[mythtv-users] Database damage?

Stephen Worthington stephen_agent at jsw.gen.nz
Sun Feb 14 09:22:09 UTC 2021


On Sun, 14 Feb 2021 08:35:44 +0100, you wrote:

>On 14/02/2021 03:29, Stephen Worthington wrote:
>>> Success!  I first tried a sleep interval of 2 seconds and it crashed after ~40 deletions.  I upped the interval to 4 seconds and it processed 120 zero byte recordings.  With an 8 second interval, it trundled through the remaining recordings and finished without error.
>>>
>>> I guess I’ve done my spring cleaning early this year!
>>>
>>> Thanks very much.
>>>
>>> Craig
>> 
>> Excellent!  Mythbackend is pretty slow at doing deletions, so my
>> recommendation of 10 seconds was based on my experience with that,
>> plus a bit to be sure.  Fortunately, MythTV is pretty forgiving of
>> crashes, but it would pay to run a full database check again, just in
>> case.
>
>Surely the crash is still a backend bug though...

Yes, but any time mythbackend crashes, there is the possibility it did
it in the middle of a database operation.  So there is the possibility
that a table is crashed.  A worse but virtually impossible to detect
error is also possible, where it updated one table (eg recorded) but
failed to update the matching data in another table (eg
recordedprogram, recordedseek).  Since it was doing deletes at the
time, that might mean that some of the recording tables have data for
the recording being deleted at that time, but others have had it
deleted.


More information about the mythtv-users mailing list