Hexchat does not auto-reconnect if a reconnect failed due to "SSL handshake timed out" #475

Open
dmlloyd opened this Issue Mar 26, 2013 · 25 comments

Comments

Projects
None yet
10 participants

dmlloyd commented Mar 26, 2013

Just as it says. I left hexchat running while I was out of town for a week, and when I came back I discovered it had disconnected, tried to reconnect, and failed. After reconnect failed due to "SSL handshake timed out", it stopped trying to auto-reconnect. Manually typing "/reconnect" did work though.

Owner

tomek commented Mar 26, 2013

What's your hexchat version & os.

Owner

Arnavion commented Mar 26, 2013

@eustachy This has been a bug since forever in all versions of Hexchat. That is, if a connection attempt fails specifically because of the SSL handshake timing out, Hexchat doesn't retry.

dmlloyd commented Mar 26, 2013

Version: 2.9.4
Compiled: Dec 7 2012
OS: Linux 3.7.6-201.fc18.x86_64
Charset: UTF-8

./configure --prefix=/home/david/local --enable-ipv6 --enable-xft --enable-openssl --enable-shm

Owner

tomek commented Mar 26, 2013

@Arnavion I really though it was fixed in git

Owner

Arnavion commented Mar 27, 2013

@eustachy Was it? I don't remember... If you're talking about #456 or #461 those are different issues.

@dmlloyd Can you try compiling from git HEAD and check if you still see it?

dmlloyd commented Mar 27, 2013

I can compile from master but I can't reproduce at will. It's just a thing that FreeNode did one time. Maybe it'd be possible to set up some kind of nc+stunnel mock server or something... not sure when I'd have time to do that though.

Owner

Arnavion commented Mar 27, 2013

@dmlloyd That's fine. I used to see it all the time back when I was on third-world internet, but now I'm not and so I haven't seen it all recently. If it happens again please let us know.

Owner

Arnavion commented Apr 2, 2013

This just happened to me when connecting failed because of proxy traversal error. It is unlikely to have been fixed for @dmlloyd 's case also.

Owner

Arnavion commented May 9, 2013

Happened to me yesterday on a build compiled from about a week ago. (Win 7 x64)

mArT-i commented Jul 5, 2013

Edit hexchat.conf and set "net_auto_reconnectonfail = 1".
That should do the trick.

On windows, there might be some troubles with the default build :
src/common/cfgfiles.c#L538

#ifndef WIN32 /* FIXME fix reconnect crashes and remove this ifdef! */
{"net_auto_reconnectonfail", P_OFFINT (hex_net_auto_reconnectonfail), TYPE_BOOL},
#endif

See #849

@mArT-i I don't have the facilities set up to recompile a fixed version for Windows, could you do that for me?

@Michaelbanks What was it previously compiled on Visual Studio? If so I have 2013 Ultimate from a BizSpark licensing if you still need it done.

Owner

Arnavion commented Jul 17, 2014

FYI: You can compile it with Express. You don't need a paid version of VS.

Also, the issue that code comment is referring is probably the same as #600

@Arnavion this is also true. I didn't know if it had been done yet or not.

@k-glendenning if you could compile a fixed version for me, that'd be great. I'm really not in the mood to set up compilers atm

Hm, apparently net_auto_reconnectonfail isn't a valid variable anyway o-o

@Michaelbanks I can, I've been dealing with hospitals over the weekend because of an assault, I'll clone the repo, and compile it.

p3lim commented Oct 27, 2014

This has not been updated since the 2.10.1 release, it is still an issue that I often encounter when using connecting over 3G or a bad connection.

I tried setting the config variable "net_auto_reconnectonfail" to 1, but it did not do anything (and it's not even listed with a default value in the files).

mArT-i commented Oct 27, 2014

As said, on windows build this option is not used to avoid some reconnect crash. Still notre fixed I guess.

Owner

Arnavion commented Oct 27, 2014

The reconnect crash mentioned in that (ancient) comment is almost certainly what we know of as #600

I'm seeing this quite often on Linux:

HexChat 2.10.2
Compiled: Nov 4 2015
OS: Linux 4.1.12-gentoo

gentoo use flags:
dbus gtk ipv6 libnotify nls plugins spell ssl -libcanberra -libproxy -ntlm -perl -plugin-checksum -plugin-doat -plugin-fishlim -plugin-sysinfo -python -theme-manager

It happens most offen with SSL connection to IRC Highway and freenode:

In network settings for both, I have selected "Connect to this network automatically", "Use SSL for all the servers on this network" and "Accept invalid SSL certificates"

irc.freenode.net/7000
irc.freenode.net/6697

irc.irchighway.net/9999

As an example, this is what I see in the network window on a failure.

