[mythtv-users] mythbackend still eats memory: the current status
Udo van den Heuvel
udovdh at xs4all.nl
Tue Nov 30 14:25:21 UTC 2010
On 2010-11-29 19:24, Gavin Hurlbut wrote:
> On Mon, Nov 29, 2010 at 9:13 AM, Udo van den Heuvel <udovdh at xs4all.nl> wrote:
>> That might be the case but that number grew since that period.
>> Why?
>> The app has been recording for a while before then and since then so why
>> would it allocate more memory?
>
> Because other things may have changed... Like scheduling
> information.. Who knows. That in and of itself is not indicative of
> a memory leak.
Growing over a longish time, more or less constantly indicates a leak.
>> Same feed as ever, same updating schedule, same source, same multiplex,
>> same stations, etc, etc. Same.
>
> Really? So your stations air the identical show lineup every day? I
> didn't think so.
So Myth doesn't free stuff after e.g. a day and fills memory with the
info for the next day?
It just keeps on stuffing schedules into RAM?
>> You might as well explain why it isn't leaking, that would be easier.
>> I just see numbers increasing when I do not expect them to be increasing
>> (after a `short` while of running...).
>
> If you can't show (with the proper tools) that it is leaking and
> where, you might as well just give up.
You are turnign around without answering the question asking you to explain.
> I can't use a crystal ball and
> tell you where in the hundreds of thousands, if not millions, of lines
> of code there is a leak.
I did not ask where, I asked why it isn't leaking.
> Asking me to explain why your accusation is false is bogus. You make
> the accusation, you prove it.
Again data points in the otehr direction.
If you do not accept the data that is your issue.
I do not expect RAM usage to grow after a certain period. A day, a week,
soemthign like that. But it does.
> And proof means valgrind output, not ps
> snapshots.
I did not indicate proof, I did indicate an indication.
>>> kilobytes, which really isn't terribly significant.
>>
>> A leak is a leak.
>
> Whatever.
You have the correct approach here.
Kind regards,
Udo
More information about the mythtv-users
mailing list