Netchan_checkforcompletion Problem

Warhammer

<div class="IPBDescription">reserved players get kicked on new map</div> Hi all,

Running 1.04 linux on RedHat 7.2 with the fixed AdminMod libs right now. However since 1.03, there is occasionally a problem where at the end of a game when the map changes, the following message comes up in the console:

Netchan_CheckForCompletion: Lost/dropped fragment would cause stall, retrying connection

This is fine for normal players, as the HL client just reconnects with no problem. But for special AdminMod players (admins, reserved slot players, etc.) the connection retry doesn't invoke the autoexec to re-read the pw-home, etc. and they get kicked. I tried searching the forums and other net resources but haven't seen this documented anywhere. I also run a TFC instance on the same box (multi-CPU server) with no problems or errors, so it seems NS specific offhand. Any thoughts?

Warhammer

Max1

Don't set people with the reserved nick permission and it wont be a problem

Warhammer

The admins, clan members, and paying members all get reserved slots based on their WON id (or name in some cases), so that's not really an option. Or are you saying something else?

GoleX

No, the reserved nick option will not allow someone with the WONId/Name listed in the users file to connect if they don't have the correct setinfo line running. It has nothing to do with the reserved slots option.

Warhammer

That's what I thought he was saying. That's not the problem - everyone that should be able to connect with the setinfo pw can. What happens is that at the end of the map, there appears to be a dropped packet of some sort, and the client has to attempt to reconnect to the server (and this happens automatically). When this happens, it has already attempted a connection and the setinfo pw-home variable has been erased from the client (as normally happens), so it kicks reserved players.

Sounds like I am the only one in the world (besides the players on my server) who have seen this problem?? Also allow me to clarify - this is <b>not</b> a case of the user's pw-home not being part of autoexec.cfg or similar. This problem happens only perhaps 50% of the time, and when it is ok the Netchan_CheckForCompletion error is not displayed.


Netchan_CheckForCompletion: Lost/dropped fragment would cause stall, retrying connection
Commencing connection retry to tv5.theville.org:27020
Connecting to tv5.theville.org:27020 ...
Connection accepted by 66.153.28.3:27020
[ADMIN] That name is reserved on this server.