Custom Query (1145 matches)

Reported by jdreed, 14 years ago.

A cluster machine that is sad (i.e. because it downgraded itself to -workstation) should know enough to call for help.

A possible method is that the installer drops an unpackaged initscript somewhere that, if debathena-cluster is not installed, displays a single X window that says "This workstation is temporarily out of service, please contact hotline@…".

The window could also do more awesome things,like prompt the user to click a button to hackboot into PXE and reinstall, or something else.

But at the very least, it should advertise its brokenness.

Reported by kaduk, 13 years ago.

Sometimes, when a cluster machine decides it wants to reboot as a user is logging out, it will end up just spinning its dots in the Ubuntu shutdown splash screen. Pressing 'del' to go to the console shows mostly standard shutdown messages about various daemons and such being killed by TERM, but also a few complaining about the session chroot not being able to be destroyed because it was still in use (from memory, may not be quite accurate). Yet a three-finger salute brings the machine down and back, cleanly.

Reported by kaduk, 13 years ago.

Sometimes, when a cluster machine decides it wants to reboot as a user is logging out, it fails to actually reboot, and ends up not at the Ubuntu splash screen, but at the text console with a bunch of things like:

INFO: task [task]:NN blocked for more than 120 seconds
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message

where [task] is things like fsnotify_mark, dbus-daemon, gdm-binar, polkitd, and N is perhaps the corresponding pid?

It is possible to soft-reboot machines in this state, though I don't remember whether ctrl-alt-del works or sysrq-b was needed.