Is UPnP a rubish protocol ???

[tarnkappe]

Also it might help to turn off all network related devices and then restart one after another, starting with the internet router.

[lyapounov]

(15-09-2022 11:16)tarnkappe Wrote:  Also it might help to turn off all network related devices and then restart one after another, starting with the internet router.

Yes, that might help.
However, I have altogether 24 devices, including switches, electric plugs, synology servers etc. etc. So no.

I am just thinking what happened since last time it was working.

And the main thing is a new iPad. But why so ?

After this was a cascade of switches, the last one being a Buffalo. I took the Buffalo off. Hooray, my iPad was no longer disconnected.

Except that it lasted 3 days. At the end of the third day, back to square one; the iPad regularly loosing the rendered, and without any change on my network.

In a nutshell: UPnP is a rubish protocol. Comparing to DHCP as an exemple: I never had any issue with DHCP. I keep on having issues with UPnP

[lyapounov]

(11-09-2022 18:22)simoncn Wrote:  In a situation like this, it is often helpful to change (back) one thing at a time until you find which change is making the difference.

I did put aside my Buffalo switch. Things went back to functional. For three days.

And after three days, problem has come back.

[stefano_mbp]

… comparing UPNP and DHCP seems quite … senseless …
In my network Minimserver and endpoints are all connected to the same switch that is connected to the router and I’ve never had any issue.
You should give a look to your network topology: it’s not a good practice to have switches in cascade, each switch should be connected to the center of the star (ie the router), in this way you’ll avoid speed issues.
… and it should be better to have switches of the same vendor/model to avoid various protocol issues

[lyapounov]

(15-09-2022 15:35)stefano_mbp Wrote:  … comparing UPNP and DHCP seems quite … senseless …
In my network Minimserver and endpoints are all connected to the same switch that is connected to the router and I’ve never had any issue.
You should give a look to your network topology: it’s not a good practice to have switches in cascade, each switch should be connected to the center of the star (ie the router), in this way you’ll avoid speed issues.
… and it should be better to have switches of the same vendor/model to avoid various protocol issues

1) switch in cascade is a not only a very common practice in the hifi world, which has shown good value, but also in a lot of corporate companies. This is the beauty of all those protocols to be decentralized and to avoid to put the load to the router (can you imagine how many ethernet points are in a corporate ? Many hundreds. Do you know of a router which has many hundred ports ? No, that does not exits. So switches are in cascade) But in my case, I have only two switch cascading; and the last one has only my server and my rendered; this to have the shortest path between the two and no influence from other things. If you google it, you will see that it is a good practice to have one switch at the end with only the server and the renderer

2) you are the very first one to tell that mixing different vendors means speed issues. I myself have always had multiple vendors in my network without any problem. Do you have literature which supports this totally incredible comment ? Again, the beauty of the internet is open norms and protocols ... Can you name a protocol which is different according to the vendor ?

But again, in the Internet world, you add a new device and bing, the DHCP assigns a new IP adress, and the device is there forever.

On the contrary, UPnP is unable to keep the communication. Voilaà

[stefano_mbp]

I have been the CIO in many large companies therefore I know quite well what I’m talking about, and I’ve many examples about how can be easy to let a network go nuts, different firmware on different stacks is the first example (and we are on the same vendor … try to think about different vendors switches).
For first in a large company there never are cascades of switches but “stackable switches” are used (like Cisco Catalyst 9300 series) … “stackable” is totally different from “cascade”
About speed : let me assume you have a 1Gbps network and switches according to this speed, then on each port will be available 1Gbps, but if you connect a switch in cascade then the original 1Gbps will be shared by all ports of the “cascade” switch hence the speed available on the cascade switch ports will be less than 1Gbps. It’s quite easy to start with 1Gbps and end up with less than 10Mbps
About protocols: each vendor will set up the firmware according to … his choices, therefore one vendor could block UPNP or something else. This is for unmanaged switches, different story for managed switched where you can (but you must be able to) define anything as you like.

[stefano_mbp]

… and about UPNP:
this is what I found “ A hop is the number of steps allowed to propagate for each UPnP advertisement before it disappears. The number of hops can range from 1 to 255. The default value for the advertisement time to live is 4 hops, which should be adequate for most home networks.”
… each switch is a hop and Wi-Fi access point is a hop too … think about it

[lyapounov]

(15-09-2022 16:54)stefano_mbp Wrote:  … and about UPNP:
this is what I found “ A hop is the number of steps allowed to propagate for each UPnP advertisement before it disappears. The number of hops can range from 1 to 255. The default value for the advertisement time to live is 4 hops, which should be adequate for most home networks.”
… each switch is a hop and Wi-Fi access point is a hop too … think about it

Hmm, so you mean there is an impact on the number of hops on the UPnP protocol.

That I should investigate; it could explain. Question: when the renderer and the server are on the same switch, but this switch is 3 hops to the router. How do you compute the number of hops in this case ?

After what you wrote, I would be ashamed to describe to you my network architecture ;-)

[stefano_mbp]

UPNP server (ie Minimserver) must be reached by you control point (tablet/smartphone) … where is Wi-Fi access point? If it is the router there should be 4 hops but if it is elsewhere … the hops could be more than 4

[simbun]

(15-09-2022 18:50)stefano_mbp Wrote:  UPNP server (ie Minimserver) must be reached by you control point (tablet/smartphone)

And I assume the renderer too - although in this case they're on the same switch, so if you can see one you can see both.

I tried to get SSDP and mDNS working across subnets a few years ago now (I never did get UPnP working even with the help of BubbleUPnP) but from memory I don't think switches do reduce the TTL do they; Layer 2 vs Layer 3 I think, although I would agree it'd be much better to reduce the hops\switches.