R Dicaire | 4 Feb 2006 16:40
Picon

openservices

Hi, I've been looking to implement openchanfix on a 2.1.7 ratbox
installation, but everything I've found points to a 2.0.8 ratbox
release. Since ratbox stable is now in 2.1.x branch, is there going to
be an update for openservices?

Thanks

--
aRDy Music and Rick Dicaire present:
http://www.ardynet.com
http://www.ardynet.com:9000/ardymusic.ogg.m3u
Lee H | 8 Feb 2006 22:48
Picon

ircd-ratbox-2.2.0rc2 'testing' released


http://www.ircd-ratbox.org/download/testing/ircd-ratbox-2.2.0rc2.tgz
Sig: http://www.ircd-ratbox.org/download/testing/ircd-ratbox-2.2.0rc2.tgz.asc
sha1sum: 62865afcfc41a8f2393d29a1f64a8460d2cba56b

This is the current version of the 'testing' tree of ircd-ratbox.

If you are upgrading from 2.1, you must read doc/whats-new-2.2.txt
If you are upgrading from 2.0, you must also read doc/whats-new-2.1.txt
If you are upgrading from 1.x, you must also read doc/whats-new-2.0.txt

-- ircd-ratbox-2.2.0rc2
- Fix compilation of contrib/
- Fix cores with unknown servers
- TESTMASK has been extended to take a gecos field, and match against CIDR
  ip masks.
- New operspy capable MASKTRACE command, works like testmask except actually
  lists those clients that match.
- Fix listener name for wildcard binds.
- Update some help files
- Fix a core in challenge.
- Rewrite the logic for match_esc() used by xlines, so that ' <at> ' doesnt match
  a literal ' <at> ' etc.
- Properly exit when we receive SIGTERM.
- Add some more log entries for when we drop new server connections.

$Id: RELNOTES 21860 2006-02-08 21:02:41Z leeh $

--

-- 
-                 Lee H // anfl
(Continue reading)

Lee H | 8 Feb 2006 22:50
Picon

Re: openservices

On Sat, Feb 04, 2006 at 10:40:34AM -0500, R Dicaire wrote:
> Hi, I've been looking to implement openchanfix on a 2.1.7 ratbox
> installation, but everything I've found points to a 2.0.8 ratbox
> release. Since ratbox stable is now in 2.1.x branch, is there going to
> be an update for openservices?

Its doubtful -- it was recommended to them not to bother as a stable release
of 2.2.0 isnt that far away.

-- 
-                 Lee H // anfl
-        I code, therefore I break things.
On Sat, Feb 04, 2006 at 10:40:34AM -0500, R Dicaire wrote:
> Hi, I've been looking to implement openchanfix on a 2.1.7 ratbox
> installation, but everything I've found points to a 2.0.8 ratbox
> release. Since ratbox stable is now in 2.1.x branch, is there going to
> be an update for openservices?

Its doubtful -- it was recommended to them not to bother as a stable release
of 2.2.0 isnt that far away.

--

-- 
-                 Lee H // anfl
-        I code, therefore I break things.
Richard Brooklyn | 15 Feb 2006 14:27

Not automatically reconnecting to hub

Hi,

This has been a issue with my server for some time now. When it splits
from our hub due to networking issues, it fails to attempt to reconnect.

I'm currently running 2.1.7. I have 5 servers set up in 'connect'
blocks. Only one of them has the autoconn flag, as it's our main hub,
and the rest are back-up re-routes should the hub go down.

There are no configuration errors showing when I issue a rehash. And
a /stats c command shows that the autoconn has been seen (some details
altered):

--- C * <at> 127.0.0.1 TZ irc.hidden 7000 server
--- C * <at> 127.0.0.1 TZ irc.hidden 9000 server
--- C * <at> 127.0.0.1 TZ irc.hidden 9000 server
--- C * <at> 127.0.0.1 TZ irc.hidden 5500 server
--- C * <at> 127.0.0.1 ATZ hub.hidden 9000 server
--- c :End of /STATS report

So, I'm not sure as to why the server does not automatically connect. My
server was doing this just fine on 2.0, but since I migrated to 2.1,
this has not been working. This would suggest a configuration issue. The
connect block for the hub is as follows (details changed again):

connect "hub.hidden" {
        host = "xxx.xxx.xxx.xxx";
        accept_password = "mypassword";
        send_password = "isnotsecure";
        port = 9000;
(Continue reading)