 Looking up irc.irchighway.net
* Connecting to irc.irchighway.net (64.150.180.229:9999)
* * Subject: /C=US/ST=Arizona/O=IRCHighWay/OU=Hosts/CN=tastywatermelon.az.us.irchighway.net/emailAddress=ssladmin@irchighway.net
* * Issuer: /C=DE/O=IRCHighWay/OU=CA/CN=IRCHighWay CA/emailAddress=admin@irchighway.net
* * Subject: /C=US/ST=Arizona/O=IRCHighWay/OU=Hosts/CN=tastywatermelon.az.us.irchighway.net/emailAddress=ssladmin@irchighway.net
* * Issuer: /C=DE/O=IRCHighWay/OU=CA/CN=IRCHighWay CA/emailAddress=admin@irchighway.net
* Connection failed (SSL handshake timed out)

It would probably help if the timeout could be extended? I'm not sure if that's possible with SSL or not, but I'm constantly seeing that timeout error and have to manually reconnect.

on a successful connection to the same server I see this:

* Connection failed (SSL handshake timed out)
* Looking up irc.irchighway.net
* Connecting to irc.irchighway.net (173.248.152.23:9999)
* * Subject: /C=US/ST=Colorado/O=IRCHighWay/OU=Hosts/CN=friedchicken.co.us.irchighway.net/emailAddress=ssladmin@irchighway.net
* * Issuer: /C=DE/O=IRCHighWay/OU=CA/CN=IRCHighWay CA/emailAddress=admin@irchighway.net
* * Subject: /C=US/ST=Colorado/O=IRCHighWay/OU=Hosts/CN=friedchicken.co.us.irchighway.net/emailAddress=ssladmin@irchighway.net
* * Issuer: /C=DE/O=IRCHighWay/OU=CA/CN=IRCHighWay CA/emailAddress=admin@irchighway.net
* * Certification info:
*   Subject:
*     C=US
*     ST=Colorado
*     O=IRCHighWay
*     OU=Hosts
*     CN=friedchicken.co.us.irchighway.net
*     emailAddress=ssladmin@irchighway.net
*   Issuer:
*     C=DE
*     O=IRCHighWay
*     OU=CA
*     CN=IRCHighWay CA
*     emailAddress=admin@irchighway.net
*   Public key algorithm: rsaEncryption (4096 bits)
*   Sign algorithm sha1WithRSAEncryption
*   Valid since Aug  7 11:15:01 2011 GM to Aug  6 11:15:01 2014 GM
* * Cipher info:
*   Version: TLSv1/SSLv3, cipher AES256-GCM-SHA384 (256 bits)
* * Verify E: unable to verify the first certificate.? (21) -- Ignored
* Connected. Now logging in.
* Welcome to the irchighway network 
* Your host is friedchicken.co.us.irchighway.net[@0.0.0.0], running UltimateIRCd(Tsunami)-3.0(02).m5
* This server was last (re)started on Mon Jan 21 2013 at 13:28:43 MST and is located in Illinois, US
* friedchicken.co.us.irchighway.net UltimateIRCd(Tsunami)-3.0(02).m5 adhioprxDOPRSWZ acehimnopqstvAKMNORS
* SAFELIST SILENCE KNOCK FNC WATCH=128 CHANLIMIT=#&:30 MAXLIST=be:60 NICKLEN=30 TOPICLEN=307 KICKLEN=307 CHANNELLEN=32 :are available on this server
* EXCEPTS=e CHANTYPES=#& PREFIX=(aohv)!@%+ CHANMODES=be,k,l,cimnpqstAKMNORS STATUSMSG=!@%+ NETWORK=irchighway CASEMAPPING=ascii STD=i-d :are available on this server
* There are 1 users and 2805 invisible on 1 servers
.... etc etc

ChipWolf commented Dec 3, 2015

Progress folks?

I can confirm this issue. I am using HexChat 2.10.2 (64-bit) on Windows.

matoro commented Mar 3, 2016

Still waiting on something for this. Can confirm it is OS-independent as I have experienced the same symptoms across Windows 8, 10, and Arch Linux.

TingPing locked and limited conversation to collaborators Mar 3, 2016

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.