On 17/10/16 18:43, Manuel Bouyer wrote: > this means that they're all stuck on a lock. > The first thing would be to use ddb (or crash(8), if this works on Xen) > to find where in the kernel these processes are stuck. > > The next step would be to which which process is actaully holding the lock > (this one is probably not in tstile state) > Thanks, Manuel... looks like crash(8) works, I'll check that out next time it happens! Does the lock show as the "wait channel" as shown by ps/w? How do I find what process currently holds it? And then I guess I do a "trace/t" on the pids involved to see what's what? I'm a little green with ddb. -- Alaric Snell-Pym http://www.snell-pym.org.uk/alaric/
Attachment:
signature.asc
Description: OpenPGP digital signature