Starting an Erlang slave node in escript fails when using custom Erlang in Ubuntu 10.4

Question

I have the following escript:

#!/usr/bin/env escript
%%! -name [email protected]

main(_) ->
    NodeName = test,
    Host = '127.0.0.1',
    Args = "",

    {ok, _Node} = slave:start_link(Host, NodeName, Args),
    io:format("Node started successfully!").

When running it on Ubuntu 10.04 I get this:

$ ./start_slave
Node started successfully!
$

I want to install my own Erlang (latest version, debug compiled files for dialyzer etc) since the stock install of Erlang on Ubuntu lacks some features. I put my Erlang binaries inside ~/Applications/bin. Starting Erlang normally works, and starting slave nodes inside an Erlang shell works as well.

However, now my escript doesn't work. After about 60 seconds it returns an error:

$ ./start_slave                                   
escript: exception error: no match of right hand side value {error,timeout}

Even if I change the first line to the escript to use my erlang version, it still does not work:

#!/home/user/Applications/bin/escript

The slave node is started with a call to erlang:open_port/2 which seems to be using sh which in turn does not read my .bashrc file that sets my custom PATH environment variable. The timeout seems to occur when slave:start_link/3 waits for the slave node to respond, which it never does.

How can I roll my own installation of Erlang and start slave nodes inside escripts on Ubuntu 10.4?

Update: I've tried to add the path to my custom Erlang inside /etc/environment (where the original PATH in Ubuntu is set) but this does not change anything...

Answer 1

Is it possible that the slave node is being run with the other Erlang install? Listed under reasons for timeout error in the documentation on slave nodes I saw "the Erlang nodes have different cookies" which might, I believe, occur in that case.

If this were the case, running ps -FC erlang while it's waiting for the timeout should show you processes with different paths.