feature idea: trigger load of nodeinfo for single node

[mar-v-in]

It would be awesome if there was a way to trigger the hopglass-server to load a nodeinfo update "out-of-order". This could be used by gateway server once a node connected, allowing "live update" of the map.

Although the delay usually is less than a minute, some people are eagerly refreshing the hopglass after connecting the node. As hopglass-server creates a new json for each request anyway, creating them "out-of-order" is not a deal (but was back in times of ffmap-backend).

This can also be used if nodeinfo is updated when the node is "old", so that it won't take 5 minutes (default settings) to update.

[ghost]

So when exactly is the server supposed to get nodeinfo? It doesn't even know the node's ip yet..

[mar-v-in]

I was thinking of some kind of trigger API. Of course hopglass does not know the IP of a node before it knows the node, but that's not necessarily the case for all other components of the system. e.g. you can know the IP through batman (via mac address->ipv6 address calculation). Or the node could actually trigger the update itself by notifying the server.
So I thought of some new http request /trigger?ip=<some-ipv6>. Not sure about possible security implications, if this should be restricted somehow (atleast the ip address should be limited) etc, but I hope you understand the idea.

[ghost]

I understand the idea, but I don't get why you would want that.

[mar-v-in]

Faster map updates.

Currently the situation is much worse than at the beginning of hopglass-server, because we are no longer requesting for nodeinfo on unknown nodes. This leads to unresolved nodes being shown in the graph and in prometheus, which could be resolved at that moment. I guess we should really bring the feature back which triggers nodeinfo request on devices that sent neighbours or statistics (independent from this issue). This would solve some of the problems of delayed map updates.

Nonetheless I still think it would be awesome to have live map updates, including the possibility to push updates to the hopglass web client via websockets. Just imagine you are plugging in a router and it pop ups on the map only seconds after configuration ;)

[ghost]

I don't understand how just adding a trigger feature will improve anything.

we are no longer requesting for nodeinfo on unknown nodes.

we can change that again, now that you mention it

I agree about the push part and I have tried to implement that into respondd, but my C skills are very limited.