Hey Guys,
We have the following problem:
Bbwin is sending traffic to 2 monitoring servers and is encrypted via stunnel. If one of the monitoring servers goes down, the number of ports being used by bbwin client increases to the point where the box is unusable - no more ports available on the windows box being monitored. If we kill the bbwin process, the problem goes away temporarily - after a short period of time, the problem returns. The problem continues even after the failed hobbit server recovers, and only clears up when we kill the bbwin process.
* Windows version Server 2003 * bbwin versions 0.11, 0.12 * stunnel version 4.25 * hobbit server version 4.2.2
Output of the stunnel log shows the following error:
2009.12.16 06:56:19 LOG7[2336:2460]: New thread created
2009.12.16 06:56:19 LOG7[2336:4016]: hobbit_ssl started
2009.12.16 06:56:19 LOG7[2336:4016]: FD 356 in non-blocking mode
2009.12.16 06:56:19 LOG7[2336:4016]: TCP_NODELAY option set on local socket
2009.12.16 06:56:19 LOG5[2336:4016]: hobbit_ssl accepted connection from 127.0.0.1:1725
2009.12.16 06:56:19 LOG7[2336:4016]: FD 396 in non-blocking mode
2009.12.16 06:56:19 LOG6[2336:4016]: connect_blocking: connecting 10.132.55.24:1982
2009.12.16 06:56:19 LOG3[2336:4016]: connect_blocking: connect 10.132.55.24:1982: No buffer space available (WSAENOBUFS) (10055)
2009.12.16 06:56:19 LOG5[2336:4016]: Connection reset: 0 bytes sent to SSL, 0 bytes sent to socket
2009.12.16 06:56:19 LOG7[2336:4016]: hobbit_ssl finished (0 left)
Can something be added to my stunnel.conf to prevent this from happening? Any help would be greatly appreciated. Thanks.
Gregory Wege wrote:
- stunnel version 4.25
That's probably not the case here, but it usually a good idea to upgrade before sending an email to ~500 subscribers of this mailing list. 8-)
2009.12.16 06:56:19 LOG5[2336:4016]: hobbit_ssl accepted connection from 127.0.0.1:1725 2009.12.16 06:56:19 LOG7[2336:4016]: FD 396 in non-blocking mode 2009.12.16 06:56:19 LOG6[2336:4016]: connect_blocking: connecting 10.132.55.24:1982 2009.12.16 06:56:19 LOG3[2336:4016]: connect_blocking: connect 10.132.55.24:1982: No buffer space available (WSAENOBUFS) (10055)
I asked Google a question about WSAENOBUFS and it replied with: http://support.microsoft.com/kb/q196271/
For some reason your server opened more connections than Windows could handle in its default configuration. Execute cmd.exe and type: netstat -bn to see the list of your connections with program names involved in creating each connection.
Mike
I upgraded to 4.29 and added TIMEOUTidle = 30 and socket = l:SO_LINGER=1:60 as I'm using the loopback address for the bbwin>stunnel connection but the problem still occurs. netstat -bn isn't displaying any processes/connections from stunnel or bbwin.
-Greg
From: [email protected] [mailto:[email protected]] On Behalf Of Michal Trojnara Sent: Friday, December 18, 2009 9:39 AM To: [email protected] Subject: Re: [stunnel-users] BBwin hanging windows
Gregory Wege wrote:
* stunnel version 4.25
That's probably not the case here, but it usually a good idea to upgrade before sending an email to ~500 subscribers of this mailing list. 8-)
2009.12.16 06:56:19 LOG5[2336:4016]: hobbit_ssl accepted connection from 127.0.0.1:1725
2009.12.16 06:56:19 LOG7[2336:4016]: FD 396 in non-blocking mode
2009.12.16 06:56:19 LOG6[2336:4016]: connect_blocking: connecting 10.132.55.24:1982
2009.12.16 06:56:19 LOG3[2336:4016]: connect_blocking: connect 10.132.55.24:1982: No buffer space available (WSAENOBUFS) (10055)
I asked Google a question about WSAENOBUFS and it replied with:
http://support.microsoft.com/kb/q196271/
For some reason your server opened more connections than Windows could handle in its default configuration. Execute cmd.exe and type:
netstat -bn
to see the list of your connections with program names involved in creating each connection.
Mike
-
Gregory Wege -
Michal Trojnara