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

[Gavin Hurlbut]

On Sun, Nov 28, 2010 at 12:37 AM, Udo van den Heuvel <udovdh at xs4all.nl> wrote:
> The ps output is just for monitoring.
> If you still want to deny we have leaks, i.e.: only accept valgrind logs
> and until then do as if nothing is wrong, you are free to do so.
> I did not say anything that says otherwise.

The only way to identify and fix memory leaks is with a tool like
valgrind.  In fact, you have shown with your ps listings that no
significant memory IS leaked (or the VMEM would be constantly
growing).  The RSS column is irrelevant.  That just shows that more of
the process is resident in memory.  That is not a leak, that is the
virtual memory system leaving larger chunks in memory as they are
active, etc.

> So apparently the memory leak situation does bother you.
> It also matters to me, so soon we might get some logs.

If you are going to claim memory leaks, then the onus is on you to
prove it.  The only proof that is worth anything is valgrind logs (or
another similar tool).  ps snapshots don't prove anything
conclusively.  We've been over this with you before.

Many of us have much heavier loaded MythTV systems than yours, and we
have not noticed memory leaks of any importance.  Get us proof of your
allegations, thanks.