[mythtv] in progress recordings and new scheduler
bob at smalltime.com
Tue Mar 16 21:25:29 EST 2004
Here's a bug that I encountered today, I'm now running in gdb to see if
I can catch it.
masterbackend is running, has two capture cards. I see that I have
conflicts, so i start a slave backend, two capture cards. I now have no
conflicts for the time period starting in 30 minutes, which is good. I
then realize that my master is running older-than-cvs code and doesn't
include the new change to try to get around the start late problem, so I
ctrl-c it. frontend keeps running, then starts complaining that it
when I ctrl-c'd the master, it was already in the process of recording
"oblivious" and "century city".
make installed the new version, restarted the master. It didn't start
re-recording oblivious and century city immediately; took about 2 minutes.
it then scheduled as follows. 13 and 14 are on the master, 15 and 16
are on the slave. note that Century City is now recording on the first
card on the slave.
2004-03-16 20:06:30 Recording starts soon, AUTO-Startup assumed
2004-03-16 20:06:30 Started recording "Oblivious" on channel: 1028 on
, sourceid 1
Unable to write to client socket, as it's no longer there
2004-03-16 20:06:30 Started recording "Century City" on channel: 1002 on
15, sourceid 1
2004-03-16 20:06:30 Changing from None to RecordingOnly
2004-03-16 20:06:30 Slave backend: frontend has left the building
2004-03-16 20:06:32 Found changes in the todo list.
Right after scheduling on the slave, the slave got the job:
2004-03-16 20:05:24 Connecting to master server: 192.168.10.50:6543
2004-03-16 20:05:24 Connected successfully
[mythtv at frontend mythtv]$ 2004-03-16 20:06:30 Changing from None to
And segfaulted. Also note that it took ~ 1 minute from the time that
the masterbackend had started (based on the NTP synced timestamps)
before the jobs were scheduled.
Don't know if any of this is significant or helps. We'll see what
happens with gdb.
More information about the mythtv-dev