Node.connect only works with fully qualified node name on OSX (Yosemite)

[joel-regen]

I tried connecting two nodes per the instructions in Chapter 15 of the Programming Elixir book (Dave Thomas) on a Mac running Yosemite (10.10.3) and it refused to work unless I used the fully qualified names (iex --name top@joel.local).

with short names:
TOP WINDOW
[ OTP ] $ iex --sname top@joel
Erlang/OTP 17 [erts-6.3.1] [source] [64-bit] [smp:8:8] [async-threads:10] [hipe] [kernel-poll:false] [dtrace]

Interactive Elixir (1.0.3) - press Ctrl+C to exit (type h() ENTER for help)
iex(top@joel)1> Node.connect :"bot@joel"
false

with FQ names:
[ OTP ] iex --name top@joel.local
Erlang/OTP 17 [erts-6.3.1] [source] [64-bit] [smp:8:8] [async-threads:10] [hipe] [kernel-poll:false] [dtrace]

Interactive Elixir (1.0.3) - press Ctrl+C to exit (type h() ENTER for help)
iex(top@joel.local)1> Node.connect :"bot@joel.local"
true

[josevalim]

If you use --sname, you need to either not add the "@joel" part or make sure it matches your machine name. Otherwise the runtime will be unable to lookup or find the node.

[joel-regen]

WINDOW 1:
[ OTP ] $ iex --sname top
Erlang/OTP 17 [erts-6.3.1] [source] [64-bit] [smp:8:8] [async-threads:10] [hipe] [kernel-poll:false] [dtrace]

Interactive Elixir (1.0.3) - press Ctrl+C to exit (type h() ENTER for help)
iex(top@joel)1>

WINDOW 2
iex --sname bot
Erlang/OTP 17 [erts-6.3.1] [source] [64-bit] [smp:8:8] [async-threads:10] [hipe] [kernel-poll:false] [dtrace]

Interactive Elixir (1.0.3) - press Ctrl+C to exit (type h() ENTER for help)
iex(bot@joel)1> Node.connect :"top"
false
iex(bot@joel)2> Node.connect :"top@joel"
false
iex(bot@joel)3> Node.connect :"top@joel.local"
false
iex(bot@joel)4>
11:40:39.603 [error] ** System NOT running to use fully qualified hostnames **
** Hostname joel.local is illegal **

[josevalim]

After you start both nodes, can you please run: epmd -names and let us know what it returns?

[joel-regen]

[ OTP ] $ iex --sname top
Erlang/OTP 17 [erts-6.3.1] [source] [64-bit] [smp:8:8] [async-threads:10] [hipe] [kernel-poll:false] [dtrace]

Interactive Elixir (1.0.3) - press Ctrl+C to exit (type h() ENTER for help)
iex(top@joel)1> epmd --names
** (RuntimeError) undefined function: epmd/0

iex(top@joel)1> epmd -names
** (RuntimeError) undefined function: names/0

[josevalim]

Sorry, please run it in your terminal, outside of IEx, after both nodes are started.

[joel-regen]

[ ] $ epmd -names
epmd: up and running on port 4369 with data:
name bot at port 51180
name top at port 51144

[josevalim]

That's so weird. I have no idea why it doesn't work, everything seems to be setup properly.

[joel-regen]

yeah. go figure. oh well... if you think of something keep me in the loop.

[pma]

@joel-regen I suspect /etc/hosts is missing something. You can try with @localhostwhen using -sname:

$ iex --sname top@localhost
Erlang/OTP 18 [erts-7.1] [source] [64-bit] [smp:4:4] [async-threads:10] [hipe] [kernel-poll:false]

Interactive Elixir (1.1.1) - press Ctrl+C to exit (type h() ENTER for help)
iex(top@localhost)1> 

$ iex --sname bot@localhost
Erlang/OTP 18 [erts-7.1] [source] [64-bit] [smp:4:4] [async-threads:10] [hipe] [kernel-poll:false]

Interactive Elixir (1.1.1) - press Ctrl+C to exit (type h() ENTER for help)
iex(bot@localhost)1> Node.connect :top@localhost
true
iex(bot@localhost)2>

[joel-regen]

ok. that did work:

[ OTP ] $ iex --sname top@localhost
Erlang/OTP 17 [erts-6.3.1] [source] [64-bit] [smp:8:8] [async-threads:10] [hipe] [kernel-poll:false] [dtrace]

Interactive Elixir (1.0.3) - press Ctrl+C to exit (type h() ENTER for help)
iex(top@localhost)1> Node.connect :"bot@localhost"
true

[josevalim]

@pma found the issue. If @joel does not resolve to the current machine, then it certainly wouldn't work. :)

[Gazler]

I don't use OSX, but I saw the same question on Stack Overflow so a quick search led me to http://forums.macrumors.com/threads/yosemite-10-10-hosts-file.1741422/

TL;DR - hosts file may have moved.

[brione-zz]

I was having the same problems that the OP reported and I was able to figure out a workaround based on comparing my system with one that works.

In short, I was having the exact same problem described at the top and like the OP, never got a satisfactory answer.

When I compared my system to my son's system, where --sname generated sessions work correctly, we found that on his system the command:

scutil --get HostName

Would return: HostName: not set

while on my system, it would return the default hostname of MyComputerName.local

So, as an experiment, I ran the command > scutil --set HostName _, where _ represents a space to unset the value. You'll have to become superuser for a moment for that to work.

Then I was able to start iex --sname one, and iex --sname two, and now my two nodes are able to connect and can see each other.

Previously it had only worked with the example above using one@localhost and two@localhost or by using --name with the FQN.

So, I believe that somehow if the HostName is set, as defined by scutil --get, that Elixir or Erlang is somehow getting confused. I see that this bug is closed. Should I open a new issue?

[Sotorin]

I was having exactly the same issue as the OP, I'm on Mac (El Capitan) and @brione work around worked for me as well. Nodes are connecting.