View Issue Details

IDProjectCategoryView StatusLast Update
0001588unrealircdpublic2004-03-19 17:36
ReporterJoolz Assigned To 
PrioritynormalSeveritymajorReproducibilityhave not tried
Status closedResolutionopen 
Summary0001588: SJOIN errors, Nick Collisions
Description[15:00:51] --- underground.hub.eu.zirc.org - Missing user can in SJOIN for #HardRockCafe from darkstar.de.eu.zirc.org (@V ~ !10Errq #HardRockCafe :can )
[15:00:51] --- underground.hub.eu.zirc.org - Missing user can in SJOIN for #zurna from darkstar.de.eu.zirc.org (@V ~ !{eizy #zurna :can )
[15:01:42] --- underground.hub.eu.zirc.org - *** Global -- Nick change collision from JAF to JustAnotherFan (darkstar.de.eu.zirc.org 1077618792 <- darkstar.de.eu.zirc.org 1077634902)
[16:44:27] --- underground.hub.eu.zirc.org - Missing user b in SJOIN for #nmp3s from darkstar.de.eu.zirc.org (@V ~ !}6EgC #nmp3s :b)
[16:44:27] --- underground.hub.eu.zirc.org - Missing user b in SJOIN for #somethingawful from darkstar.de.eu.zirc.org (@V ~ !wTnu6 #somethingawful :b)
[16:44:27] --- underground.hub.eu.zirc.org - Missing user b in SJOIN for #crackhouse from darkstar.de.eu.zirc.org (@V ~ !zZ}q} #crackhouse :b)
[16:44:27] --- underground.hub.eu.zirc.org - Missing user b in SJOIN for #idlegames from darkstar.de.eu.zirc.org (@V ~ !ymFnT #idlegames :b)
[16:44:27] --- underground.hub.eu.zirc.org - Missing user b in SJOIN for #muthafukka from darkstar.de.eu.zirc.org (@V ~ !ymGQY #muthafukka :b)

With the SJOINs I've noticed that it's happening only with darkstar.* at the moment..

(darkstar is a leaf of undeground.)

Nick Collisions:
On darkstar it looks like this:
11:32:43 -!- baafie is now known as baafie[away]
11:32:43 * baafie[away] is away: out
11:32:43 -!- baafie[away] [[email protected]] has quit [Killed
          (underground.hub.eu.zirc.org (baafie[away](?) <-
          darkstar.de.eu.zirc.org[62.80.114.24]))]
11:33:04 * EienWork pokes Joolz
11:33:09 <@Joolz[gone]> ow
11:33:19 <@EienWork> Do you know why that just happened?
11:33:31 <@Joolz[gone]> I don't know, you're mean and felt like poking me?
11:33:39 <@EienWork> No, no... Look what happened to baafie
11:33:49 <@Joolz[gone]> yes, he's demented isn't he?
11:34:10 <@Joolz[gone]> you mean that?
11:34:18 <@EienWork> I mean him getting killed.
11:34:31 <@Joolz[gone]> hey, how did you know about it before it happened?
-------
On underground: (same time, different timezone)
[16:32:11] <EienWork> That's not good.
[16:33:03] * EienWork pokes Joolz
[16:33:08] <Joolz[gone]> ow
[16:33:18] <EienWork> Do you know why that just happened?
[16:33:30] <Joolz[gone]> I don't know, you're mean and felt like poking me?
[16:33:39] <EienWork> No, no... Look what happened to baafie
[16:33:48] <Joolz[gone]> yes, he's demented isn't he?
[16:33:55] <-- baafie ([email protected]) has quit (Killed (NickServ (Nick kill enforced)))
[16:34:09] <Joolz[gone]> you mean that?
[16:34:17] <EienWork> I mean him getting killed.
[16:34:30] <Joolz[gone]> hey, how did you know about it before it happened?
(I don't know quite why Nickserv killed 'baafie' as baafie had already identified hours before.)

I thought that relinking underground and darkstar might help , I had this error shortly afer when I did:
[18:30:18] --- underground.hub.eu.zirc.org - Missing user baafi in SJOIN for #evil-zirc from darkstar.de.eu.zirc.org (~ !zxaFj #evil-zirc +sntirRQSC :@baafi)
[18:32:35] --- underground.hub.eu.zirc.org - *** Notice -- do_numeric( 401, darkstar.de.eu.zirc.org, darkstar.de.eu.zirc.org, 2, { darkstar.de.eu.zirc.org, undergroun } )!
Additional InformationAll servers complied with SSL, Remote Includes, ZIP
ircservices-5.0.28 services.zirc.org :build #1, compiled Sun Feb 15 08:05:53 EST 2004

Underground & darkstar: libcurl/7.11.0 OpenSSL/0.9.7c zlib/1.2.1
Unreal3.2-RC1. darkstar.de.eu.zirc.org :Fhin6XeOoZE
Unreal3.2-RC1. underground.hub.eu.zirc.org :Fhi6XeOoZE
ReleaseID (1.1.1.1.2.1.2.34 2004/02/14 20:22:37)
Actual Numerics:
Underground - 3
Darkstar - 31
3rd party modules

Activities

Joolz

2004-02-24 14:02

reporter   ~0005180

Oh, both are using linux-2.4.25.

codemastr

2004-02-24 18:01

reporter   ~0005186

Could you do:
/tsctl alltime

and paste the output?

