- stunnel-users - stunnel.org

Keep initial session-ID when stunnel-client reconnect to a stunnel-server
by Martin Got 01 May '19

01 May '19

Trying to establish link using pppd connection overlapped into stunnel 5.44. Server runs on OpenBSD 6.4, client uses the same stunnel and OS versions. From the client's side I have to reconnect to stunnel-server in case of Internet link is dropped. And expecting that pppd session is kept by stunnel over reconnections possible. But stunnel-server runs another pppd instance when stunnel-client reconnects. After reconnection in system processes I see two pppd processes (as shows below), and pppd link has no pings after it. # ps -aux | grep pppd user 45359 0.0 0.0 440 1672 ?? I Tue11PM 0:00.01 lock passive 10.0.1.1:10.0.1.2 local debug noauth (pppd) user 1156 0.0 0.0 321 1210 ?? I Tue12PM 0:00.01 lock passive 10.0.1.1:10.0.1.2 local debug noauth (pppd) 1. The question is how to keep initial session established by stunnel-client and do not run second pppd process by stunnel-server? 2. May I keep session by cache-ID on the client's side only or do I need to do it on both client and server? Can it help not to run another pppd process? Both server config and client config parts are below: # cat /etc/stunnel/stunnel-server.conf ... [ppp] ;SERVER accept 1.2.3.4:443 exec = /usr/sbin/pppd execargs = lock passive 10.0.1.1:10.0.1.2 local debug noauth pty = yes CAfile = ca.crt cert = server.crt key = server.key verifyChain = yes TIMEOUTclose = 0 ... # cat /etc/stunnel/stunnel-client.conf ... [ppp] ;CLIENT client = yes retry = yes connect = 1.2.3.4:443 exec = /usr/sbin/pppd execargs = persist lock 10.0.1.2:10.0.1.1 local debug noauth name ppp-client pty = yes CAfile = ca.crt cert = client.crt key = client.key verifyChain = yes checkHost = hostname ;checkIP = 1.2.3.4 ... Thank you for answer in advance.

1 0

Re: [stunnel-users] illegal instruction crash 5.50 Debian/Stretch
by Michal Trojnara 29 Apr '19

29 Apr '19

On 24.04.2019 16:06, Florian Lohoff wrote: > with the stunnel4 in Debian/Stretch i am experiencing pretty regular > crashes with illegal instruction. Version is 5.50. > In Bug https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=880659 > you can see the drill down to a Lock beeing unlocked although > not locked. When i look at the code its some usage of openssl > primitives. (Unlock happens in openssl code) > I had a quick look at the changelog and this: > "Service threads are terminated before OpenSSL cleanup to prevent occasional stunnel crashes at shutdown." > sounds a bit like the issue i am experiencing. > Can you confirm this? Yes, I do confirm this is the issue described in the Debian bug #880659. It wasn't a simple coding error, but rather a design flaw introduced in version 3.0 that took 20 years to identify and fix. Before version 5.18, released almost 4 years ago, the shutdown race condition mostly manifested itself while terminating stunnel under extremely heavy load, such as stress testing. The probability has increased with the introduction of periodic DH parameter regeneration in version 5.18. I didn't test it, but I suspect some changes in the OpenSSL internals may have also cause the crashes to only surface fairly recently, and not 20 years ago. The fix itself involves a major redesign of the stunnel threading subsystem, so I made a beta version and I'd appreciate your help with testing it: https://www.stunnel.org/downloads/beta/stunnel-5.54b2.tar.gz Best regards, Mike

2 1

Either "verifyChain" or "verifyPeer" has to be enabled
by kovacs janos 23 Apr '19

23 Apr '19

sorry to bother, again this is probably something simple, but i couldn't find anything about it. i'm trying to make stunnel NOT verify the certificate of the address it connects to. specifically because the certificate date will give an error, and i know its safe to connect there. whenever i put 'verify = 0' or 'verifyChain = no' or 'verifyPeer = no' in the configuration, i get an error saying 'Either "verifyChain" or "verifyPeer" has to be enabled', and so the configuration can't be applied. and i found nothing about this anywhere. or this could only be set on server Services? and there is no way to not verify if 'client = yes'?

