[Openswan Users] When tunnel goes down, socket freeze

[Paul Wouters]

On Mon, 2 May 2011, Frédéric Lamorce wrote:

> In our application I set the socket as non blocking to try to prevent it,
> but it changes nothing, when this happen, the call to send() block, until
> the tunnel is re-establish.
> 
> When the tunnel is re-established, the ping comes back with some stats
> like:
> 
> 1 packets transmitted, 1 packets received, 0% packet loss
> 
> Is it a bug in the kernel? IPSec? Others?

Sounds like the -EAGAIN non-posix case when using NETKEY?
We think it is a kernel bug, but stopped trying to convince them.

Paul