UnrealIRCd Bug Tracker - Issues (acknowledged features)

0006422: query security-groups

Fri, 31 Oct 2025 13:35:51 +0100

I would like to be able to query
* which security-groups there are
* members of a given security-group.

The former could possibly best be done with /STATS?
The latter would possibly best be done with a /WHO flag?

0005304: Allow different options/features for different IP addresses or ports of the same ircd

Sun, 14 Sep 2025 18:03:27 +0200

Currently we have to run multiple instances of the ircd to achieve that.
One possible solution would be to allow "set" options in "class" blocks, and allow selecting ports in "allow" blocks.
Example (i have added "no" prefix to disable the following option in case it's enabled in the global set block):

class normal {
pingfreq 100;
maxclients 1000;
sendq 100k;
recvq 4000;
set {
prefix-quit "Quit";
options {
no websocket;
identd-check;
show-connect-info;
};
};

class websocket {
pingfreq 100;
maxclients 1000;
sendq 100k;
recvq 4000;
set {
prefix-quit "Web client exited";
options {
websocket;
no identd-check;
no show-connect-info;
};
};

allow {
ip *@*;
// all ports except those "caught" by allow blocks below
class normal;
maxperip 10;
};

allow {
ip *@*;
port 8080;
class websocket;
maxperip 2;
};

listen {
ip *;
port 6667-6669;
};

listen {
ip *;
port 8080;
options { ssl; };
};

0006567: server_ban_exception.get return strange response!

Mon, 25 Aug 2025 15:57:37 +0200

Hi,
Case 1:
When trying getting this server_ban_exception (*@::1) i have this error.
Error: For technical reasons you cannot start the host with a ':', sorry

Case 2:
There is also something strange. When i try to get information about the name (*) i have also this error
"Error: Nickname not found"

Thanks.

0006499: Some spamfilter options are currently config-file-only and not in SPAMFILTER command

Tue, 18 Feb 2025 19:03:34 +0100

The following spamfilter features are currently only available in the spamfilter { } block but not in the SPAMFILTER command:
* spamfilter::rule (on its own or as a filter)
* spamfilter::input-conversion
* combining multiple spamfilter::action's
* spamfilter::action: set, report, stop. With the first two also having the thing that they accept parameters (mandatory for set, optional for report)

Obviously some day we would want this to be available in the SPAMFILTER command as well. And in RPC.

0005306: Support PROXY protocol used by TCP load balancers

Tue, 21 Jan 2025 11:28:05 +0100

Many TCP load balancers support the PROXY protocol to pass along client IP information at the start of a connection.
UnrealIRCD should support for trusting the PROXY data and using the provided client IP for connection throttling and ident.

Docs can be found here: https://docs.nginx.com/nginx/admin-guide/load-balancer/using-proxy-protocol/

0005290: U5: extban to prevent channel history

Sun, 15 Sep 2024 19:10:57 +0200

'i' suggested an extban to ban people from seeing channel history.
Similarly, if combined with +e, you could create a whitelist for channel history.
In all cases, +H will need to be set.

0006281: Allow commas in "port" or multiplie "port" lines in listen{} blocks

Thu, 06 Jun 2024 17:46:43 +0200

I think the "port" option for listen{} blocks should allow commas or multiple "port" lines

Rather than having:
listen {
ip *;
port 6660-6669;
options { clientsonly; }
}
listen {
ip *;
port 9000;
options { clientsonly; }
}

we should be able to do:
listen {
ip *;
port 6660-6669, 7000;
options { clientsonly; }
}

0006285: More ways to control reputation

Sun, 17 Sep 2023 12:00:03 +0200

I noticed #define REPUTATION_SCORE_CAP 10000 is the hardcoded cap for reputation. I think this should be a configurable feature as it is not hard to idle on a network long enough to get 10000 reputation, even with clones. Would also allow for more customization on making a security group for "long time chatters".

Another idea is the ability to reduce a clients reputation over time if they have not sent any data to the server in x amount of time. After that x amount of time, the reputation can go down by x amount every z(days,hours,etc).

0005775: ./Config will copy modules from previous install even if a newer one exists with the same name

Sat, 09 Sep 2023 09:52:38 +0200

As the title implies, while it is quite useful that ./Config can import settings/modules from a previous install, it does not check for existing modules, and will overwrite even a newer version of the same module without warning or confirmation.

0004055: [feature] Move shun permissions to a shun block or set::shun

Sun, 23 Jul 2023 21:30:22 +0200

This will allow admins to define permissions for users that are shunned to match their network policies.

Example:
shun {
options {
join "no";
join-help-chan "yes";
part "yes";
msg-opers "yes";
msg-users "no";
msg-chans "no";
keep-chanops "no";
};

Shun should always allow ADMIN, QUIT and PONG. For backward compat, all options as well as the block it's self should be optional. Values not included should be an implied "no".

0006311: More max length customizations.