1 0

Stunnel 5.44 client syntax to accept connections on pty?
by Martin Got 18 Apr '19

18 Apr '19

How to use 'stunnel' in client mode to accept connections on pty from local programs like 'ppp' instead of 'accept=IP:PORT'? 'pty = yes' seems work in server mode only, so when I removed 'accept = 127.0.0.1:1723' from 'stunnel' client config I'm getting the error: [!] Service [ppp]: Each service must define two endpoints on stunnel-5.44 I need to call 'stunnel' client from 'pppd' and communicate with 'stunnel' using pty if it was possible by using a command: pppd /dev/ptyp3 10.0.1.2: local debug noauth passive noccp novj novjccomp nopcomp noaccomp name ppp-client connect 'stunnel /etc/stunnel/stunnel-client.conf' Both 'pppd' and 'stunnel' started by executing the command above, but 'pppd' can't communicate to 'stunnel' server using pty. Please advice. [ppp] client = yes accept = [127.0.0.1:1723 ](http://127.0.0.1:1723) ^^^^^^^^^^^^^^^^^^^^^^^ connect = STUNNEL-SERVER-IP:723 CAfile = /etc/stunnel/ca.crt cert = /etc/stunnel/client.crt key = /etc/stunnel/client.key verifyChain = yes checkHost = hostna.me ;checkIP = 1.2.3.4

2 1

Sending original IP (proxy) for smtp
by James Brown 16 Apr '19

16 Apr '19

I’d like stunnel to pass the incoming IP along, not stunnel's IP. My traffic flow is: Stunnel (accept on port 465) -> ASSP (an smtp proxy) -> Postfix (accept on port 10026) I tried: [ssmtp] client = no accept = 465 connect = 25 protocol = proxy But ASSP had: Apr-15-19 18:50:22 [Worker_1] Connected: session:7FE104AA1870 127.0.0.1:58954 > 127.0.0.1:25 > 127.0.0.1:10026 Apr-15-19 18:50:23 [Worker_1] 127.0.0.1 warning: got reply '502 5.5.2 Error: command not recognized' from 127.0.0.1 Apr-15-19 18:50:23 id-18223-03317 [Worker_1] 127.0.0.1 info: sending EHLO instead of HELO to 127.0.0.1 Apr-15-19 18:50:23 id-18223-03317 [Worker_1] 127.0.0.1 disconnected: session:7FE104AA1870 127.0.0.1 - processing time 1 seconds And Postfix had: 2019-04-15 18:50:22.995042+1000 localhost smtpd[33360]: connect from localhost[127.0.0.1] 2019-04-15 18:50:23.108756+1000 localhost smtpd[33360]: improper command pipelining after EHLO from localhost[127.0.0.1]: RSET\r\n 2019-04-15 18:50:23.273664+1000 localhost smtpd[33360]: disconnect from localhost[127.0.0.1] ehlo=2 rset=1 quit=1 unknown=0/1 commands=4/5 So I take it protocol=proxy is not the way to do it. Sounds like this only works with haproxy which I do not have installed. Is there a way to do this? The problem I am trying to solve is ASSP has lines like: warning: SMTP authentication failed on 127.0.0.1 - obviously I don’t want fail2ban to ban 127.0.0.1. Stunnel log has: 2019.04.15 18:52:26 LOG5[23]: Service [ssmtp] accepted connection from ::ffff:185.222.209.66:53642 2019.04.15 18:52:27 LOG3[23]: s_connect: connect ::1:25: Connection refused (61) Any suggestions? Thanks, James.

2 3

Support for TLS 1.3 Early Data?
by hshh 13 Apr '19

13 Apr '19

Is it stunnel 5.54 support for TLS 1.3 Early Data? Stunnel 5.54, TLSv1.3, Cipher is TLS_AES_256_GCM_SHA384. I was tested by openssl s_client, return "Early data was not sent".

