Deploy seed nodes to assist p2p network discovery

[bisq-network]

Motivation

Server-side seed nodes are necessary such that that first-time Bitsquare users can "bootstrap" into the p2p network, i.e. discover enough other nodes and communicate with them. From that point forward, clients maintain their own list of last-known-good peers, meaning that clients no longer rely on the seed nodes.

Next steps:

• Steve to merge seed node command line startup from the seedpeer branch (eaaa0e0)
• Steve and Manfred to get their local routers working with UPnP and Bitsquare (per Deploy seed nodes to assist p2p network discovery #3 (comment))
• Wait for Thomas Bocek to look into TomP2P issues and get back to us (per Deploy seed nodes to assist p2p network discovery #3 (comment))

[bisq-network]

I tried again with the current master to test the serverside bootstrapping, but it still has the same issue that the relay fallback causes an exception on the server. Thomas Bocek told me on the mailing list that he will look into the issue. There is also a warning with the signing for data protection. That seems to be another open bug, he will investigate.
I added code that the last successful bootstrap method is saved and reused so a repeated connection in relay or port forwarding mode will be faster.

[notmandatory]

I'll investigate after I've got my akka prototype going.

[bisq-network]

Probably we need to wait until Thomas has fixed the problem...

cu,
manfred

Am 22.08.2014 um 20:51 schrieb stevemyers [email protected]:

I'll investigate after I've got my akka prototype going.

—
Reply to this email directly or view it on GitHub.

[ManfredKarrer]

@cbeams Thanks Chris! I will try it on my own first, if I have troubles I will ping you...

[ManfredKarrer]

Good news from Thomas, he said we think he can fix the open problem with directSend (UPnP) until mid of the week.

[ManfredKarrer]

@stevemyers @cbeams @remyers

Hi Steve,

Great!
Is directSend working for you?

Good to have the master seeds connected. I will add a test where one peer connected to my server stores data and the 2. peer connected to your server read the data.
Will check afterwards to get the offer storage working again in the app and to change back to async handling so the app does not freeze during startup.

Maybe we can extend later the tests in the way that we have a number of normal peers (5-10) running on the server and then check if shutdown of nodes will affect the DHT storage. It should automatically manage the storages, so a leaving node will be replaced by others. As long 1 node remains the data should be accessible.
Would be cool if we could manage to get a test environment (#51) which runs those scenarios automatically. Will be a bit complex though, but I think we should have something in the WAN for testing real life conditions.
Maybe there are some test cases in the tomp2p lib we can extend. There is also a sim grid (https://github.com/tomp2p/TomP2P-SimGrid) test setup. I have not looked into that yet, but I think it is only for local testing of massive nodes (10 000s). Would be more interesting to get something like that for the WAN.
But of course that complex test environment is no hi prio yet.

Do you have any ideas how we should go on with the directSend problem? I hope Thomas can fix it soon, he wrote that he thinks he can fix it until mid of the week, so chances are good that we get it soon. Otherwise probably its worth to dive deeper into tomp2p code, will probably be not the last problem and better to be not too dependent on the help of Thomas. Though that will be a time consuming adventure... but I think we need to go that path. What do you think?

[ManfredKarrer]

Good news:
Direct send and DHT in relay mode is now working in the latest head of the master (just tested in our test class). Unfortunately I dont get port forwarding anymore. Will investigate further...

[ManfredKarrer]

Trade process with relay mode works now. Offer storage + messaging!
Will continue to find out why port forwarding does not work anymore. And there is a problem with offer removal. I will wait update our fork soon.

[ManfredKarrer]

Added tests for add (used for offers to add an offer to a list), remove and all tests are green with relay. There are 2 small issues with incorreect success results, whcih i will report Thomas. I will update the our fork to the master head.

[ManfredKarrer]

Port forwarding is working again, dont know why it was not working before. Maybe my router behaves strange with UPnP? But there are still problems, DHT basically works but delivers wrong success flags, and directSend is still not working. If we dont get it working we can deactivate it so users can only connect directly or via relay mode (just as temporary solution).

[ManfredKarrer]

I deactivated port forwarding for the moment as that is not working correctly. In relay mode startup is slower but all works so far as I have tested.