[mythtv] Failed recording marked recorded anyway

[Joseph A. Caputo]

Bruce Markey wrote:

>Joseph A. Caputo wrote:
>  
>
>>DISCLAIMER:  I am running SVN from just before the LiveTV overhaul 
>>(revision 7569), but I've been keeping up on the -dev and -commits list 
>>and haven't seen anything addressing this.  Please beat me with a wet 
>>noodle if this has actually been fixed.
>>
>>
>>OK, so in fiddling with my system recently I apparently screwed up my 
>>second tuner device, so everything that's been scheduled to record on 
>>that card has silently failed.  Worse, though... even though the 
>>recordings failed, they were still inserted into the 'oldrecorded' 
>>table, so later showings didn't get recorded.
>>
>>Here are some relevant entries from my logfile (the stuff about not 
>>finding the input are because I screwed up my second tuner):
>>
>>****************************************************************************************
>>2006-01-12 20:57:03.554 TVRec(1): Changing from None to RecordingOnly
>>2006-01-12 20:57:03.931 ChannelBase: Could not find input: Tuner 0 on 
>>card when setting channel 4
>>2006-01-12 20:57:03.934 TVRec(1) Error: Failed to set channel to 4. 
>>Reverting to kState_None
>>2006-01-12 20:57:03.936 TVRec(1): Changing from RecordingOnly to None
>>2006-01-12 20:57:03.967 scheduler: Finished recording
>>2006-01-12 20:57:03.991 scheduler: Started recording
>>2006-01-12 20:57:10.828 scheduler: Scheduled items
>>2006-01-12 20:57:10.863 TVRec(1): Changing from None to RecordingOnly
>>2006-01-12 20:57:11.228 ChannelBase: Could not find input: Tuner 0 on 
>>card when setting channel 68
>>2006-01-12 20:57:11.231 TVRec(1) Error: Failed to set channel to 68. 
>>Reverting to kState_None
>>2006-01-12 20:57:11.233 TVRec(1): Changing from RecordingOnly to None
>>2006-01-12 20:57:11.262 scheduler: Finished recording
>>2006-01-12 20:57:11.283 scheduler: Started recording
>>2006-01-12 20:57:16.043 scheduler: Scheduled items
>>****************************************************************************************
>>
>>You can see that it tried to start a recording on Channel 4, failed, 
>>then (since it thought a tuner had freed up), it tried to record a 
>>program that had previously been conflicted on Channel 68, and failed 
>>that one as well.  Yet, both programs now have entries in the 
>>'oldrecorded' table, so later showings of either program get skipped as 
>>duplicates.  Bad!
>>    
>>
>
>I was all psyched up to answer your question but you didn't ask one.
>So is this just a complaint that you're not going to lift a finger
>to do anything about and just brow beat the people who spent hundreds
>of man hours creating what you have so far that you downloaded for free?
>Joseph, you should know better. I'll try to answer the questions you
>should have asked.
>  
>

Sorry if I was unclear... the 'question', though, should be obvious: is 
this behavior known, and if so, has it been fixed or is it under 
investigation?  If this hasn't been noticed before, then this serves to 
bring it to the attention of the community at large.  If this is a 
new/newly noticed issue, I can start taking a look at the code, if 
nobody else wants to tackle it.

>Is anything being done to track failed recordings on oldrecorded?
>Yes. One of the biggest invisible changes between 18 and 19 is that
>there were a lot of changes related to how we keep track of shows
>during their airtime, what is stored in oldrecorded and how that
>can be reused, what happens when master and slave backends restart
>and it all leads to the ability to change the end time of recordings
>in progress. Part of this is a new state rsAborted to mark a failed
>recordings to make you aware of the problem, allow reactivate, allow
>re-record and display this info after the fact on the Previously
>Recorded page. This state is set certainly for a backend crash and
>should be incorporated for other recording failures that can be detected.
>  
>

>Do I have to cower like a little girl and use SVN from before the live
>TV changes? [You would have been better off phrasing this for yourself.]
>  
>
A bit harsh!  Actually I held off updating for a good 6-8 weeks while 
the LiveTV update was stabilizing; then the holiday season began and I 
just didn't have the time to fiddle with my system... now it's after the 
holidays and I have a little time so I'll probably upgrade this weekend.

>No. It's been working just fine for quite a while. There are rough
>spots to smooth out like anything else but the last major piece, the
>transition from live to scheduled recording, went in this week. A lot
>of other fixes have gone in in the meantime possibly related to the
>issues you had above. Current SVN is in much better shape than 7569.
>
>  
>

>How can I fix the oldrecorded entries for these shows that didn't
>record? If you'd found the problem from trying to play the recordings
>from the Watch Recordings page, I would hope that you chose the
>allow re-record option (duh =). Generic showings aren't supposed to
>have this button but this doesn't always work out right.
>
No, they didn't show up in the "Watch Recordings" page.  That's part of 
the problem... if I had seen a recording listed but it couldn't play it 
because it was a 0-byte recording, or some other problem, then I would 
have noticed sooner.  No files were created for any of the failed 
recordings, no DB-logged entries to the effect that the recordings were 
attempted, nothing.  The only reason I noticed the failed recordings was 
because I was actually *looking* for it and the repeat showing for next 
week was deactivated as "previously recorded", which got me poking into 
the logfile.

> You can
>always go to the Previously Recorded page. Sort by time may help
>cluster the shows in question. The 'Enter' key shows buttons to
>delete the one entry or all entries for that title. Also, on the list,
>"D" is a shortcut to delete the one highlighted item. If you're not
>sure of which shows were affected, go through your backend to find
>the times and channels when you had this problem. If you use "-v record"
>you would see the titles also.
>  
>
Right, except there was no indication (except in the logfile & my 
recollection that something should have been recorded) that anything had 
gone wrong.

Anyway, thanks for your reply.  I'll try and see if this is remedied in 
current SVN, but as I've fixed the underlying problem with my tuner I'll 
have to figure out if I can reproduce the conditions now...

-JAC