Hi, On Fri, Mar 29, 2013 at 11:04:34AM +0100, Fabien Wernli wrote: > We just encountered a serious problem when switching from ipmitool to > ipmiconsole: on at least two out of several hundred hosts, multiple > 'SysRq' breaks were sent just after the sol session establishment, resulting > in one server reset, and one power off.
I just managed to reproduce the problem: 1. On the console, generate some output, e.g. while date; do usleep 500000; done 2. On the console server, loop a SIGKILL, e.g. while sleep 30; do pkill -9 -f 'ipmiconsole -h server12'; done After several thousand iterations, I was able to trigger the SysRq. As before, I get quite some garbage on screen, plenty of sysrq requests in '/var/log/messages' (mostly HELP commands), and also "INIT: Id "S0" respawning too fast: disabled for 5 minute". Initially, the SIGKILL was sent by the init script we're using (killproc in redhatish 'rc.d/functions'). I will work around this by editing the init script, but I think it's worth mentioning here that it is reproductible. If someone would like to dig into this, I'd be most happy to assist. Cheers _______________________________________________ Freeipmi-users mailing list [email protected] https://lists.gnu.org/mailman/listinfo/freeipmi-users
