[mythtv-users] mythbackend still eats memory: the current status

Allan Stirling dibblah.allan.stirling at googlemail.com
Sun Jan 25 23:12:05 UTC 2009

Udo van den Heuvel wrote:
> Allan Stirling wrote:
>> The patch in 19705 had nothing really to do with EIT parsing. It was 
>> in the handling of freeing memory for NITs and SDTs.
>> Have you actually tried disabling EIT?
> It was my misunderstanding/miscommunication then.
> I throw EIT and NIT/SDT (?) on the same heap in my unfamiliarity. ('just 
> another packet...')
> I ran the backend without EIT when I found out about the 'eats memory' 
> issue.
>> Myth actually has some support for increased buffers to work around 
>> increased latency caused by valgrind:
>> ./configure --enable-valgrind --compile-type=profile
>> Can you please try these on your production system, for a short (4 
>> hour) run?
> Does building with these options cause issues for a normal run?
> If not I could add these options by default for my next build.
It shouldn't do - All it does is increases the timeouts and 
associated buffers, as far as I know.

I would suggest building this copy of Myth in a seperate 
prefix, if you are uncertain. For the backend, just 
configuring with --prefix=/usr/local/mythtest, compiling and 
installing then running /usr/local/mythtest/bin/mythbackend 
instead of your usual backend process should isolate this 
environment. Obviously, this does not isolate the database, 
however, you made backups, right...? ;)

>> I would suggest that without DVB cards matching your production 
>> system, a test system will not reproduce this leak.
> Is the driver for the card causing the leak you mean?
> Or is the data handed over from the card to the backend causing this?
> Please explain.
The data from your provider combined with your particular 
use-case is probably causing the leak, since it is not seen 
on many (if any) other systems.

>> It may also be valuable to back up your database and try current trunk.
> I ran trunk for a long while but was then cursed^H^H^H^H^H^Hasked to try 
> -fixes only to find that -fixes had the same issues.
> So I am still on fixes beause -fixes also provides me with multirec and 
> other features  which wasn't the case in the past.
It is, of course, up to you. I was just suggesting options.



More information about the mythtv-users mailing list