http://bugzilla.novell.com/show_bug.cgi?id=415607
User nfbrown@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=415607#c47
Neil Brown changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|ASSIGNED |NEEDINFO
Info Provider| |R.Vickers@cs.rhul.ac.uk
--- Comment #47 from Neil Brown 2009-09-30 19:35:19 MDT ---
I'm not having a lot of luck. I'm no longer thinking of that patch...
I compared the trace with the kernel code and I'm fairly sure that
the trace is only possible if either XPRT_CLOSING is set, or
XPRT_CONNECTING is set.
The first should only be set if the TCP connection is in one of the states:
FIN_WAIT_1 FIN_WAIT_2 CLOSING TIME_WAIT LAST_ACK
It is very unlikely the connection would stay in those states for more than
a couple of minutes providing the server was up and responding. If it
happens again you could check if the TCP connection is still present and
what its state is with
netstat -nt | grep 2049
Setting of XPRT_CONNECTING will be accompanied by a debug message mentioning
xs_connect, then it will be cleared again within at most 5 minutes which will
result in a "worker connecting ..." message.
We don't see either of those messages over a period longer than 5 minutes.
So my only hope is that "netstat -nt" might show something useful
if the problem happens again (assume you are willing to try running
with TCP for a while again....)
So: setting to "NEEDINFO" for "Can I get 'netstat -nt' output while the
problem is happening'.
--
Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.