BUG cries on Bugfarm Fortress and Sumo

[Z-Man]

Code: Select all

[4] [sick]Lucifer: BUG - 5 users named in server browser, but 2 shown in column

Durka, Killa! and Player 3 came from the same client, and the server swears it told you there were three clients online. Nemo filed a bug like this in the SF database, I'm waiting for a clientside recording.

No, grepping for "magic ping" didn't show up much What is is it? Manta had a ping of about 300, but that alone can't be magic. I noticed the ping charity was rather high (around 180), which means you took over a large portion of his ping. There is the possibility to cheat the ping charity mechanism, but only with nasty client code changes.

[Lucifer]

Um, I might be running a cvs checkout that's older. Let me update before you get too worried.

Magic ping = Manta won all the ping races. You know, where you're side by side or racing to the same point, and you go ahead and push it and let ping pick the winner. This is unusual to me, Manta usually loses those races with me.

I'm chalking it up to me being sick. It's a joke, hoping to bring you a smile while grepping the logs.

[Lucifer]

Ok, updated cvs, but unfortunately none of your servers showed inaccurate counts. Recording attached anyway, I stepped through each server that showed people in the master server browser so I could count and compare, probably enough time hovering for you to do the same. There were discrepancies, but the servers are out of your control and I didn't go in. Anyway, if it's my client, recording should show it, right?

[Z-Man]

Sorry, deviations of one on a server that's not mine or where I can request a debug recording aren't of much use here. Small devialtions are normal, spectators make the client count one more than the player count, and several people playing on one server make the player count bigger.

I don't say this bug doesn't exist, but to really track it down, I need something like this:

http://sourceforge.net/tracker/download ... id=1466657

If possible with a clientside bug recording and you entering the server immediately and clamoring about it.

[Lucifer]

Heh. I'll try to make sure I'm recording for my first server browser look for awhile. If I dont' turn anything up in a week or so, we might need to stage something.

[dlh]

Looks like it was off by one in this recording.

[Z-Man]

From when is your checkout? It doesn't play back with current CVS HEAD And just before you entered the server, this happened:

Code: Select all

[15] Player 1 plays for Team blue.
[15] Player 2 plays for Team gold.
[15] Player 3 plays for Team blue.
[15] Player 4 plays for Team gold.

Four players from one client
The user count you saw should have been 15, and the number of players, not counting P1-P4, should have been 14.

[dlh]

From when is your checkout? It doesn't play back with current CVS HEAD :( And just before you entered the server, this happened:

Code: Select all

[15] Player 1 plays for Team blue.
[15] Player 2 plays for Team gold.
[15] Player 3 plays for Team blue.
[15] Player 4 plays for Team gold.

Four players from one client :(
The user count you saw should have been 15, and the number of players, not counting P1-P4, should have been 14.

0.2.8.1. The user count was 14, then I refreshed and it was 15.

[Z-Man]

I think I know what it is, and it is a server bug. When counting the number of users, the server excludes clients that, according to some old logic, would be kicked if a new client joined. One of the criteria for that was that a certain amount of network packet timeouts had occurred, and that happens frequently between rounds when there is heavy network traffic from object deletions and other stuff (which we probably should optimize away some time). Luckily, the timeouts are no longer a criterion for kicking a user out, otherwise the lower displayed number would mean that you can push your way into a full server I'll fix this for 0.2.8.3.

Edit: to let you verify it, I also updated Fortress to that code, although it is mostly running the code targetted for 0.2.8.2. I didn't have the heart to force a restart, it already did one half an hour ago So you'll have to wait four hours before you can test the new code.

[Z-Man]

Code: Select all

[7] SuBlImInAl.FreQ: BUG: REQUEST BAN, THIS POSER IS GOING to each server
[7] SuBlImInAl.FreQ: BUG: ZMAN, plz add paco, kolley, and kevin's ips to the ban list in forums
[7] SuBlImInAl.FreQ: BUG: IF KEVIN's IP MATCHES THE OTHERS THEN BAN

All three of them came from the same IP, apart from using language Tank wouldn't want to see on this forum (and I'd expect you to use the silence feature for), they all were in permanent chatbot mode as far as I can tell. I can't ban them on the forum (and wouldn't see a reason to do so anyway), but on my servers, this guy is now forced to bear the name JoLLeY.

Hmm, should perhaps, if a kick for one player from an IP succeeds, all other players from that IP get automatically kicked as well?

[subliminal]

nah, we kicked 1, the others remained

[Z-Man]

Yes, I saw that, the question was whether this is the right behavior.

I'll let you figure out this one for yourself

Code: Select all

[5] featherfcuk: core dumped Gnorty for 2 points.
...
[2] Gnorty: BUG! I was alive when the report of my deat arrived!
[5] featherfcuk: hehe
[2] Gnorty: BUG from previous round

[Jonathan]

Makes me wonder what happened.

[gnorty]

Round started, I was on the grid and the client annonced my demise.

In my shock at this, I died ifor real! hence 2 deaths in a single round for me!

[Z-Man]

This was your real death message, sorry, I forgot it:

Code: Select all

[0] Driverr core dumped Gnorty for 2 points.

Now you can play "find the difference".