[erlang-bugs] Race condition in TLS distribution

[Fred Hebert]

On 10/19, Magnus Henoch wrote:
>I came up with the attached patch, which waits for the code server to 
>start before proceeding, and that fixes the problem for me. What do you 
>think about it?  Might there be a better way to solve this?
>

I think one of the gotchas is that the code server could be loading code 
from over the network when running on diskless node, a thing the OTP 
team supports.

In such cases, it would be impossible to use SSL as a transport 
mechanism as it would require to load the crypto stuff from over the 
network, but the crypto library would be required to establish a 
connection over the network, leaving you in an unbootable state.

I have never tried this and have no proof it's an actual problem, but 
it's a possible circular dependency between protocol and diskless nodes 
that comes up in a thought exercise.

Regards,
Fred.