Joolz

2004-02-24 18:04

reporter   ~0005187

TSCTL ALLTIME output:

[23:02:35] --- underground.hub.eu.zirc.org - *** Server=underground.hub.eu.zirc.org TStime=1077663754 time()=1077663755 TSoffset=-1
[23:02:35] --- darkstar.de.eu.zirc.org - *** Server=darkstar.de.eu.zirc.org TStime=1077663754 time()=1077663755 TSoffset=-1
[23:02:35] --- ferret.uk.eu.zirc.org - *** Server=ferret.uk.eu.zirc.org TStime=1077663754 time()=1077663756 TSoffset=-2
[23:02:35] --- entropy.hub.us.zirc.org - *** Server=entropy.hub.us.zirc.org TStime=1077663754 time()=1077663755 TSoffset=-1
[23:02:35] --- avalon.hub.ca.zirc.org - *** Server=avalon.hub.ca.zirc.org TStime=1077663754 time()=1077663755 TSoffset=-1
[23:02:35] --- devo.ny.us.zirc.org - *** Server=devo.ny.us.zirc.org TStime=1077663754 time()=1077663755 TSoffset=-1
[23:02:35] --- vaix.va.us.zirc.org - *** Server=vaix.va.us.zirc.org TStime=1077663753 time()=1077663759 TSoffset=-5
[23:02:35] --- lunchbox.il.us.zirc.org - *** Server=lunchbox.il.us.zirc.org TStime=1077663754 time()=1077663756 TSoffset=-2
[23:02:35] --- nemesis.ca.us.zirc.org - *** Server=nemesis.ca.us.zirc.org TStime=1077663753 time()=1077663825 TSoffset=-71
[23:02:36] --- deliverance.ca.us.zirc.org - *** Server=deliverance.ca.us.zirc.org TStime=1077663754 time()=1077663755 TSoffset=-1
[23:02:36] --- bones.wa.us.zirc.org - *** Server=bones.wa.us.zirc.org TStime=1077663754 time()=1077663755 TSoffset=-1

syzop

2004-02-26 10:54

administrator   ~0005215

seems like darkstar thinks various users exist, sends them in SJOIN etc...
When another server receives stuff from those users it will say "Missing user <whatever> in SJOIN" because <whatever> doesn't exist...
and...
11:32:43 -!- baafie[away] [[email protected]] has quit [Killed
(underground.hub.eu.zirc.org (baafie[away](?) <- darkstar.de.eu.zirc.org[62.80.114.24]))]
this too is caused by a server receiving something from an unknown user ('baafie[away]') [done by remove_unknown()]

in short, something looks seriously f*cked up...
I wonder if other people/nets have this too, especially since, as you say, darkstar* is every time involved here one would think it's something box-specific..

Especially the:
I thought that relinking underground and darkstar might help , I had this error shortly afer when I did:
[18:30:18] --- underground.hub.eu.zirc.org - Missing user baafi in SJOIN for #evil-zirc from darkstar.de.eu.zirc.org (~ !zxaFj #evil-zirc +sntirRQSC :@baafi)
[18:32:35] --- underground.hub.eu.zirc.org - *** Notice -- do_numeric( 401, darkstar.de.eu.zirc.org, darkstar.de.eu.zirc.org, 2, { darkstar.de.eu.zirc.org, undergroun } )!"

Is weird and pretty much says it's not something only in simple /join, but happends during synching too... I also haven't seen the do_numeric() notice before in my life ;).

Btw, you didn't have this before? like with beta19/18/etc?

There's also a small chance of race conditions, although I cannot entirely see how it would happen, but if you get these messages a lot suddenly after upgrade to RC1 (which seems to be the case?) then I would guess it's not, but...!

Fun!

codemastr

2004-02-26 12:10

reporter   ~0005217

Yeah it certainly seems odd. If there were such huge problems, it would seem other people would have reported this as well...

I too am leaning towards a problem with that particular machine.

Joolz

2004-02-27 10:44

reporter   ~0005224

We didn't have problems with beta19 or any of the of the previous betas..

We moved darkstar to another hub from underground and that seems to have minimized the problem (but it hasn't gone away completely)..

