Bug 4232 - Slow clients got disconnected on mapchange
Status: RESOLVED FIXED
Alias: None
Product: ioquake3
Classification: Unclassified
Component: Platform
Version: unspecified
Hardware: PC Linux
: P3 normal
Assignee: Zachary J. Slater
QA Contact: ioquake3 bugzilla mailing list
URL:
Depends on:
Blocks:
 
Reported: 2009-07-08 12:55 EDT by Dmitry Ermakov
Modified: 2011-02-08 13:31:25 EST
2 users (show)

See Also:


Description Dmitry Ermakov 2009-07-08 12:55:46 EDT 
With warmup enabled(g_warmup 5), if a client is not in-time to load before match starts, it gets  disconnected after ~2 seconds after spawn on map.
This doesn't happen with 1.32b.

Steps to reproduce:
1. Latest SVN build of ioq3ded.i386 (checked from rev 460 till current, all affected)
2. Clean Q3 install,  only pak0-pak8 in baseq3 + ioq3ded, no QVM or .so mods (problem persists with mods running aswell)
3. g_warmup 5 (or less, if needed)
4. Connect to server and issue vote for the nextmap, or just manually change a map.
5. If client wont load in less than 5 seconds, it will get disconnected


If you do the same under id q3, you will just see quick "Fight" message, and everything's fine after.
Comment 1 Tim Angus 2009-07-08 13:20:54 EDT 
The changes at r460 don't look like they could cause the problem you're seeing. Are you sure this is the *exact* revision the bug starts to occur?
Comment 2 Dmitry Ermakov 2009-07-08 13:50:37 EDT 
(In reply to comment #1)
> The changes at r460 don't look like they could cause the problem you're seeing.
> Are you sure this is the *exact* revision the bug starts to occur?

Sorry, i meant that rev 460 was last which started successfully. Lower rev's are failing with SEGFAULT after qvm load
Comment 3 Dmitry Ermakov 2009-07-08 19:47:35 EDT 
(In reply to comment #1)
> The changes at r460 don't look like they could cause the problem you're seeing.
> Are you sure this is the *exact* revision the bug starts to occur?

Tim, I've found it out, rev 91, your patch. Rev 90 works just fine, in 91 it goes like this: you respawn on map, but lagometer shows green 999, you cant move. Then after some time, usually less than minute, everything normalizes.

And, as i suppose (i couldn't check that), after rev 192 it starts just to kick after 2 secs, because r192 seems to be the only commit on that bug since r91.
Comment 4 Tim Angus 2009-09-18 18:25:42 EDT 
As you say this is related to the changes in r91, but also bug #2454 which was resolved by r192.
Comment 5 Dmitry Ermakov 2009-10-01 00:43:11 EDT 
Yes, it started from r91, then after 192 behaviour changed, no visible changes after 2454 and it remains the same in current builds, you saw it youself.
Comment 6 Tim Angus 2009-10-15 19:40:35 EDT 
This should be fixed in r1670, but please check it and reopen if you still have problems.
Comment 7 Dmitry Ermakov 2009-12-02 09:17:27 EST 
Issue still exists, though behaviour changed a bit. Now client respawns on map, and scoreboard shows normal ping, but client itself has frozen picture and 999 on lagometer.
Comment 8 Tim Angus 2010-01-07 06:17:27 EST 
The client should eventually unfreeze though?
Comment 9 Tim Angus 2010-02-15 11:17:09 EST 
Reclosing due to lack of further information.
Comment 10 Thilo Schulz 2011-02-08 13:31:25 EST 
Additional note: this was a bug which concerned all games. It is fixed in latest SVN.