That's a lot of fast disk storage for an HLDS server. What else are you running on that box? WTF? 4 UW SCSI drives just for an HLDS server??
I'd guess you're using this box for other purposes as well as HLDS with that much storage, and those processes are probably what is causing your HLDS server to lag. If this machines is dedicated to HLDS, why all the hi-po storage? Are you using software mirroring or striping of the drives? About the only other thing that comes to mind is maps. How many maps are in your rotation? If you're running a shit load of maps, that might cause it. When this happens on our Linux server, it's usually due to low memory, and a lot of VM paging. I would suggest setting up PerfMon to log disk activity, VM usage, and physical memory usage. If you don't know perfmon very well, install terminal services in admin only mode, install the RDP client on your remote wks, connect to the terminal server and launch task manager. Then watch the resource usage in realtime when the map changes. Better yet, I'd recommend you actually run your HLDS server from a term service session anyway. That way, you can remote into the machine, and have direct access to the HLDS console. When you kill your RDP session, all your user processes still run, and you can reconnect to the session anytime. I'm not currently running a win32 HLDS W2K server, but I do run 3 seti@home instances on my W2K IIS server thru an RDP initiated session. I logon thru an account with admin privs, so I can completely manage the server via RDP. The machine is headless (no display connected), and I can reboot it remotely thru RDP, and reconnect after it boots. It's NICE. If you're not using terminal services to admin that box, you should be! StanTheMan TheHardwareFreak www.hardwarefreak.devastation.cc [EMAIL PROTECTED] -----Original Message----- From: Ronnie Enriquez [mailto:[EMAIL PROTECTED]] Sent: Tuesday, October 09, 2001 3:55 PM To: [EMAIL PROTECTED] Subject: Timeout During Map Changes Hi Everyone! I'm not sure if this was previously addressed or not.... but... In a nutshell, my CS server takes way too long to change maps. It is NOT because no one is pressing fire. I'll be in the server myself and will press fire to initiate the map change. I know the map is changing because i get a "flatline" on my netgraph. It is NOT due to the custom.hpk size. It is 7k. In addition, I have disabled uploads. Sometimes, it can take up to 1 minute to change maps. When this happens, people have reported seeing the dreaded message: "Warning: Connection Problem..." They will see this for a few seconds and will go away. When it goes away, they'll see the LOADING... message on the screen. Then they will enter the new map. Some people won't even get this far. They'll just be dropped altogether. When I do a manual changelevel, it takes even longer! The server does not crash during map change. Since the server is co-lo'd at a data center, I cannot physically see or hear if the HD is crunching away during the delay period. So I really don't know if there's anything happening on the server end. Bandwidth is not an issue. There's plenty of it. Also, some people say that this is a known issue. On the other hand, I've seen plenty of other win32-based CS servers out that don't have this issue. Anyone want to sell me a clue? BTW: Everything listed above has been done with a fresh new install of the OS, HL, and CS. My CS server is running on a fairly decent system: ---Hardware--- Pentium3 / 1GHz Asus CUSL2 Motherboard 768MB SDRAM Adaptec 2940UW SCSI Controller 2 x Seagate Barracudda 9.1GB UW-SCSI Drives 2 x IBM Deskstar 18GB UW-SCSI Drives ---Software--- Win2K Server SP2 HLDS 4.1.0.8a Counter-Strike 1.3 Full 20 players -ronnie
