[OpenAFS] Re: busy .backup volumes
Andrew Deason
adeason@sinenomine.net
Thu, 2 May 2013 12:44:50 -0500
On Thu, 02 May 2013 11:19:41 -0600
"Kristen J. Webb" <kwebb@teradactyl.com> wrote:
> I've omitted the volumes from backup so I can control the
> order of things better next time. Is there anything else
> useful I can get before restarting the server?
Capture a core. What seems likely is the Dump RPC is executing, the
relevant Rx call is errored, but for whatever reason the Dump RPC is not
returning. I don't know what it could be hanging on if the Rx call is
errored, so all you can do is see where it is stuck.
A core itself obviously has a lot of sensitive information in it.
However, just the stack trace will not, as long as you make sure to trim
all of the function arguments that may appear in it. (In gdb, get a
stack trace with 'thread apply all bt'.) But just save the core even if
you don't do anything besides get a stack trace, so you can poke around
in it later if necessary.
--
Andrew Deason
adeason@sinenomine.net