> Additionally, I just set up a directory, on the head node, with the > entire filesystem from the compute nodes, and chrooted to it, copying the > /proc/cpuinfo + meminfo + stat entries into where they should be, and > gmond starts up fine in the chrooted environment... essentially what it > 'should' be like on the cluster. > > I'm going to play with it more tomorrow - it's most likely a stupid > error on my part, which some sleep might bring to light. :)
Hmm - check the filesystem. Is /tmp writable? Is /proc and/or /dev mounted? If you interrupt gdb, where's the backtrace? I'm just guess here. Try strace-ing or ltrace-ing the process to see what it's doing towards the end. Regards, Robert. -- Robert Walsh Amalgamated Durables, Inc. - "We don't make the things you buy." Email: [EMAIL PROTECTED]
signature.asc
Description: This is a digitally signed message part
