You are not logged in.
I have not found messages indicating error in the /var/log files, but my system crashes and requires reboot fairly frequently when I ssh to my office from home. This just happened while installing R (from the source, not pacman). But this has also happened while running programs written in c (just ./a.out). I generally cannot reproduce the crash by running the same program at my office (not over ssh but "in person"). Reinstall does not help (not surprisingly).
Offline
What package is "R"?
Last edited by whatshisname (2011-01-12 15:21:24)
Offline
R-2.12.1 from http://cran.r-project.org/
Offline
OK. Thanks. Never heard of "R" so wanted to make sure it wasn't a typo.
When you're sitting at the console, have you tried ssh'ing back into your machine to see if you can reproduce the problem that way? Perhaps you might see an error message at the console you wouldn't otherwise see.
Otherwise, I would be suspicious of R. This might be a question for them instead of the Arch community. To be sure, of course, remove R and see if the problem persists.
Good luck.
Offline
I think I now understand. If I accidentally start an infinite loop that involves reading and writing, ssh kills the system. I wonder why this happens in ssh, but not "in person."
Offline
I think I now understand. If I accidentally start an infinite loop that involves reading and writing, ssh kills the system. I wonder why this happens in ssh, but not "in person."
Someone far more versed in ssh would have to answer that one. But at least you found the source of your problem. Congratulations!
Offline
I think I now understand. If I accidentally start an infinite loop that involves reading and writing, ssh kills the system. I wonder why this happens in ssh, but not "in person."
This does not make sense to me. First, what is the meaning of "crash"? Kernel oops/panic, or spontaneous reboot? sshd is run as your user... I don't understand how it can kill the machine. Also, what is your /etc/security/limits.conf? You could try executing program with script(1) to capture the output...
Arch Linux is more than just GNU/Linux -- it's an adventure
pkill -9 systemd
Offline
Agree my last suggestion did not entirely make sense. I guess I am just generating system crashes with code, but I am highly puzzled. The usual story is
(1) develop a shell script or do some data looping in c or sometimes R
(2) run and debug code and run again many times
(3) Eventually crash the system even though no evidence that a lot of memory is getting eaten
(4) The ssh version is:
(a) ssh connection dies
(b) Can't ping or re-ssh
(c) show up to work the following day and find computer on, but display off (monitor on yellow instead of green light)
(d) restart computer, run same code, no crash, runs fine.
(e) go through logs and find no evidence of a failure.
In person ssh doesn't generally crash the way the remote ssh did. So it seems like system crashes just happen with some probability.
I have recently crashed the system in person. I get a garbled display and system doesn't respond. Again, log files provide no evidence of anything.
This is not an arch issue, apparently, as same thing just happened on a fresh install of debian.
Offline