[MB Premium Twin HD] Enigma looping on signal error

[birdman]

I've just noticed my box was displaying the spinner, unresponsive to the remote and not recording.

Here's the debug log: View attachment Enigma2-25-07-2015_20-46-21.log

NOTE that the SIGSEGV signal at the end put the whole thing into a loop - it was generating the same signal ~200 times a second (from looking at the later debug log, which is just a series of repeating errors) until I logged into the box and rebooted it.

 

[birdman]

Following on from the reboot enigma2 was crashing and restarting about every hour.
The debug log indicated this was happening when reloading epg.dat (which it is set to do every hour).
I've deleted the cache file and rebooted, so will see how it goes.
But the process should be able to handle a corrupt data file without crashing - it should just ignore the malformed data.

 

[bbbuk]

I don't believe it would be possible to ignore the corrupted data as it would be impossible I would have thought.

However, I would guess it would be possible to handle errors bit more gracefully and display error message and then maybe delete epg data.

 

[birdman]

I don't believe it would be possible to ignore the corrupted data as it would be impossible I would have thought.

However, I would guess it would be possible to handle errors bit more gracefully and display error message and then maybe delete epg data.

Your second sentence shows that your first sentence is wrong - you do believe it is possible.
Any program that crashes has a bug (unless it was written with the express intention of producing a crash, e.g. for testing purposes).

 

[bbbuk]

Your second sentence shows that your first sentence is wrong - you do believe it is possible.
Any program that crashes has a bug (unless it was written with the express intention of producing a crash, e.g. for testing purposes).

I don't think it does. It's probably how I read your second post

How I read your second post, you wanted it to ignore the malformed data (ie strip the malformed data out) and thus read the epg data after stripping it out!!

This is why I replied I don't think it could be easily done.

However, gracefully handling any error message that could be generated trying to read a corrupted EPG and maybe prompting about corrupted epg data and giving option to delete it, I would like to think is possible in python to code.

 

[adm]

Following on from the reboot enigma2 was crashing and restarting about every hour.
The debug log indicated this was happening when reloading epg.dat (which it is set to do every hour).
I've deleted the cache file and rebooted, so will see how it goes.
But the process should be able to handle a corrupt data file without crashing - it should just ignore the malformed data.

This may be similar to the problem I reported a few month back where my box would gradually grind to halt with a constantly spinning VIX and no response to the remote. The debug/error log was Gbytes and continually being written at speed with blocks of 32 hex formatted words. I suspect the log would have filled the disk if I hadn't manually switched off the box and deleted the EPG on next switch on. The gigantic logs were all triggered after a report of a EPG cache problem (the report in the log file). The only way I managed to stop this behaviour was to disable the Automatic Refresh and Automatic save options in the EPG settings. I originally had them on a 3/4 hour cycle.

 

[birdman]

This may be similar to the problem I reported a few month back where my box would gradually grind to halt with a constantly spinning VIX and no response to the remote. The debug/error log was Gbytes and continually being written at speed with blocks of 32 hex formatted words.

I've set a limit on my debug log size, which seemed to be honoured - I have a 5MB log file, but it has definitely had entries missing from the "start" (that 5MB only covers 18s of log entries). From what I can see a new log gets started at each enigma2 restart, so normally I don't even get to 1MB (and old logs are being deleted after 10 days, so the whole thing looks after itself).

Yes - the problem sounds the same. Deleting the EPG cache resolved the immediate problem.