1 0

Possible memory leak at asn1_lib.c:412: 20417 allocations
by Marc Smith 12 Apr '19

12 Apr '19

Hi, I'm using stunnel version 5.44 with OpenSSL version 1.0.2r on Linux. I noticed if stunnel has been running for a while, we start observing these messages in the logs: Apr 12 19:12:11 localhost stunnel: LOG4[971]: Possible memory leak at asn1_lib.c:412: 20417 allocations Apr 12 19:12:11 localhost stunnel: LOG4[971]: Possible memory leak at tasn_new.c:180: 20416 allocations And indeed, the process is consuming memory at a steady rate, and continues to grow until stopped. I updated to the latest stunnel version (5.53) and I no longer see these "possible memory leak" messages in the logs, yet the process still appears to consume memory as before (until it gets very large and things stop working). Is this actually a bug in the OpenSSL library? I'm using the latest in that release series (1.0.2), but perhaps this is a "known" leak and resolved in OpenSSL 1.1.1? Any help would be greatly appreciated. Thanks, Marc

1 0

OCSP validation via AIA responders through a proxy
by Olops, Daniel 12 Apr '19

12 Apr '19

Hello, stunnel v5.53 ignores proxy settings when performing OCSP validation via AIA responders. An open issue for OpenSSL at their GitHub suggests that this is a feature request for them (https://github.com/openssl/openssl/issues/6965) thus not being a stunnel limitation itself, and I've asked for clarification in the openssl-users mainling list (https://mta.openssl.org/pipermail/openssl-users/2019-April/010285.html) In either case, I want do double-check it with you guys: is this really an OpenSSL limitation, or is this something that can be addressed in stunnel code? Regards, Daniel O. Amazon Web Services EMEA SARL, 38 avenue John F. Kennedy, L-1855 Luxembourg, R.C.S. Luxembourg B186284 Amazon Web Services EMEA SARL, Irish Branch, One Burlington Plaza, Burlington Road, Dublin 4, Ireland, branch registration number 908705

