[confirm] Development issues (3 Viewers)

[morpheus_xx]

Ah ok, then this is the more probable case. For me it also takes a while until the multiseat server reconnects again. This I also tried to improve, but didn't find a way to force a quicker reconnection

 

[disaster123]

What about triggering an explicit reconnect if the connect fails? Which file / code is it?

 

[morpheus_xx]

This is not that easy.

I have already tested to shutdown all UPnP services in suspend and restart them on resume. Even this didn't lead to immediate reconnects for me.

But it might be a local issue I have here: the HTPC (server) is only found via it's IPv6 address, while the IPv4 is not found. I expect some additional troubles by this.

 

[disaster123]

But can you Point me please to the code where the connect and requests are handled?

 

[morpheus_xx]

Let's start here, from TV provider. On startup it registers the UPnP proxy device:
https://github.com/MediaPortal/Medi...der/UPnP/NativeTvProxyRegistration.cs#L60-L66 at the control point
Method is defined here: https://github.com/MediaPortal/Medi...unication/UPnPClientControlPoint.cs#L121-L132

The registered delegate is used when a new root device is added: https://github.com/MediaPortal/Medi...unication/UPnPClientControlPoint.cs#L153-L159.

This event is invoked by the UPnPNetworkTracker: https://github.com/MediaPortal/Medi...erCommunication/UPnPClientControlPoint.cs#L65, which is defined https://github.com/MediaPortal/Medi...P/Infrastructure/CP/UPnPNetworkTracker.cs#L45.

The UPnPNetworkTracker also has a "Search" method to lookup specific or all UPnP devices. It is executed at init of class. One idea I had already is to submit a new Search after resume to speed up connection, but this didn't work.

 

[disaster123]

I don't know that much about UPnP may be there's something special but normally shouldn't it work like this?

Client <=> Server - Status: connected
Client suspend - close connectiion / upnp

Client Resume try to reconnect to the same server without a research (may be in a loop of 5-10 retries)

So why is a new search needed at all?

Thanks for all the links i'll have a look at it.[DOUBLEPOST=1452105806][/DOUBLEPOST]Mhm i'll try to add some debug code first to get an idea what happens. The UPnPNetworkTracker doesn't have any logging right now.

 

[disaster123]

The ServerConnectionManager seems to be interesting too or is it not relevant?

 

[disaster123]

@morpheus_xx can you play recordings after standby?

If i remove the upnphelper / this code:
// _upnpServer.UpdateAndAdvertise();

it works again. It seems this does not work.

 

[morpheus_xx]

@morpheus_xx can you play recordings after standby?

If i remove the upnphelper / this code:
// _upnpServer.UpdateAndAdvertise();

it works again. It seems this does not work.

Really? This is the change based on your patch, only extracted a new method from the network connection event handler...

 

[disaster123]

Yes i know. I was focused on livetv while implementing this but never tested recordings and never had the idea that this could break them ;-( still no idea what's the reason. I'll try to find out.