[mythtv-users] Web interface scalability issue: Video Gallery

[Paul Harrison]

On 11/09/2023 04:07, f-myth-users at media.mit.edu wrote:

> I haven't looked at the code and I haven't tried the setup, but one
> thing that jumped out at me in following this discussion is that 10.01
> second timeout in the web timing chart, plus statements like
>
>> For me the slowest requests always spend just over 10 seconds
>> waiting. The file that is slow isn't always the same one it seems
>> to be random which one is slow
> I'm wondering if this isn't CPU load but some sort of 10 second timeout
> somewhere, or maybe two five-second timeouts in opposite directions,
> etc.  But I don't know where those might be coming from.  Local DNS/DHCP
> resolution?  Obviously this is a total guess, but I'm suspicious.


I think you are correct the fact the slow requests all take just over 10 
seconds can't be a coincidence.


It's strange the only thing that appears to have the problem is the 
WebApp which makes me think the problem is in the BE server but why 
aren't others seeing it. It could be like de-interlace artifacts or mpeg 
blocking on dark backgrounds most uses can blissfully go through life 
and never see them but once someone points them out to you that is all 
you can see. Maybe I'm just seeing the delays  because I know they are 
there.


Another thing I just thought of is it depends on what request is being 
delayed if it is one of the js files or css files etc then that is bad 
because the page has to wait for the data to arrive before it can start 
to render. If it's an image or API call then the page can usually start 
to render more quickly you just see a space where the image should be or 
maybe the loading spinner continues for a little longer than normal.


> Is there any possibility you could run wireshark?  You'd have to have
> some way of capturing the packet stream, but if you could do it on the
> affected host(s), or perhaps if you have a smart switch where you could
> clone traffic on one of the ports (or a really dumb hub :), a packet
> trace might shed some light.


Stuart A. and I did start to debug this back 18 months? or so ago when I 
first noticed it but circumstances changed and we never got very far. It 
was Jen's message that reminded me about the problem and I was just 
wondering if it was the same problem but it is clearly not. One rainy 
day when I'm bored I may try to debug this further.


Paul H.