Burn Notice - Monday 1st Sep

[Vortical]

Thursday and Friday have a tonne of Seven HD & Ten HD timers that simply dissappeared off the Upcoming Recordings list fore device 0 in Ice Interactive but they are still on the pvr luckily.

7 HD Digital 05/09/2008 13:00 120min Once Only Ice Interactive Timer
Ten HD 05/09/2008 18:30 30min Series Ice Interactive Timer
Ten HD 05/09/2008 20:30 130min Once Only Ice Interactive Timer
7 HD Digital 05/09/2008 22:00 120min Once Only Ice Interactive Timer
7 HD Digital 06/09/2008 19:30 120min Once Only Ice Interactive Timer
7 HD Digital 06/09/2008 21:30 90min Once Only Ice Interactive Timer
7 HD Digital 06/09/2008 23:00 120min Once Only Ice Interactive Timer

Monday and Tuesday are ok for the timers that have already been sent and there are a few queued timers for Tues & Wednesday that are yet to be sent.

[Daniel Drysdale at IceTV]

I believe I have an answer for you.

The short answer is that this disappearing timers issue is a combination of a bug in the TF7100 firmware and a database maintenance task running on our server.

The correct procedure for a task coming down from the IceTV server to a PVR is as follows..

1) Task sent down with state WAITING
2) Server updates local copy of task to state SENT once the message is sent to the PVR
3) PVR responds by sending the task back to the server with state PENDING
4) When the task is running (recording) on the PVR the state is sent up to the server as RUNNING
5) When the task is complete (recording finished) on the PVR the state is sent up to the server as COMPLETE

If an error occurs in scheduling or recording the task it is sent to the server with state FAILED

What we are seeing with these disappearing tasks is

1) Task sent down with state WAITING
2) Server updates task state to SENT
3) Server updates task state to MISSING
4) Task sent down again with state MISSING
5) Task sent back to server as a DELETE with state FAILED

The reason we are seeing the task state being updated to MISSING is that we have a process that runs at 2:00 am every day that performs a number of routine database cleanup functions; One of these functions is to find tasks that are still in a SENT state and have been that way for more than 12 hours. Since tasks are supposed to be acknowledged by setting the state to PENDING on the next update we would not normally see them in this state unless the PVR fails to acknowledge them (as in this case) or the PVR is off for 12 hours after accepting the task but before it can acknowledge it.

Unfortunately the TF7100 incorrectly takes the MISSING state as some kind of error condition and then deletes the task locally as well as sending a DELETE that is incorrectly set to a state of FAILED.

We have confirmed that the latest firmware v1.11.06 still has this "auto delete" bug, fortunately it appears to be correctly acknowledging the new tasks by sending them to the server with state PENDING.

We will be reporting this latest bug to Topfield today but in the meantime I would urge you to update to v1.11.06 as soon as possible. We will also disable the process that sets the state to MISSING to give people time to update their firmware.

[tonymy01]

Very interesting post Daniel!   Can we please have a summary like this in the Beyonwiz threads (assuming it does things a tiny bit differently... or not?)  Someone mentioned the other day about getting recording failures reported by ICE, I thought they must have meant timer failures, but this protocol you describe here also tracks recordings!  Would be nice to get a full run-down of what the system is capable of, or at least what has been provided for the various units out there.
Regards

[Vortical]

Thanks Daniel

I'm glad you've found the explanation as to why it's happening.

Hopefully Topfield Korea can rectify it soon.