Lee H | 15 Feb 2006 17:04
Picon

Re: Not automatically reconnecting to hub

On Wed, Feb 15, 2006 at 01:27:12PM +0000, Richard Brooklyn wrote:
> This has been a issue with my server for some time now. When it splits
> from our hub due to networking issues, it fails to attempt to reconnect.
> 
> I'm currently running 2.1.7. I have 5 servers set up in 'connect'
> blocks. Only one of them has the autoconn flag, as it's our main hub,
> and the rest are back-up re-routes should the hub go down.
> 
> --- C * <at> 127.0.0.1 TZ irc.hidden 7000 server
> --- C * <at> 127.0.0.1 TZ irc.hidden 9000 server
> --- C * <at> 127.0.0.1 TZ irc.hidden 9000 server
> --- C * <at> 127.0.0.1 TZ irc.hidden 5500 server
> --- C * <at> 127.0.0.1 ATZ hub.hidden 9000 server
> --- c :End of /STATS report

You should look at what the max links (or whatever its called) is in your
server class.  Whatever that number is set to, if that number of servers are
already linked to your ircd, it will not attempt autoconnects.

You probably just need to seperate hubs into a seperate class.

--

-- 
-                 Lee H // anfl
-        I code, therefore I break things.
On Wed, Feb 15, 2006 at 01:27:12PM +0000, Richard Brooklyn wrote:
> This has been a issue with my server for some time now. When it splits
> from our hub due to networking issues, it fails to attempt to reconnect.
> 
(Continue reading)

Richard Brooklyn | 15 Feb 2006 17:44

Re: Not automatically reconnecting to hub

On Wed, 2006-02-15 at 16:04 +0000, Lee H wrote:
> You should look at what the max links (or whatever its called) is in your
> server class.  Whatever that number is set to, if that number of servers are
> already linked to your ircd, it will not attempt autoconnects.
> 
> You probably just need to seperate hubs into a seperate class.

This did indeed fix things. I created a new class just for hubs, as that
testlink server was already linked in as a 'server' class, and my max
connections was only one, this filled the class and stopped the server
re-joining the hub in case of a split.

With the hubs class created, and the max connections being just 1
(that's all I need for proper hub connections), the server is now
behaving.

Thanks for the help.

Dustin Marquess | 20 Feb 2006 07:54

devpoll in 2.2.0rc2

Just a quick note... It looks like the devpoll.c module in 2.2.0rc2 was 
never updated to handle the new dynamic maxclients system.  It is still 
using:

static short fdmask[POLL_LENGTH];

-Dustin
Aaron Sethman | 21 Feb 2006 19:11

Re: devpoll in 2.2.0rc2

Committed a fix for this.  I don't have access to a Solaris box to test it 
at the moment, so it at least *looks* correct now.

-Aaron

On Mon, 20 Feb 2006, Dustin Marquess wrote:

> Just a quick note... It looks like the devpoll.c module in 2.2.0rc2 was never 
> updated to handle the new dynamic maxclients system.  It is still using:
>
> static short fdmask[POLL_LENGTH];
>
> -Dustin
> _______________________________________________
> ircd-ratbox mailing list
> ircd-ratbox <at> lists.ratbox.org
> http://lists.ratbox.org/cgi-bin/mailman/listinfo/ircd-ratbox
>
Dustin Marquess | 21 Feb 2006 19:57

Re: devpoll in 2.2.0rc2

Okay, that was the push I needed.  I tried fixing it myself, and I ended 
up with an ircd with broken connectivity :).

This should fix the rest of it.  Tested on Solaris & Tru64...

-Dustin

--- devpoll.c   2006-02-21 12:14:46.600354763 -0600
+++ /home/jailbird/devpoll.c    2006-02-21 12:54:24.797807221 -0600
 <at>  <at>  -26,10 +26,11  <at>  <at> 
   */

  #include "config.h"

  #include "stdinc.h"
+#include <poll.h>
  #include <sys/devpoll.h>

  #include "config.h"
  #include "commio.h"
  #include "class.h"
 <at>  <at>  -52,11 +53,11  <at>  <at> 
  #include "memory.h"

  static void devpoll_update_events(int, short, PF *);
  static int dpfd;
  static short *fdmask;
-static pollfd *npollfds;
+static struct pollfd *npollfds;
  static void devpoll_update_events(int, short, PF *);
(Continue reading)

Aaron Sethman | 21 Feb 2006 20:13

Re: devpoll in 2.2.0rc2


On Tue, 21 Feb 2006, Dustin Marquess wrote:

> Okay, that was the push I needed.  I tried fixing it myself, and I ended up 
> with an ircd with broken connectivity :).
>
> This should fix the rest of it.  Tested on Solaris & Tru64...

Committed.

-Aaron

Gmane