Sun, 23 Jul 2023 18:55:24 +0200

You already allow some maximum lengths to be set in the config, however are missing the following,

Maximum Channel Name Length is hard-coded to 32 characters, Insp this is customizable and defaults to 64.
Maximum Modes per line is hard-coded to 12, Insp this is customizable and defaults to 20.
Maximum Real Name/Gecos length is hard-coded to 50, Insp this is customizable and defaults to 128.

Moot point ones, but will be listed for completeness:

Maximum hostname/ident is hard-coded to 10, Insp this is customizable. Moot point as mIRC limits the ident to 10 chars anyway.
Maximum nick is hard-coded to 30 characters, Insp you can go higher.

0006310: Make /WATCH and /MONITOR entry list sizes customizable.

Sun, 23 Jul 2023 18:48:47 +0200

Both /WATCH and /MONITOR entry lists are hard-coded in UnrealIRCd to 128. While this may seem relatively large, some people may want to lower it to save system resources, or make it higher. InspIRCd allows entry size customization for both. I don't feel things like this should be hard-coded into the IRCd. We want to discourage people from modifying source code.

0006270: SPAMFILTER/Action : viruschan has stranges behaviors and we're unable to free users trapped

Tue, 16 May 2023 05:14:32 +0200

I have tested a rule with spamfilter with action viruschan. When some link is sent, the user is trapped and correctly isolated in the viruschan.

USER SIDE :

At the same time, user can see these returns (depends of what his client does when it opens a new chan) :
17:01:42 i : MODE: Unknown command
17:01:44 i : WHO: Unknown command

And since he's trapped, all what he wants to do returns an error "Unknown command" :
17:02:03 i : PING: Unknown command
17:05:14 i: Notice: Unknown command
...
And almost every 20secondes, the user has a return "unknown command" :
(depends of his client and his raw's treatment I guess)
- On kiwiirc we just see "Unknown command"
- On weechat, there's no return (or not shown) but after 5mn the lag is enough to be disconnected by client :
17:07:03 i : irc : le lag est élevé, déconnexion du serveur
17:07:03 i : irc : déconnecté du serveur
- On Kvirc we see "NOTICE : unknown command" (user didn't sent any notice) :
[17:18:41] [erreur] NOTICE Commande invalide
[17:19:01] [erreur] NOTICE Commande invalide
[17:19:21] [erreur] NOTICE Commande invalide
...

SERVER/OPER SIDE :

The other side, the user seems not to be shuned or tempshuned :
- No shuned info in his whois
- Stats s is empty
- /shun or /tempshun -nick return that the user is not shuned/tempshuned

And there is no way to free the user :
- shun/tempshun -nick returns is not shuned
- Kick or sapart him changes nothing, user is just stuck in the void no way to rejoin viruschan neither

Only solution is to deco/reco.

Reproductible every time

Thanks for reading :)

0006084: Please allow '/AUTH'ing after connection handshake

Wed, 06 Apr 2022 09:39:13 +0200

Hi there!

Request:
Please could you add the ability for a user to be able to '/AUTH' after they have connected, instead of it being restricted to the authprompt module (before connect).

This is for completeness, as an emulation of SASL is still a SASL in this respect, and clients should still be able to take full use of the SASL mechanism as a form of identification using including this compatibility layer, as it is a very useful command.

Thanks for considering! Have a great week

0006017: Add support for the netsplit and netjoin batches

Sat, 04 Dec 2021 08:39:02 +0100

Hi,

IRCv3 defined two batch types, to group QUIT and JOIN messages in case of netsplits: https://ircv3.net/specs/batches/netsplit

It would be nice if Unreal could implement them, so clients can easily fold long groups of QUIT/JOIN messages instead of displaying them all.

Thanks!

0005920: Include git commit in VERSION output.

Sun, 27 Jun 2021 17:59:45 +0200

When compiling from git the current output from connecting and using /version has "UnrealIRCd-5.2.1-git".

This is not very descriptive since I do not know which git commit the server is running on.

This can be an issue when you are trying to debug something on network (like pissnet) with 100+ servers each run by different people with some of them using tarball releases and some using whatever existed in git head/master at the time they compile.

Suggestion:

UnrealIRCd-5.2.1-git-XX-YYYYYYY

"XX", number of git commits since the creation of 5.2.1
"YYYYYYY", the first seven digits of the git commit.

0005531: Option to deny/allow channels on the fly

Fri, 27 Nov 2020 17:24:37 +0100

This all started because of this:
<@Jobe> ammusingly enough in ircu you can gline channel names

Which gave me the following idea:
Since we have "deny channel" block would be nice to have some kind of option to do the same in the spamfilter.

My thoughts were to have something like this:
/SPMAFILTER add -simple C redirect #helpop The_official_help_channel_is_#help #help

In the above command, we have a new type: C for channel name, a redirect option and the target channel.
Other actions such as block, tempshun should work as they work until now.

I also thought in having a specific command like:
/DENYCHAN #channel #redirect_channel/- reason

On this command, the second parameter is the #channel to be redirect to or none if "-" is given.

