From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.9) Gecko/20020412 Debian/0.9.9-6 Description of problem: I have successfully installed and tested (using the crash.o module) the netdump client (0.6.6-1) on our RHAS server. However, when the server actually crashes "for real", no netdump gets sent :-(. This may of course be because the kernel is too fucked up to be able to send a crash dump, but it *could* depend on that the kernel just hasn't been able to detect that it has crashed. In case 2, I'd like to be able to explicitly tell the kernel that it has crashed. Unless you have a better idea, an alt-sysrq key combo for doing just that would be great. When crashed, our server does respond to alt-sysrq-'b'oot, but it seems not to want to do any alt-sysrq-'u'nmounting. So, a magic sysrq key combo could help us out (or not...).
Our kernel actually responds to Alt-SysRq-* when crashed, and it does send the result codes to the netdump server. Thus, having this implemented has a great chance of actually helping fixing bug 67608 (kernel hang).
Agreed, a dump-now sysrq combo would be convenient.
This feature has been implemented for the next enterprise product.
Hello .. can you tell me what this dump-now switch is ?? I cannot get ALt-Sysrq-m to work at all it just does nothing .. I can do Alt-Sysrq- t .. so I know that the Alt-Sysrq function works anyhow .. please adivse ..we are running this is currently being tested on ES 3.0 v2 ..thanks
It's Alt-sysrq-c, which will force an oops upon receiving the keycode.