1 0

stunnel 5.44 + pppd client side problem
by Martin Got 12 Apr '19

12 Apr '19

Trying to set up pppd link with stunnel wrapped between two OpenBSD 6.4amd64 machines. I use this reference article as an idea: http://bremford.org/tips/QuickStunnelVPN.html While connecting from client's side by command: /usr/sbin/pppd ptypA 10.0.1.2: local debug noauth passive noccp novj novjccomp nopcomp noaccomp name ppp-clnt connect 'stunnel /etc/stunnel/stunnel-client.conf stunnel-client starts, pppd starts on client's end according to stunnel-clnt.log, but has LCP timeouts: # tail stunnel-clnt.log stunnel: LOG5[ui]: Configuration successful pppd[5421]: Connect: ppp2 <--> /dev/ptypA pppd[5421]: LCP: timeout sending Config-Requests pppd[5421]: Connection terminated. pppd[5421]: Connect script failed It seems no pppd pty client connection to stunnel-local nor remote stunnel-server afterwards. But when I tried to connect to stunnel-client port 1723 using telnet: telnet localhost 1723 I received pppd advertisements from remote stunnel-server. It seems exec = /usr/sbin/pppd on stunnel-server is running when client's stunnel-client connection appeared. Can it be a problem with pppd and stunnel-client using pty? Please advice. # cat /etc/stunnel/stunnel-server.conf ;chroot = /var/stunnel # chroot is disabled for testing ;setuid = _stunnel # stunnel started by root for testing currently ;setgid = _stunnel ; PID file is created inside the chroot jail (if enabled) ;pid = /stunnel.pid foreground = yes debug = 7 ;output = log/stunnel.log # disabled sslVersion = TLSv1.2 socket = l:TCP_NODELAY=1 socket = r:TCP_NODELAY=1 ; Enable support for the insecure SSLv3 protocol ;options = NO_SSLv3 options = NO_TLSv1 options = NO_TLSv1.1 ; Fix for Eudora "error reading network" can be useful for changing packet length options = DONT_INSERT_EMPTY_FRAGMENTS ; These options provide additional security at some performance degradation ;options = SINGLE_ECDH_USE ;options = SINGLE_DH_USE ; *** TLS server mode services [ppp] accept = 723 exec = /usr/sbin/pppd execargs = 10.0.1.1: local debug noauth pty = yes CAfile = /etc/stunnel/ca.crt cert = /etc/stunnel/srv.crt key = /etc/stunnel/private/srv.key verifyChain = yes TIMEOUTclose = 45 [default] ; HTTP connections ;ciphers = ALL ;options = CIPHER_SERVER_PREFERENCE accept = 1111 connect = 127.0.0.1:80 CAfile = /etc/stunnel/ca.crt cert = /etc/stunnel/srv.crt key = /etc/stunnel/private/srv.key verifyChain = yes TIMEOUTclose = 0 [ntp] connect = 127.0.0.1:123 sni = default:ntp CAfile = /etc/stunnel/ca.crt cert = /etc/stunnel/srv.crt key = /etc/stunnel/private/srv.key verifyChain = yes TIMEOUTclose = 0 -------------------- # cat /etc/stunnel/stunnel-client.conf chroot = /var/stunnel setuid = _stunnel setgid = _stunnel pid = /stunnel-clnt.pid foreground = yes debug = 7 ;output = log/stunnel-clnt.log sslVersion = TLSv1.2 socket = l:TCP_NODELAY=1 socket = r:TCP_NODELAY=1 ; Enable support for the insecure SSLv3 protocol ;options = NO_SSLv3 options = NO_TLSv1 options = NO_TLSv1.1 ; Fix for Eudora "error reading network" can be useful for changing packet length options = DONT_INSERT_EMPTY_FRAGMENTS ; These options provide additional security at some performance degradation ;options = SINGLE_ECDH_USE ;options = SINGLE_DH_USE [ppp] client = yes accept = 127.0.0.1:1723 # 'accept' is absent in client's configuration http://bremford.org/tips/QuickStunnelVPN.htmlbut stunnel reports: [!] Service [ppp]: Each service must define two endpoints on stunnel-5.44 connect = STUNNEL-SERVER-IP:723 CAfile = /etc/stunnel/ca.crt cert = /etc/stunnel/client.crt key = /etc/stunnel/client.key verifyChain = yes checkHost = hostna.me ;checkIP = 1.2.3.4 --------------------

2 1

Stunnel 5.51 Pop3 Problem?
by Colin 10 Apr '19

10 Apr '19

Hello, I'm new to this list and looking for advice/help with release 5.51 for Windows. I have three accounts, Yahoo, Aol and Gmx and I use Foxmail 5 with Stunnel. I use pop3 with all accounts and sometimes when retrieving mail I may decide to 'Not Fetch' any mail on the server that looks 'iffy' untill I've looked at it via the mail providers website. Now I've just upgraded to version 5.51 (over 5.50) and when checking any of the said accounts I can't see any mail that hasn't been fetched (the server looks empty.) But if I login to the mail website the unfetched mail is there to view for all accounts. I reverted to version 5.50 and now I can see the unfetched mail for all accounts in my 3rd party software. Therefore my question is: Is there a problem (bug??) with 5.51 which stops me from seeing unfetched pop3 mail or do I have to add something to the config file to make it work properly? Thanks...Colin ============ The problem of mail not being seen on the server seems fixed in 5.52 but now another glitch has arisen. When trying to download mail from a server the programme keeps freezing at a certain point. Let's say if I have ten mails to download the programme will freeze at maybe the third one and won't download the rest. I can try it several times and everything freezes at the same point. Again I've gone back to 5.50 and all works perfectly. Any ideas on how to fix this or shall I stay with 5.50?

2 1