zombies new incarnation?

[taxman]

current beta. uptime less than hour. stop by right-click on tray icon than 'stop'.

Code: Select all

No clock movement in 30 seconds.  Dumping thread status.
*Thread Info*
Semaphore PID: 0
Scripts Thread Checkpoint: 55
Last Script: scripts/misc/logoff.ecl PC: 207
Escript Instruction Cycles: 469767871
Tasks Thread Checkpoint: 9
Active Client Thread Checkpoint: 116
Current Threads:
2404 - ThreadStatus
2576 - Main
Waiting for 154 child threads to exit

[taxman]

79 clients online. 'Stop' selected.

Current Threads:
1388 - ThreadStatus
2952 - Main
Waiting for 471 child threads to exit

[Nando]

Taxman, please review your logoff.src. It might have been stuck into some infinite loop. If that's not the case, please report while the whole thread status.
(And with the few lines near the reported PC of the script)

[Nando]

And please, add the "[threadstatus]" command to your console.cfg (it's the same as [lock/unlock]) and paste here a few reports when the shard is running so we can compare with them afterwards.

Thanks!

[taxman]

I think we have serious problem somewhere near changes about listener. From friday evening i start current beta on live server and players have big troubles with connect when players number is approx. 200+.
Sometimes it hangs on 'verifying account', next try it may hangs on selecting shard.
In first case i do not see even 'client connected' in console, in second - here is connected/disconnected (login server) and no 'connected' then (game server). If you can connect - you can play normally. The same players may connect and may not so it is not ploblems with providers/network..

[taxman]

Nando, i found your secret lines in pol.log

Code: Select all

error in create_thread: 12 8 "Not enough space" "Not enough space" 5947 5647584 SocketClientThread 5635168 0 793810800

P.S. POL uses 1.4G RAM from 2G onboard.

[MuadDib]

nando will be commiting some changes soon. I will build a special core and send it to you. Remember to enable Variable memory dump type too

[Nando]

Well, this is caused because when you run out of memory, POL can't create new threads but was increasing the thread count nevertheless. Fixed this, and it is probably fixed in other places.

Now... this doesn't fix the out-of-memory problem. My best bet would be that 098 uses more memory (naturally, even though it has less leaks), and your memory is possibly getting too fragmented for Windows to find a contiguous space where it could create a new thread. I *THINK* it would help if you used the WinLFH, but that
needs 2003 or XP. :/