[mythtv-users] Dying Backend

Gavin Haslett gavin at nodecaf.net
Tue May 17 23:48:35 UTC 2005


Nothin' there I'm afraid. This is all I got from the backend;

2005-05-17 18:41:36.022 ClearListMaps...
2005-05-17 18:41:36.078 Sort by time...
2005-05-17 18:41:36.134 PruneRedundants...
--- print list start ---
Title - Subtitle                    Chan ChID Day Start  End   S C I  T
N Pri
2005-05-17 18:41:36.190 Host mythtv.alopex.nodecaf.net updating endtime
to 20050517184100
The Simpsons                          30 1030  17 18:37-19:00  1 1 1  S
S   0
2005-05-17 18:41:36.305 Host mythtv.alopex.nodecaf.net renaming
/video/recordings/1030_20050517183700_20050517190000.nuv to
/video/recordings/1030_20050517183700_20050517184100.nuv
Futurama - "Love's Labour Lost in S  296 1296  18 01:00-01:30  1 1 1  T
1   0
2005-05-17 18:41:36.448 1       1
MythBusters - "Breaking Glass"       278 1278  18 23:00-00:00  1 1 1  W
1   0
Futurama - "Fear of a Bot Planet"    296 1296  19 01:00-01:30  1 1 1  T
1   0
McLeod's Daughters - "The Bridle Wa  260 1260  20 14:00-15:00  1 0 0  F
R   0
Keeping Up Appearances - "How to Go  264 1264  20 14:20-15:00  1 1 1  W
1   0
McLeod's Daughters - "To Have and t  260 1260  21 21:00-22:00  1 1 1  F
1   0
McLeod's Daughters - "To Have and t  260 1260  22 00:30-01:30  1 1 1  O
1   0
2005-05-17 18:41:37.014 50      BACKEND_MESSAGE[]:[]DONE_RECORDING 1
248[]...
Futurama - "My Three Suns"           296 1296  22 01:00-01:30  1 0 0  O
X   0
2005-05-17 18:42:01.665 New DB connection, total: 1
Starting up as the master server.
 
Note that I was deliberately recording a show (The Simpsons), stopped
the recording and was using Mythweb. Nothing odd in the logs, no
errors... Just this. Note that it got as far as "Done recording" then
died almost immediately. The restart was forced by my job that runs
every minute.

Not sure about the binary packages... It did the same thing (though
slightly more frequently) with 0.17 as well.

Thoughts anyone? It still looks like a database problem to me, the
failing seems to happen when the database is being utilized a lot (and
watching "top" I saw usage of memory of the mysqld process spike a
little higher than usual right before the crash). Could this be a mysql
bug we're fighting? Version perhaps?

-----Original Message-----
From: mythtv-users-bounces at mythtv.org
[mailto:mythtv-users-bounces at mythtv.org] On Behalf Of Risto Treksler
Sent: Tuesday, May 17, 2005 3:04 PM
To: Discussion about mythtv
Subject: Re: [mythtv-users] Dying Backend

> since I use binary packages is there a way I can up the verbosity of 
> these logs to see if anything weird is going on?

you can edit /etc/init.d/mythbackend
go to the "start)" section
find the line that calls mythbackend
add " -v all "

or just run mythbackend manually in a shell

log in remotely or switch to a virtual terminal locally with
<ctrl><alt><f1> (you can go back with <ctrl><alt><f7> usually)

comment out the cron job for now
this way, if the backend dies, you can switch to that vt to see tha last
log messages printed

just type
"mythbackend -v all"

or loop it with
"while : ; do mythbackend -v all ; sleep 5 ; done"

> Another time the backend dies is if I've been using Mythweb, that 
> seems to cause more crashes than it's sometimes worth (though I do 
> prefer using Mythweb to locate shows that I want to record).

does mythweb make the backend die for you only on the recorded_programs
screen or on other screens as well?

> Currently I'd say I lose backend about twice per day on average at the

> moment. Version 0.18.0 (not had time to do 0.18.1, but doesn't seem 
> like it fixes any backend or commflag bugs).

try current cvs with debugging turned on and send a backtrace to the
mythtv-dev list

every now and then this seems to be happen to someone, but since none of
the devs can seem to reproduce it, it's hard to fix.

best to follow these instructions
http://mythtv.org/docs/mythtv-HOWTO-21.html#ss21.2

who knows, the crashes could be caused by your binary packages
themselves so, does it still happen with CVS with debugging turned on?

also,
look up the "ulimit" command
you can set it so that core files are written to disk when a program
crashes "ulimit -c unlimited"

that way you can run "mythbackend -v all" and do a "gdb mythbackend
core"
or similar when it dies to make a backtrace even though you weren't
using gdb to run mythbackend


--
Risto Treksler
Elkhorn Lodge
Banff, Alberta, Canada
_______________________________________________
mythtv-users mailing list
mythtv-users at mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users



More information about the mythtv-users mailing list