[mythtv-users] Re: Backend crash with current CVS

Michael J. Sherman msherman at dsbox.com
Tue Jul 6 16:56:48 EDT 2004


 > When is "over the weekend". That's a 48-72 hour period
 > depending on whether you mean Sat/Sun or Friday night to
 > Monday night since in the U.S. a lot of people had 3-day
 > weeknds.

I grabbed it Friday night.  Again Saturday, and again Sunday.  Didn't 
notice anything in the commit logs all the way until Monday afternoon 
when I checked last.  So when I said "over the weekend", it really way 
over the entire weekend, grabbing from CVS each day.  Same behavior. 
I did not grab anything last night, so I missed your changes that you 
committed.

 > I think others have mentioned something like this before prior
 > to the commflag changes, but if you can get gdb to show where
 > it's happening that would help.

Yes, I know a gdb trace would be spectacular, but as I said for some 
reason gdb was not cooperating.  I'll get the exact error message 
later, but I'm sure there was no call stack to print out.  I think it 
was getting confused with all the threads.  In fact at one point I 
couldn't even start the frontend because gdb had mucked up the threads.

Even with the June 15th CVS snapshot I grabbed, I had the backend 
crash this morning.  It used to be rock-solid until about late May, if 
memory serves...  Now I have to continually check mythweb from work, 
and if the backend has crashed, ssh into the box and restart it.

I'll do my best to get a backtrace out of it tonight with a latest CVS.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: msherman.vcf
Type: text/x-vcard
Size: 295 bytes
Desc: not available
Url : http://mythtv.org/pipermail/mythtv-users/attachments/20040706/bbced1cb/msherman.vcf


More information about the mythtv-users mailing list