mpt timed out and Re: port math/gnuplot hangs and ignores "kill -9"

[Anton Shterenlikht]

On Thu, Sep 17, 2009 at 07:41:59PM +0200, Mel Flynn wrote:
> On Thursday 17 September 2009 15:57:43 Anton Shterenlikht wrote:
> > On Thu, Sep 17, 2009 at 02:40:24PM +0100, Anton Shterenlikht wrote:
> > > On Thu, Sep 17, 2009 at 03:16:16PM +0200, Roland Smith wrote:
> > > > On Thu, Sep 17, 2009 at 01:34:04PM +0100, Anton Shterenlikht wrote:
> > > > > > It could be that the process is stuck in the 'D' state
> > > > > > (uninterruptable wait). You can veryfiy that by running 'ps -u' and
> > > > > > looking in the eight column when gnuplot is running.
> > > > > >
> > > > > > Does the window with the plot actually appear?
> > > > > >
> > > > > > Interactive use of gnuplot-4.2.6 is fine on amd64 7.2-RELEASE-p2.
> > > > >
> > > > > I reinstalled gnuplot-4.2.6 and (hopefully) all ports on which it
> > > > > depends. I still get the same behaviour.
> > > > >
> > > > > top -PISu shows:
> > > > >
> > > > > last pid:   108;  load averages:  0.88,  0.35,  0.19    up 2+02:23:38
> > > > >  13:27:52 109 processes: 4 running, 88 sleeping, 17 waiting
> > > > > CPU 0:  0.0% user,  0.0% nice,  0.0% system,  0.0% interrupt,  100%
> > > > > idle CPU 1:  0.0% user,  0.0% nice,  100% system,  0.0% interrupt, 
> > > > > 0.0% idle Mem: 105M Active, 2074M Inact, 363M Wired, 768K Cache, 827M
> > > > > Buf, 5322M Free Swap: 19G Total, 19G Free
> > > > >
> > > > >   PID    UID    THR PRI NICE   SIZE    RES STATE   C   TIME   WCPU
> > > > > COMMAND 11      0      2 171 ki31     0K    64K RUN     0  77.9H
> > > > > 100.00% idle 99992   1001      2  48    0 98240K 55608K CPU1    1  
> > > > > 0:00 100.00% gnuplot
> > > > >
> > > > > so gnuplot is using 100% and all in system state.
> > > > >
> > > > > and ps -u:
> > > > >
> > > > > USER    PID %CPU %MEM   VSZ   RSS  TT  STAT STARTED      TIME COMMAND
> > > > > mexas 99992 98.1  0.7 98240 55608   5  R+    1:25pm   0:00.72 gnuplot
> > > > >
> > > > > so the state is not "D".
> > > > >
> > > > > The window does appear (just using simple gnuplot> plot sin(x), and
> > > > > the terminal is set to 'wxt', but nothing ever apears in the window.
> > > >
> > > > The wxt terminal is only available when gnuplot is compiled with the
> > > > wxWidgets toolkit. Try using the plain x11 terminal, and see if that
> > > > works better?
> > >
> > > yes, that works fine! Thank you!
> > >
> > > So the problem must be with wxgtk2-2.8.10_1 and wxgtk2-common-2.8.10_1 ?
> > >
> > > How can I kill the offending gnuplot process?
> > 
> > On reboot I see this on the console:
> > 
> > System shutdown time has arrived
> > Stopping cron.
> > Stopping sshd.
> > Stopping ntpd.
> > Stopping devd.
> > Writing entropy file:mpt0: request 0xa0000000000d2140:52792 timed out for
> >  ccb 0x
> > 
> > e000000019ece800 (req->ccb 0xe000000019ece800)
> > mpt0: completing timedout/aborted req 0xa0000000000d2140:52792
> > mpt0: Timedout requests already complete. Interrupts may not be
> >  functioning. Sep 17 14:49:59 mech-cluster241 syslogd: exiting on signal 15
> > Sep 17 14:49:59 init: timeout expired for /bin/sh on /etc/rc.shutdown:
> >  Interrupt
> > 
> > ed system call; going to single user mode
> > Sep 17 14:50:19 init: some processes would not die; ps axl advised
> > Waiting (max 60 seconds) for system process `vnlru' to stop...done
> > Waiting (max 60 seconds) for system process `bufdaemon' to stop...done
> > Waiting (max 60 seconds) for system process `syncer' to stop...
> > Syncing disks, vnodes remaining...3 1 2
> > 
> > 
> > I'm a bit worried about mpt0 messages - this is the SCSI driver.
> > Does this indicate a problem with mpt?
> 
> Since gnuplot was spinning in kernel mode, all bets are off. This timeout is 
> most likely a side effect from that, unless you see this every reboot not just 
> with an unkillable gnuplot. 
> 
> If your system has the ability to run procstat -k, you might find out what 
> gnuplot is spinning on. You'll need at least a 7.x system, but I'm not sure if 
> kernelthreads are supported on ia64 and kernel needs to have STACK or DDB 
> options.

Marcel, are kernel threads supported on ia64?

My ia64 system is 9.0-current.
I've DDB in the kernel.
I've rebuilt gnuplot without wxWidgets toolkit, so I don't have this
problem now, but for the future, does this look right:

> procstat -k 51055
  PID    TID COMM             TDNAME           KSTACK                       
51055 100114 gnuplot          initial thread                                
>

many thanks
anton

-- 
Anton Shterenlikht
Room 2.6, Queen's Building
Mech Eng Dept
Bristol University
University Walk, Bristol BS8 1TR, UK
Tel: +44 (0)117 331 5944
Fax: +44 (0)117 929 4423