I assume that we would need the counterpart somehow (the allow channel option) to be implemented also.
My idea for spamfilter is:
/SPAMFILTER add -simple C allow - Official_help_channel #help

As for a separated command:
/ALLOWCHAN #channel - reason
(I've kept the "-" since we would be using this syntax in the denychan command)

I'm just requesting this, as this would be a handy way to allow admins to allow/deny channels on the fly without the need to edit config files and therefore speed up the way admins would be able to deny a channel reported as a warez channel or something like that.

Cheers

0005789: Toggle *-with-override ircop privs with dedicated usermode

Fri, 27 Nov 2020 17:17:36 +0100

Summary gives the rough idea, usage of *-with-override privs (from hereon refered to as "actual override") should be "toggled" via a dedicated user mode, for example +O (implies oper-only to a degree, and it not in use). The reason why this might be beneficial is that opers with "actual override" can accidentally trigger these with e.g. flood protection scripts, or use them deliberately and blame a script on it afterwards. Setting a usermode like +O should be a conscious effort and thus the abovementioned situations could be avoided better.
Further enhancement would be if the umode sends an SNOTICE (perhaps to EYES snomask +e?) when set and unset (perhaps a config option whether it sends the SNOTICES?).
Also it should be (config-sided) timed how long it is active and then auto-unset (e.g. Charybdis uses 30 minutes for this umode, which is +p on their end, but in use on Unreal).
One difficulty may be limiting this usermode to just the "actual override" opers, although this could be done via a new privilege that would be given by default to all *-with-override opertypes (and none that do not have that in their name).

0005733: disallow character 0x7F in channel names

Mon, 24 Aug 2020 17:55:22 +0200

It is possible to create a channel with 0x7F character in its name (ASCII "DEL"). This character does not show up in irssi (I haven't tried other clients) and it's misleading. Despite RFC 1459 and RFC 2812 implicitly allowing this character, I believe it shouldn't be allowed.

0005663: antimixedutf8: detect more transitions

Sun, 03 May 2020 15:09:09 +0200

right now it only detects cyrillic <-> non-cyrillic

0005548: Logging format of user@host and in oper notices

Thu, 06 Feb 2020 17:19:38 +0100

At various places we show a nickname with user@host in a notice or logging. Eg:
sendto_snomask(SNO_FNICKCHANGE, "*** %s (%s@%s) has changed their nickname to %s"
client->name, client->user->username, client->user->realhost, nick);
It would be better to replace it with a function that returns %s (%s@%s), like:
sendto_snomask(SNO_FNICKCHANGE, "*** %s has changed their nickname to %s"
something(user), nick);

And then it will be configurable if the format should be "nick (user@host)", "nick (user@ip)" or "nick (user@host) [ip]" or something else. etc. <- no these are not final thoughts, but just to get the idea
eg via set::log-whatever-some-nice-name [userhost|userip|somethingelse]; <- again, not the final name :D

This so we can have a uniform way to communicate user information in logs and oper notices.
The function is not meant to be used for sending notices (or other information) to non-ircops.

0005441: Current ping timeout configuration unreasonable for radio/satellite communicatons

Wed, 11 Dec 2019 16:11:08 +0100

I have been a fan of UnrealIRCd for well over a decade now and in that time I have also worked with various network typologies employing mobile communications, everything from household smartphones to VSAT units, both in the public and private sectors. While some of the environments I have worked in have had the budgets to employ specialized network equipment to ensure high ping-through rates even with higher bit-error rates, other environments I have worked in have not been so fortunate and running applications such as UnrealIRCd becomes impractical due to the extremely unforgiving ping timeout thresholds.

I have read from various sources that UnrealIRCd currently enforces a strict 2-strike rule when it comes to missed pings, "missed pings" being defined as pings that did not reciprocate a reply in a predefined amount of time and therefore timed out. Currently the only custom user-defined value that can be set here is the interval between pings, or pingfreq, which is obviously an extremely limited feature set when taking into account the other variables here that could and should easily be user-defined, as well, for a higher range of flexibility, especially as it pertains to mobile communications.

For better flexibility and resiliency, especially for mobile communications, I recommend also allowing custom user-defined values for how many "missed pings" constitute reason for disconnect as well as how long to wait for each ping to consider it timed out, in milliseconds. All of the currently defined defaults are completely acceptable and should remain as the defaults, but the aforementioned custom values should be allowed to deviate from the defaults as needed.

0005297: Hook for denying (ch)mode flags

Wed, 20 Nov 2019 17:17:56 +0100

In my case I'd be using it to deny +o and higher from getting banned, except if it's done by +a/+q/opers. Maybe something like:
HookAddChmode(modinfo->handle, 'b', ......)

Then the specified callback will receive all and only +b related stuff. =] It should probably be able to work with any mode flag though.

0005348: U5: X509_check_host()

Wed, 20 Nov 2019 15:55:32 +0100

Check if X509_check_host() code can be used when X509_check_host is available. Right now we always use the "missing X509_check_host" code.
So add in autoconf and
1) verify the configure code (both missing and present)
2) verify the X509_check_host() code actually does it job

0005301: U5: Profiling

Fri, 15 Nov 2019 06:58:19 +0100

Do some actual profiling to see where the weak spots are with regards to performance.