stretch distro repositories. Instruction (which need a
> >> bit
> >> of updating but should get you where you need to be) are at:
> >>
> >> https://www.psand.net/services/public-repository.html
> >>
> >> These kernels are minimally tested and as usual come with
> >> absolutely
> >> no guarantees whatsoever.
> >>
> >> Cheers,
> >> Ben
> >>
>
>


I don't seem to be able to start vservers. I get the following error with a newly built stretch vserver:

# vserver -v vsrvdev1 start
The following problem(s) were encountered while verifying vshelper
functionality:
* The vshelper state-directory '/var/run/vshelper' does not exist; since
it is created by 'make install', this indicates a serious problem with
your util-vserver installation

To fix this, you can:
* disable vshelper entirely by executing
  | touch "/etc/vservers/.defaults/apps/vshelper/disabled"
* disable only this message by executing
  | touch "/etc/vservers/.defaults/apps/vshelper/warning-disabled"
* configure the util-vserver vshelper script, e.g. by adding
  | kernel.vshelper = /usr/lib/x86 64-linux-gnu/util-vserver/vshelper
  to /etc/sysctl.conf and rebooting the machine, or by executing
  | echo "/usr/lib/x86 64-linux-gnu/util-vserver/vshelper" >/proc/sys/kernel/vshelper

If I disable vshelper, I get the following when I try to start the vserver:

# vserver -v vsrvdev1 start
/proc/uptime can not be accessed. Usually, this is caused by
procfs-security. Please read the FAQ for more details
http://linux-vserver.org/Proc-Security

Failed to start vserver 'vsrvdev1'

My kernel version is 4.9.78-vs2.3.9.6-beng-reptoline

I have all the util-vserver systemd packages installed:

# dpkg -l | grep util-vserver
ii  libvserver0                                   0.30.216-pre3126-stretch0.1-1  amd64        dynamic libraries for util-vserver
ii  util-vserver-systemd                          0.30.216-pre3126-1.3           amd64        utilities for managing Linux-VServer guests
ii  util-vserver-systemd-build                    0.30.216-pre3126-1.3           amd64        tools which can be used to build vservers
ii  util-vserver-systemd-build-dbgsym             0.30.216-pre3126-1.3           amd64        Debug symbols for util-vserver-systemd-build
ii  util-vserver-systemd-core                     0.30.216-pre3126-1.3           amd64        core utilities of util-vserver-systemd
ii  util-vserver-systemd-core-dbgsym              0.30.216-pre3126-1.3           amd64        Debug symbols for util-vserver-systemd-core
ii  util-vserver-systemd-dbgsym                   0.30.216-pre3126-1.3           amd64        Debug symbols for util-vserver-systemd
ii  util-vserver-systemd-legacy                   0.30.216-pre3126-1.3           amd64        legacy scripts for util-vserver-systemd
ii  util-vserver-systemd-legacy-dbgsym            0.30.216-pre3126-1.3           amd64        Debug symbols for util-vserver-systemd-legacy
ii  util-vserver-systemd-python                   0.30.216-pre3126-1.3           amd64        python-bindings for util-vserver-systemd
ii  util-vserver-systemd-python-dbgsym            0.30.216-pre3126-1.3           amd64        Debug symbols for util-vserver-systemd-python
ii  util-vserver-systemd-sysv                     0.30.216-pre3126-1.3           amd64        initscripts for util-vserver-systemd



On Mon, Feb 5, 2018 at 6:54 AM, Tor Rune Skoglund <trs@swi.no> wrote:
Hi,

did not notice any such issue --- on the 4.9.76 at least. Will be
installing the 4.9.78 now on at least two test servers today, and will
report to the ML if we see any issues like the one you mentioned.

BR,
Tor Rune Skoglund

Den 05. feb. 2018 09:47, skrev Andrew Ruthven:
> Hey,
>
> Has anyone noticed big I/O issues with this kernel?
>
> I ran it for a few days, and found that processes (even on the main
> system) were often sitting in a D state, seemingly blocked on I/O. But
> there didn't seem to be much I/O happening. Even ssh'ing in to any
> vservers or physical host was incredibly slow.
>
> Cheers,
> Andrew
>
> On Tue, 2018-01-30 at 22:50 +0000, Ben Green wrote:
>> Hi all,
>>
>> I've just finished and uploaded Linux kernel 4.9.78 with patch  
>> vs2.3.9.6. They are installable directly from the psand.net  
>> repository, or via the metapackage:
>>
>> linux-image-vserver-4.9-beng
>>
>> For i386 and amd64 architectures, and there's a version I've
>> compiled  
>> with gcc-8 which can be installed directly, or via the metapackage:
>>
>> linux-image-vserver-4.9-beng-reptoline
>>
>> That one is amd64 only of course. These images are only provided in  
>> the Debian Stretch distro repositories. Instruction (which need a
>> bit  
>> of updating but should get you where you need to be) are at:
>>
>> https://www.psand.net/services/public-repository.html
>>
>> These kernels are minimally tested and as usual come with
>> absolutely  
>> no guarantees whatsoever.
>>
>> Cheers,
>> Ben
>>