PDA

View Full Version : Multiple clients => 1 zone == Sporadic disconnect?


Malevolent
04-26-2002, 04:11 PM
This problem doesn't happen all the time, but does so for the majority. The behavior occurs in a depopped or fully spawned zone.

Problem:
When two (or more) clients are in the same zone, then either one is disconnected. Usually, when the other client zones in.

I've also been with another client and helped them sack freeport. So I know that most likely it isn't my tweaks. If anything, I'd say it has to do with how the client has handled when they first zone in -- perhaps related to the 'stickman zone in' when the client becomes a stuck camera object within a zone (another rarity -- once a week for me)?

Anyone else see or have this problem? I've seen it with a nonmal-tweaked zone server, so I'd think I wouldn't be the only person with this problem.

I'm wondering if increasing the client timeout def has any affect?

Malevolent
04-27-2002, 02:07 AM
I added another digit to the client timeout define. This seems to largely solve the problem.

Malevolent
04-27-2002, 05:56 AM
I stand corrected. It didn't really help. I have noticed that lag nearly doubles.

Which leads me to believe that maybe its because of all the invocations to entity_list. But, that doesn't explain why it also happens in the arena.

Malevolent
05-06-2002, 02:16 PM
The problem exists centrally around the network code for I am quite capable of logging in from an external location and bring in other clients to the same zone as myself with no problems, outside of the visual lag induced from zone wide activity. Thus, one might suspect that the network code might have an issue or two, but this is a known issue.

To quote the hogie, "Our network code sucks."

Therefore, for anyone who finds this thread here is the solution: do not play on the same box that you are hosting from.

--MV

Merkur
05-06-2002, 09:44 PM
I think in this case the problem is not our networkcode but the EQ-Client. It eats up all CPUtime and the server is slowed down. Try to set EQClient priority to very low and all EQEMu threads/processes to high - I haven't tested this, but it could help.