VDS crashes while running updates

[Kirby420]

Got a single core VDS for some small hosting for my gaming community, namely to run a Mumble server, a FastDL for our two TF2 servers, Minecraft server and an occasional 8 man tr_walkway TF2 training map server.


Which brings me to my problem, I attempt to run ./steam -command update -game tf to update SRCDS from no install, and the VDS simply hard-locks from the load and restarts itself after about 2 minutes. That's pretty incredible, considering any little 800MHz home PC could run that update without choking. I've replicated this twice now, once when the server had no connections to it and was sitting idle, another when there were about 15 users connected to the Mumble server.


In addition to that little annoyance, I just attempted to run a simple apt-get install openjdk-6-jdk (Running Ubuntu 10.04x64) on the box to install Java VM, and the box does it again. I watched the box via a 2nd SSH window running htop, the system load slowly crept up to around 3.84 before the box crashed, and now it's entirely FUBAR'ed.

The VDS will not boot, it's been down for almost an hour now. Opening the web control panel to the server control page nets me a # 727 error, says it can't even query my server. The box is 100% unavailable to me, I couldn't even perform a OS wipe and reinstall if I wanted to.

This is the 2nd time trying to install Java VM on my VDS has completely destroyed it. Last time it happened, I had run the same command via SSH, but my PuTTY instance crashed on me, so I attributed the breakage to me, assuming I missed some user-input and the box died waiting. This time however,it did not, and the SSH connection simply died when the VDS crashed. I could get to the Web CP last time and reinstalled Ubuntu, but that really is a crappy option for me this time around. I've spent the past two days getting everything on the server set up to my liking, and I really did not expect a simple software install to literally break my server.

[Kirby420]

-edit-

I was directed to post here for help by Brendan via my support ticket, because "it is software related so it needs to be handled in the forums." - but yet I can't access the box in any way, via SSH or your website. The VDS is dead on that side.

Not really very helpful there...

[Edge100x]

./steam -command update -game tf to update SRCDS from no install, and the VDS simply hard-locks from the load and restarts itself after about 2 minutes.

That's very strange and definitely a software problem of some sort. What is showing in your /var/log/messages?

I just attempted to run a simple apt-get install openjdk-6-jdk (Running Ubuntu 10.04x64) on the box to install Java VM, and the box does it again. I watched the box via a 2nd SSH window running htop, the system load slowly crept up to around 3.84 before the box crashed, and now it's entirely FUBAR'ed.

This isn't something that I've heard of happening before for anyone, but I don't know if we have any other renters running Ubuntu 10.04, either. I will have to do some further testing, since it sounds like that version of Ubuntu either might ship with a buggy kernel, or be triggering a bug of some sort in Xen. (I have tested Ubuntu 8.04-LTS, CentOS, and 32-bit and 64-bit versions of Gentoo, and I didn't see this problem on them.)

The VDS will not boot, it's been down for almost an hour now. Opening the web control panel to the server control page nets me a # 727 error, says it can't even query my server.

Control panel access isn't a software problem, so please do contact us directly about it.

Is there an event posted to your Events log about the machine crashing? If so, please take a gander at that. I just had to reboot a physical machine today in Chicago because it crashed, and from the log, it appeared to be related to one of its VDSes frequently rebooting (which really shouldn't cause issues.. but Xen does have bugs), which sounds an awful like what was happening to you. Whenever there's an overall problem with a machine or bandwidth, we do our best to post about it to that Events log.

If there's *isn't* an event posted to indicate that the control panel problem is known, please contact us ASAP through the "Help!" button in your control panel so that we can investigate. We have a monitoring system that should catch all issues like this, but there's always the possibility of a further glitch.

[Edge100x]

I just tested with a fresh Ubuntu 10.04-LTS install on a machine in Dallas and both the install of Java and run of ./steam worked. If you were on the machine that crashed tonight, I suspect that your problem may have been related to that situation -- could you test again now and see if these work for you?

If they still don't work, you may have a corrupted virtual hard drive on your hands, and we'll have to run fsck to check it out.