@folaht created a local network on his machine and wanted to make it available to others.
Nobody could access it and @bochaco explained this possibility needs features provided by a new PR
I looked at this PR and I found that it added a secondary IP address to an endpoint. Previously I successfully created a local IPv6 network but it was exclusively IPv6 (an IPv4 nodes cannot connect to it). So I said that the secondary IP address could be better used to create a mixed network (managing both IPv6 and IPv4)
@dirvine explained that allowing mixed nodes is much more complex (and I agree with this)
Another problem with an IPv6 network is that an IPv6 client cannot connect to it using usual ~/.safe/node/node_connection_info.config file
Then I looked into sn_client code and I found that another config file could be used (~/.safe/client/sn_client.config). After some hacking I found out the right syntax of this file to allow an IPv6 client to connect to an IPv6 network.
As soon as @folaht saw this possibility, he turned his network to an IPv6 network and updated the OP with new directives to allow others to connect to it.
After some trials (ports had to be opened for all the elders) I could access his network
This means that current Maidsafe code allows creation of an IPv6 network with nodes and clients over the Internet.
Oh, wow! I hadn’t initially realised that was the case, out of the box. Considering all the issues with nat/upnp with ipv4, this sounds like pretty good news. I think I may need to brush up on my ipv6 understanding!
It’s another reason for us to look at ipv6 as soon as feasible. The quandary is that where ip6 is available then we get many more folk able to run a node, but the global coverage of ip6 is still not great and it’s worse in poorer areas. So Secure Access For Everyone is challenged. IP4 is a nightmare due to NAT and ip6 is a nightmare due to coverage.
With many recent changes though we have a really nice feature that’s currently not visible to us all. This is the CDRT data types, the mutation is signed by the “owner” or defined by policy. So it does not matter how the message gets delivered or by whom (so src becomes irrelevant). As messages are currently (almost but not quite just yet) we don’t care what route they take to deliver, even via flash drive and postal service
AT2 and BRB are a bit different though, but in any case this is something we should all be aware off, mixed network may be possible quite soon.
Ipv6 is the future and I’m sure coverage will increase, especially if there are good use cases for it. If it avoids having to hack around ipv4 limitations, I’d say it is worth it for that alone.
Given safe network doesn’t need the various software infrastructure above (e.g. DNS) and won’t have dependencies on other legacy connectivity, this seems like a no brainer to me.
That said, I’m not aware of how bad the coverage is. It seems like something that will only improve with time though.
That is the big question. At least with IP4 you can still manually port forward. There are ip6 tunnels etc. but servers man, servers and proxies. It’s an interesting area though and one we need to choose wisely
Maybe we shoudl run an IPv6 testnet and see how that goes? It might offer more people the chance to start nodes and earn test-safecoins? Put the IP4 NAT issues behind us? What do you think @lionel.faber et al’
Please look at ipv6 adoption by countries world wide.
The top countries have around 50% adoption, Countries like Sweden has 6-8%. How would that affect farming if only 6-8% of an population can get an ipv6 address?
David, please dont even think about it until the planned testnet is released. Please no more bumps in the road.
Dont delay what you are doing for a feature that, no matter how desirable, has hardly been mentioned until a few days ago. Let the community work on it without distracting you lot in any way. We promise to let you see the logs later
Deliver what we have all been waiting for and seems so near.