Opened 10 years ago

Closed 10 years ago

#7537 closed defect (fixed)

mythcommflag does not run at start of recording but immediately shows finished status 272

Reported by: mikpolniak@… Owned by: cpinkham
Priority: minor Milestone: unknown
Component: MythTV - Mythcommflag Version: 0.22
Severity: medium Keywords: mythcommflag
Cc: Ticket locked: yes

Description

I have a master backend where commflag works as expected. My slave backend has a single tuner and its commflag job is created at the start of a recording with status 1 queued. But then immediately the jobqueue shows finished status 272 and comment "0 commercial break(s)". And of course the job never starts.

I have to select "Begin commflagging" in job options menu. Then another job number for the same recording appears in the jobqueue with the normal status 4 and comment "Building Logo Detection Buffer" as normal.

My setup remains the same as in 21-fixes where this was working. I have set commflag to run on original recording host and to start simultaneously with the recording. There are no other jobs running on the slave.

Change History (4)

comment:1 Changed 10 years ago by mikpolniak@…

The HDPVR is the only tuner on the slave and sometimes it starts recording by blinking (big blue lights) a few times and sometimes the HDPVR blinks 3 or 4 times with pauses at start of recording. This may indicate why the commflagging sometimes runs normally and sometimes aborts prematurely with 'finished status' and 'o commercial breaks'.

I have been using this HDPVR for over a year and always begin commflagging when the recording starts, but this premature abort started when i switched to 0.22

comment:2 Changed 10 years ago by bob.freemer@…

I can confirm this bug exists in 0.22-fixes. The bug exists for a setup with one master backend and one frontend on the network.

comment:3 Changed 10 years ago by stuartm

Ticket locked: set

Bob, Please read the TicketHowTo in future.

comment:4 Changed 10 years ago by cpinkham

Resolution: fixed
Status: newclosed

Closing this as probably fixed. There appear to have been some frame number issues in 0.22 (and -fixes) with H264 material that have been fixed along the way in trunk, possibly by the libavcodec resync. These would potentially cause issues in the flagger if assumptions it made about frame counts were not true. I'm going to close this for now unless this issue can be reproduced with trunk. Lots of people are using trunk with H264 and not seeing this issue.

Note: See TracTickets for help on using tickets.