White_Magic

2004-02-27 10:45

reporter   ~0005225

yeah, ive seen a few ( and only a few 3-10 ) from the net im oper on of " SJOIN problems ", mostly when a user connects and it auto killed via mirc on the connect, and they attempt to join a channel, it often gives out a error msg suck as

Missing user b in SJOIN for #nmp3s from darkstar.de.eu.zirc.org (@V ~ !}6EgC #nmp3s :b)

not exactly the same but, Similar to the SJOIN...
only due to the fact a oper with one of my auto killed addons kill them,

Ive also found the same when servers get linked together, and services comes into effect on the previous unlinked server, they r killed and the same kind of msg is generated, but it doesnt always happen O_o

syzop

2004-02-27 12:25

administrator   ~0005228

Yeah what I ment with "coincidence" is that the following could happen, lag between server 1 and server 2 is 0.06sec in this example, time is absolute:
From server 1 side:
0.00 User connects
0.10 User joins #a,#b,#c
0.12 Receives kill

From server 2 side:
0.06 Receives new user, sends a kill back immediately
0.16 Receives a SJOIN for the user that doesn't exist anymore (was killed)

This is probably also the reason why it's in the junk snomask because it can happen in "normal cases". The thing that seemed weird of Joolz story is that it happened a lot and even during synching and stuff (although that is on 2nd thought not very unusual), and that it suddenly increased after upgrading.. it may still be this "normal stuff" however :).

codemastr

2004-02-27 14:11

reporter   ~0005229

Perhaps, but I've _never_ seen that do_numeric thing. As far as I know, that debug message has been in Unreal since before there was an Unreal. So if in 4+ years no one ever reported seeing that message, something certainly seems odd about it showing up now.

Joolz

2004-02-27 14:28

reporter   ~0005230

I doubt it's lag between darkstar and underground, a) they're so close, the rpings are usually between 0 and 1. and b) we've relinked darkstar to a different hub much further away and the number of errors have decreased so I don't think it's likely to be lag..

I thought I posted this earlier but it appears not. I've had other do_numeric errors which didn't occur around syncing and they were very consistant as well (ie the same action produced the same error again and again):

Feb 25 16:10:34 --- [dave] ([email protected]) : vbBot
Feb 25 16:10:34 --- [dave] is connecting from *@<removed>.freedom2surf.net
Feb 25 16:10:34 --- underground.hub.eu.zirc.org - *** Notice -- do_numeric( 307, darkstar.de.eu.zirc.org, darkstar.de.eu.zirc.org, 2, { darkstar.de.eu.zirc.org, Joolz } )!
Feb 25 16:10:34 --- []
Feb 25 16:10:34 --- [dave] ^#ntl.talk ^#subfusc
Feb 25 16:10:34 --- [dave] darkstar.de.eu.zirc.org (ZiRC Germany)
Feb 25 16:10:34 --- dave :is a Bot on ZiRC
Feb 25 16:10:34 --- [dave] idle 01:34:24, signon: Wed Feb 25 14:36:09
Feb 25 16:10:34 --- [dave] End of WHOIS list.

(I'm on underground by the way)

syzop

2004-02-27 14:31

administrator   ~0005231

I've never seen the do_numeric() stuff either, on a sidenote... this "lag thing" only occurs when there's a kill (eh yes or newly added *line) involved, so... if you didn't see a kill it can't be that (it cannot be nickchanges since nickchanges come from the "source" even in case of svsnick blablabla)

Joolz

2004-03-03 12:48

reporter   ~0005287

I'm not seeing any SJOINs or nick collisions at the moment but apparently darkstar is losing some PRIVMSGs, especially DCC requests.

syzop

2004-03-19 00:05

administrator   ~0005542

so how is it going with this? (now you are running RC2)

baafie_

2004-03-19 17:27

reporter   ~0005548

AFAIK, we haven't experienced any more problems.

syzop

2004-03-19 17:36

administrator   ~0005549

Roger :p

Issue History

Date Modified Username Field Change
2004-02-24 14:00 Joolz New Issue
2004-02-24 14:02 Joolz Note Added: 0005180
2004-02-24 18:01 codemastr Note Added: 0005186
2004-02-24 18:04 Joolz Note Added: 0005187
2004-02-26 10:54 syzop Note Added: 0005215
2004-02-26 12:10 codemastr Note Added: 0005217
2004-02-27 10:44 Joolz Note Added: 0005224
2004-02-27 10:45 White_Magic Note Added: 0005225
2004-02-27 12:25 syzop Note Added: 0005228
2004-02-27 14:11 codemastr Note Added: 0005229
2004-02-27 14:28 Joolz Note Added: 0005230
2004-02-27 14:31 syzop Note Added: 0005231
2004-03-03 12:48 Joolz Note Added: 0005287
2004-03-19 00:05 syzop Note Added: 0005542
2004-03-19 17:27 baafie_ Note Added: 0005548
2004-03-19 17:36 syzop Status new => closed
2004-03-19 17:36 syzop Note Added: 0005549