Linode.com Forum

[owlmanatt]

Howdy folks, new customer here.

I set up an Ubuntu 8.04 Linode earlier tonight (ahem...yesterday evening now, I guess). Around 1 AM I got an email from the friendly monitoring program telling me something was horribly wrong and my server was performing an average of 422 disk IO operations a second.

I poked around trying to figure out what could be doing that and I noticed /var/log/syslog and /var/log/messages were getting spammed up with this:

Quote:
Apr 28 02:46:46 elsa init: tty0 main process (6163) terminated with status 1
Apr 28 02:46:46 elsa init: tty0 main process ended, respawning
Apr 28 02:46:56 elsa init: tty0 main process (6166) terminated with status 1
Apr 28 02:46:56 elsa init: tty0 main process ended, respawning
Apr 28 02:47:06 elsa init: tty0 main process (6181) terminated with status 1
Apr 28 02:47:06 elsa init: tty0 main process ended, respawning
Apr 28 02:47:16 elsa init: tty0 main process (6184) terminated with status 1
Apr 28 02:47:16 elsa init: tty0 main process ended, respawning
Apr 28 02:47:26 elsa init: tty0 main process (6212) terminated with status 1
Apr 28 02:47:26 elsa init: tty0 main process ended, respawning
Apr 28 02:47:36 elsa init: tty0 main process (6217) terminated with status 1
Apr 28 02:47:36 elsa init: tty0 main process ended, respawning


I wasn't sure if that was causing my disk IO woes, but that sure as hell looks like it merited fixing. I kept an eye on ps aux and the log to figure out what process was dying, and it turned out to be processes like this:

Quote: root 6423 0.0 0.1 1628 580 ? Ss 02:52 0:00 /sbin/getty 38400 tty0

My friend and I did some googling, and we turned up with this forum post.

Ubuntu handles starting the ttys slightly differently than described in that post: there's a file in /etc/event.d that tells it to respawn /sbin/getty for tty0 if it dies. I commented the respawn lines out and rebooted.

Quote: nevans@elsa:~$ cat /etc/event.d/tty0
# tty0 - getty
#
# This service maintains a getty on tty1 from the point the system is
# started until it is shut down again.

start on stopped rc2
start on stopped rc3
start on stopped rc4
start on stopped rc5

stop on runlevel 0
stop on runlevel 1
stop on runlevel 6

#respawn
#exec /sbin/getty 38400 tty0

Lassie was kind enough to turn my VM back on for me. It booted up without a problem, and the out-of-band console looked happy enough. It's using tty1 now, but I didn't check to see if its using tty0 before or not. I'd wager it wasn't.

I ssh'd in and found that my logs were no longer being spammed with messages about init dying. Yay!

Can anyone confirm whether or not this is a problem caused by Xen trying to kill tty0? Is this a bug? Is there a better solution to the problem...?

[bdonlan]

tty0 doesn't exist on xen (only tty1 does), so that's why getty on tty0 was dying.

[MaineCoon]

I had a similiar problem - some package I installed added a file to /etc/event.d called hvc0, and I had a bunch of similiar but with hvc0.

Thanks to duplicity backups I was able to tell the file hadn't been there that morning...