Re: [mh] interesting fails - hang debugging help

[Rick Bolen(gm) <ric...@gm...>]

Well if the process appears to hang, it might be worth running the 
instance in the perl debugger, then breaking into it once it is "hung" 
and dumping a callstack to see where it's at.

So here's the steps I remember from recently setting up mh to "run 
under" and "be interruptable" by the perl debugger:

1) modify the main mh script "<path-to-mh-code>/bin/mh" and comment out 
the line that enables mh to hook CTRL-C to exit cleanly:

somewhere around line 726:

$SIG{INT}   = \&sig_handler;              # Exit cleanly with CTL-C
to...
#$SIG{INT}   = \&sig_handler;              # Exit cleanly with CTL-C

now you can hit CTRL-C when running under the perl debugger and halt mh 
execution in the debugger.

2) run mh under the debugger

perl -d <path-to-mh-code>/bin/mh

3) once mh halts look at the command terminal it launched in. Hit CTRL-C 
if it hasn't already stopped at the debugger prompt:

  DB<1>

then type "T" at the debugger prompt to dump a callstack (stack 
backtrace). Continue debugging from there.

Hitting "c" allows the process to continue.


Good luck,

Rick





On 05/08/2013 11:21 AM, Joel Davidson wrote:
> Hi all,
>
> I recently updated my misterhouse box (Intel Atom with 2G of ram)
> from Fedora 11 to Fedora 18.  I know I should have left well
> enough alone, as the box was running fine with FC11, but there
> is some new s/w I want to run on the box that requires a newer
> kernel.
>
> The system itself is running fine, but misterhouse seems to hang
> after anywhere from 30 minutes to an hour of run time.  There are
> no messages that I can find either in the mh log file or in the
> system logs.  I had been running on a very old mh release (105),
> so I grabbed the latest copy from git yesterday and am seeing the
> same results.  I use only x-10 to control my lights, some scripts
> that call digitemp (until I get owfs working reliably), and a
> driver that talks to some old HCS-2 lcd-links via rs-485.  I have
> no insteon devices at all.
>
> Anyone have any suggestions as to where to look, and how to start
> diagnosing this?  I turned on debugging for x10, serial, and a
> bunch of other things, but haven't seen anything yet.  Thanks.
>
> Joel
>