Experiment to test TCP keepalive for pop3d proxies

[Wesley Craig]

On 04 Jun 2010, at 12:23, Gary Mills wrote:
> I wonder if an upgrade will solve this problem?

Perhaps.  The test accompanying that change suggests it might:

	Prevented SSL_accept() from blocking indefinitely when using TLS/SSL

Of course, without a better backtrace, we don't know if you were in  
SSL_accept().  Do you have syslog from that process, suggesting that  
it had just accepted the connection and perhaps hadn't authN'd the  
user, yet?  I'd take that as confirmation that the problem was in the  
SSL_accept() code.

:wes

ps I still think prot needs a re-write, whether the above is the  
problem you're seeing.