Globally Distributed Elixir over Tailscale
richardtaylor.dev
Globally Distributed Elixir over Tailscale
1–10 of 21 posts
Re: Globally Distributed Elixir over Tailscale
#2I would be very curious to see how well this actually works in different data centers. My understanding of empd is that it basically expects all nodes will be on the same switch. I don't believe there is any default handling for spikes in latency or netsplits, and epmd is known to be pretty chatty.
The common wisdom seems to be to do federation of clusters rather than clustering multiple data centers together.
Re: Globally Distributed Elixir over Tailscale
#3Very cool concept.
Reminds me of the intended use case of Nebula, which seems very similar to this. If you're interested in a bare-bones and totally self-hosted option, it could be a good choice here. https://github.com/slackhq/nebula
Re: Globally Distributed Elixir over Tailscale
#4Very cool concept. Reminds me of the intended use case of Nebula, which seems very similar to this. If you're interested in a bare-bones and totally self-hosted option, it could be a good choice here. https://github.com/slackhq/nebula
There’s a reimplementation of the Tailscale control plane called Headscale, if you want to self-host.
I’d try to use that first, because sadly Nebula and ZeroTier don’t have a relay/TCP/HTTPS fallback option and still “have no plans of implementing one” as of Mar 2023, which leaves you out of luck as soon as you encounter NATs or try to access your cluster from an airport/hotel wifi
Maybe you know of other mesh/p2p VPNs that do support TCP fallback though? Would be great to see some alternatives in this area
Re: Globally Distributed Elixir over Tailscale
#5Very cool concept. Reminds me of the intended use case of Nebula, which seems very similar to this. If you're interested in a bare-bones and totally self-hosted option, it could be a good choice here. https://github.com/slackhq/nebula
There’s a reimplementation of the Tailscale control plane called Headscale, if you want to self-host. I’d try to use that first, because sadly Nebula and ZeroTier don’t have a relay/TCP/HTTPS fallback option and still “have no plans of implementing one” as of Mar 2023, which leaves you out of luck as soon as you encounter NATs or try to access your cluster from an airport/hotel wifi Maybe you know of other mesh/p2p V…
Re: Globally Distributed Elixir over Tailscale
#6Very cool concept. Reminds me of the intended use case of Nebula, which seems very similar to this. If you're interested in a bare-bones and totally self-hosted option, it could be a good choice here. https://github.com/slackhq/nebula
There’s a reimplementation of the Tailscale control plane called Headscale, if you want to self-host. I’d try to use that first, because sadly Nebula and ZeroTier don’t have a relay/TCP/HTTPS fallback option and still “have no plans of implementing one” as of Mar 2023, which leaves you out of luck as soon as you encounter NATs or try to access your cluster from an airport/hotel wifi Maybe you know of other mesh/p2p V…
we're focused on this very thing - https://bowtie.works
Re: Globally Distributed Elixir over Tailscale
#7Very cool concept. Reminds me of the intended use case of Nebula, which seems very similar to this. If you're interested in a bare-bones and totally self-hosted option, it could be a good choice here. https://github.com/slackhq/nebula
They actually have a (semi) hosted solution these days.
I evaluated it for one of my projects but it was not fitting the bill for reasons outside of their control, https://www.defined.net/
Re: Globally Distributed Elixir over Tailscale
#8Very cool concept. Reminds me of the intended use case of Nebula, which seems very similar to this. If you're interested in a bare-bones and totally self-hosted option, it could be a good choice here. https://github.com/slackhq/nebula
There’s a reimplementation of the Tailscale control plane called Headscale, if you want to self-host. I’d try to use that first, because sadly Nebula and ZeroTier don’t have a relay/TCP/HTTPS fallback option and still “have no plans of implementing one” as of Mar 2023, which leaves you out of luck as soon as you encounter NATs or try to access your cluster from an airport/hotel wifi Maybe you know of other mesh/p2p V…
Nebula implemented UDP(?) Relay support (1) last year although it is marked experimental.
Re: Globally Distributed Elixir over Tailscale
#9I would be very curious to see how well this actually works in different data centers. My understanding of empd is that it basically expects all nodes will be on the same switch. I don't believe there is any default handling for spikes in latency or netsplits, and epmd is known to be pretty chatty. The common wisdom seems to be to do federation of clusters rather than clustering multiple data centers together.
Doesnt seem to be a problem for all the elixir apps running as clusters on fly.io.
Re: Globally Distributed Elixir over Tailscale
#10Earlier quoted context omitted.
There’s a reimplementation of the Tailscale control plane called Headscale, if you want to self-host. I’d try to use that first, because sadly Nebula and ZeroTier don’t have a relay/TCP/HTTPS fallback option and still “have no plans of implementing one” as of Mar 2023, which leaves you out of luck as soon as you encounter NATs or try to access your cluster from an airport/hotel wifi Maybe you know of other mesh/p2p V…
Nebula implemented UDP(?) Relay support (1) last year although it is marked experimental. 1. https://github.com/slackhq/nebula/pull/678
I've been using the relay for months. It is very stable BUT, it sometimes takes minutes before peers realize they can also talk to each other on the local network.