[stunnel-users] stunnel 4.33 stops (crashes?) after undetermined running time

[Markus Borst]

To repeat:

- stunnel 4.33 service just crashes after some time.

- older stunnel versions do not crash

This looks like a bug in stunnel. Who do I contact because of this? How 
do I contact him or her? Is more information needed, if so, which?

Greetings
Markus Borst


-------- Original Message  --------
Subject: [stunnel-users] stunnel 4.33 stops (crashes?) after 
undetermined    running time
From: Markus Borst <M.Borst at hrz.tu-darmstadt.de>
To: stunnel-users at mirt.net
Date: 26.05.2010 15:47
> Hi,
>
> I have a strange problem with current stunnel (4.33): It just crashes 
> after some time. The time is variable, I suspect it stops after a 
> number of connections.
>
> System:
> - Windows Server 2003 SP2 (also happens on Windows Server 2008R2)
> - stunnel runs as service
> - stunnel is configured to run as ssl server for imap and pop3 with 
> our certificate (including key-chain)
>
> stunnel was running fine for years on the Windows Server 2003. After 
> upgrade to stunnel 4.33, the stunnel service process first worked 
> fine, the just stopped working after some time:
> - The windows service "stunnel" was stopped
> - the stunnel.log (standard debug level) showed not error or other 
> entry at the time of the stop.
> - The windows eventlog shows:
> ======
> Event 7034, Service Control Manager
> The stunnel service terminated unexpectedly.  It has done this 1 time(s).
> ======
> (I'm not 100% sure, whether this event log message corresponds to the 
> stunnel crash I described or to another crash.)
>
> I have downgrade to version 4.32, the stunnel service now runs 
> uninterrupted as before.
>
> Since the time of crash varies somewhat, seemingly with the current 
> load of the server, I suspect that stunnel crashes after a certain 
> amount of ssl connections, but I'm not sure. This has happened on two 
> different servers (Windows Server 2003 SP2 and Windows Server 2008 R2).
>
>
> Is this a known problem?
>
>
> I'd love to do some further tests, but since this is a production 
> system, I cannot test it in our normal environment. A program to 
> generate random ssl connections would come handy to further analyze 
> the problem.
>
> Greetings
> Markus Borst
>


-- 
TU Darmstadt
Hochschulrechenzentrum (HRZ)
Markus Borst
Adresse: 	Petersentrasse 30, 64287 Darmstadt, Germany
Tel.: 	06151/16-2056
Email: 	M.Borst at hrz.tu-darmstadt.de