[5946] in Release_7.7_team
Re: [Vmware-release] running Athena virtual machine from off-campus
daemon@ATHENA.MIT.EDU (Jonathan Reed)
Tue Mar 25 08:29:50 2008
Cc: release-team@mit.edu, gettes@mit.edu, vmware-release@mit.edu
Message-Id: <219587E7-A344-43BF-AA4D-D9BC5F344262@mit.edu>
From: Jonathan Reed <jdreed@MIT.EDU>
To: Alex T Prengel <alexp@mit.edu>
In-Reply-To: <200803140422.m2E4MlGG016998@dit.mit.edu>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
Content-Transfer-Encoding: 7bit
Mime-Version: 1.0 (Apple Message framework v919.2)
Date: Tue, 25 Mar 2008 08:29:08 -0400
X-Spam-Flag: NO
X-Spam-Score: 0.00
I finally started playing around with this, and was able to
successfully suspend and resume an Athena VM, both in logged-in and
logged-out state.
The VM is a default install (public, not dev cell - via PXE) of Athena
9.4.44. VMware tools is NOT installed (I can run 1024x768 just fine
without it, and time synchronization is done over the network).
On Friday, I logged out of the Athena machine, and suspended it in
VMware Workstation. I then left for the weekend. Yesterday morning,
I resumed the machine, and was able to successfully log in as root,
and confirm that AFS and network were working with no action on my
part. After synchronizing the time, I was able to log in as joeuser.
Yesterday at 4pm, I logged in as joeuser, and then suspended the
machine. This morning, I resumed it, and confirmed that AFS and
network still work (I was able to ls /afs/athena.mit.edu and ping the
DNS servers). I then cd'd to /tmp, so my cwd was not in AFS. I then
su'd to root (which took about 30 seconds, longer than normal,
undoubtedly do to time sync issues). After synchronizing time,
xscreensaver immediately activated due to the 16 hour time warp.
After unlocking the screen, I still had network and AFS, and, most
impressively, my zephyr subs. As to the latter, is there code in xss
that runs "renew", if the user's authentication expires while the
screen is locked? Because it would seem to have done that. Prior to
synchronizing time, I had tickets for afs, gms/lert, zephyr, and imap,
all of which expired at 2 am this morning. After sync'ing time, and
unlocking the screen, I had new TGTs, and tickets for afs and zephyr,
all of which are set to expire at 6pm this afternoon. Which really
sounds like what would have happened I had typed renew.
For those of you for whom resuming failed (resulting in network and
AFS lossage), did you start using the VM as soon as it came up?
VMware uses "background resuming", in which you can start using the
machine before it's fully restored. This is a preference, and can be
turned off. In Workstation, there's a progress bar at the bottom
right, next to the CD/floppy/HD/Network blinky lights. I waited until
this progress bar had disappeared before attempting to do anything
with the VM. I wonder if not doing so is the source of the network
problems.
-Jon
On Mar 14, 2008, at 12:22 AM, Alex T Prengel wrote:
>
> Hi,
>
> to follow up on Bill's report today:
>
>> Alex Prengel and I are experimenting with Athena in a VMware image.
>> Yesterday he brought a problem to me that I have reproduced myself:
>>
>> Repeat by:
>>
>> Start Athena VMware image.
>> Log in.
>> Suspend image.
>> Let sit over night.
>> Resume image.
>>
>> Actual behavior:
>>
>> The network is down.
>> AFS is dead.
>> A little while later the panel disappears with the helpful message,
>> "Killed"
>> in the console window.
>
>> ...
>
>> I was able to recover by:
>>
>> su
>> /etc/sysconfig/network-scripts/ifdown eth0
>> /etc/sysconfig/network-scripts/ifup eth0
>> /etc/init.d/openafs stop
>> /etc/init.d/openafs start
>
> There is a further problem, most likely related to this one, in
> attempting to
> run an Athena virtual machine from off-campus. I have an Athena
> virtual machine
> that works reliably on campus (at least for time periods up to
> several hours).
> I can suspend and resume it correctly (though only tested over a few
> hours).
>
> When I copy the machine to one of my home computers (I have a Verizon
> FIOS connection) and try to restore the machine, network and AFS are
> dead and unlike Bill's experience, I can't restart them with "ifup
> eth0", "network start" or "openafs start" commands.
>
> I did a further test- I put a working Athena virtual machine on a
> laptop on
> campus. I verified that I could suspend/resume the machine
> repeatedly and
> reliably. I then suspended the machine, took it home and resumed.
> Again
> no network or AFS, and no way to restart them.
>
> I tried turning off both the PC firewall and the router firewall on my
> home system to see if those might be causing this. Didn't make any
> difference.
>
> I checked /etc/sysconfig/network-scripts/ifcfg-eth0 in the virtual
> machine and there was no HWADDR entry. The IPADDR entry was
> correct. My virtual machine has a fixed IP address, like Bill's. I set
> this and the hostname in /etc/athena/rc.conf in the virtual
> machine. Most on my testing was on Windows host OSes, but I
> experienced the same thing at home with a Linux (Fedora 7) host OS.
>
> Alex
