[mythtv-users] Mysterious recording dropouts and frontend disconnects

[Jerry Rubinow]

Setup consists of a separate FE and BE, running mythbuntu autobuilds of
.22-fixes, build 23193.  The BE has an HDHomerun and an HD-PVR.  FE is a
Revo 1600.  BE is an AMD Athlon 64 3500.  Recordings on the backend are on a
separate drive than the system/mysql.

I set up three simultaneous records, one on each HDHR tuner and one on the
HD-PVR.  I wait a few minutes, then go to Watch Recordings, and one by one,
tell each recording to halt.  When I tell the first one to halt, I get a
very long pause after the confirmation screen, and it usually says it lost
the connection to the BE.  But it does end the recording for that program.
 I go to the second one and tell it to halt, usually the same thing.  The
third one usually halts without a problem.

Then I go to watch the recordings.  They all start having dropouts (missed
frames and audio, sometimes severe) around the time I started telling it to
end recording.

Backend long consists of:

- endless "ATSCStreamData:HandleTables(): Unknown table 0xc0", about one
every 18 milliseconds

punctuated by:

- a GetPlaybackURL: 'filename' should be local, but it cannot be found
- MainServer::ANN Monitor
- adding: frontendname as a client (events:0)
- again about 10 seconds later, followed immediately by MainServer, Warning:
Unknown socket closing(number)
- Empty LocalHostName
- MythSocket(996c498: -1): writeStringList: Error, socket went unconnected
We wrote 0 of 10 bytes with 1 errors
- and about 2 seconds later, messages indicating it's changing from
RecordingOnly to None for the first program I halted

On the frontend log:

about 10 seconds after I confirmed the halt of the first record,
- MythContext: Connecting to backend server
- MythSocket: readStringList: Error, timed out after 7000 ms
- MythContext, Error: Timeout connecting to server socket to master backend
- QObject: Cannot create children for a parent that is in a different thread
- Connection to master server timed out
- Preview Error: Failed to open sockets
- and a bunch more of the same

Anyone have any ideas on this?  I saw a thread from December relating to
file sizes being updated too often (milliseconds instead of seconds).
 Symptoms sounded possibly similar.

-Jerry
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mythtv.org/pipermail/mythtv-users/attachments/20100119/6b2a26f0/attachment.htm>