[stunnel-users] Re: Connection reset by peer on same local network?

[Gabe Martin-Dempesy]

On Jan 4, 2006, at 5:00 AM, stunnel-users-request at mirt.net wrote:

> Please try the latest stunnel and OpenSSL first.
> Also send detailed logs (debug=7).

Same behavior occurs after upgrade.

on debug 7:

Server log:
2006.01.04 09:01:35 LOG7[14085:0]: Waiting 43200 second(s) for 4 file  
descriptor(s)
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 1, FD=4, (IN)->()
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 1, FD=6, (IN)->()
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 2, FD=8, (OUT)->()
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 2, FD=7, (OUT)->(ERRHUP)
2006.01.04 09:01:35 LOG7[14085:0]: Waiting 43200 second(s) for 4 file  
descriptor(s)
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 1, FD=4, (IN)->()
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 1, FD=6, (IN)->()
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 2, FD=8, (OUT)->()
2006.01.04 09:01:35 LOG7[14085:0]: CONTEXT 2, FD=7, (OUT)->(ERRHUP)
2006.01.04 09:01:35 LOG3[14085:2]: transfer() loop executes not  
transferring any data
2006.01.04 09:01:35 LOG3[14085:2]: please report the problem to  
Michal.Trojnara at mirt.net
2006.01.04 09:01:35 LOG3[14085:2]: socket open: rd=yes wr=yes, ssl  
open: rd=yes wr=yes
2006.01.04 09:01:35 LOG3[14085:2]: socket ready: rd=no wr=no, ssl  
ready: rd=yes wr=no
2006.01.04 09:01:35 LOG3[14085:2]: ssl want: rd=no wr=no
2006.01.04 09:01:35 LOG3[14085:2]: socket input buffer: 16384 byte 
(s), ssl input buffer: 16384 byte(s)
2006.01.04 09:01:35 LOG3[14085:2]: check_SSL_pending=0, ssl_closing=0
2006.01.04 09:01:35 LOG5[14085:2]: Connection reset: 1345269015 bytes  
sent to SSL, 213024 bytes sent to socket
2006.01.04 09:01:35 LOG7[14085:2]: 973 finished (0 left)
2006.01.04 09:01:35 LOG5[14085:2]: stack_info: size=65536,  
current=4620 (7%), maximum=4620 (7%)
2006.01.04 09:01:35 LOG7[14085:2]: Context 2 closed
2006.01.04 09:01:35 LOG7[14085:0]: Waiting -1 second(s) for 2 file  
descriptor(s)

Client log:
2006.01.04 09:00:39 LOG7[9605:0]: Waiting 43200 second(s) for 5 file  
descriptor(s)
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 1, FD=4, (IN)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 1, FD=6, (IN)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 1, FD=7, (IN)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 2, FD=8, (OUT)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 2, FD=9, (OUT)->(OUT)
2006.01.04 09:00:39 LOG7[9605:2]: SSL_write returned WANT_WRITE:  
retrying
2006.01.04 09:00:39 LOG3[9605:2]: SSL socket closed with 16384 byte 
(s) in buffer
2006.01.04 09:00:39 LOG5[9605:2]: Connection reset: 348124 bytes sent  
to SSL, 1345102011 bytes sent to socket
2006.01.04 09:00:39 LOG7[9605:2]: 8001 finished (0 left)
2006.01.04 09:00:39 LOG7[9605:2]: Context 2 closed
2006.01.04 09:00:39 LOG7[9605:0]: Waiting -1 second(s) for 3 file  
descriptor(s)



This part of the logs:
2006.01.04 09:00:39 LOG7[9605:0]: Waiting 43200 second(s) for 5 file  
descriptor(s)
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 1, FD=4, (IN)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 1, FD=6, (IN)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 1, FD=7, (IN)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 2, FD=8, (OUT)->()
2006.01.04 09:00:39 LOG7[9605:0]: CONTEXT 2, FD=9, (OUT)->(OUT)

loops continuously in both server and client lots the entire time,  
repeating several times each second.  One rsync session over stunnel  
generates stunnel logs over 50MB.  I have no idea if this is typical  
LOG7 behavior or not, but I'm mentioning it just in case.