Re: [hlds_linux] Speakeasy inquiry
Eric (Deacon) wrote: eeek. that surely had to be an issue w/ your phone lines/distance from co or something else out of speakeasys control, unless it was a fleeting issue. Fleeting? Main Entry: fleetĀ·ing Function: adjective Date: 1563 : passing swiftly : TRANSITORY ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] Is anti-terrorism killing my ping?
Andrew Smith wrote: Thats probably exactly right. You can also try using tracepath instead of traceroute. It will mark the async hops for you. While none of the rest of your message really meant much to me (gg ignorance on my part), I'm very interested to hear about this tracepath utility. What is it? Where can I get it? It's part of iputils on RH and I think also on Mandrake, but I have no idea about Debian, FreeBSD etc. akira:~# apt-cache search tracepath iputils-tracepath - The tracepath utility from iputils ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] init3
Mad Scientist wrote: On Sunday 01 December 2002 05:04 am, Florian Zschocke wrote: lastshooter wrote: errrnow that I can run my server without x, how do I access another terminal, say to run "free"? From the console? Most Linux distros come with six or more consoles enabled. You can switch between them with Alt-F1 to Alt-F6. I believe it's Ctrl-Alt-F1 etc. ALT-x (Where X is the console identifier) is for switching between text consoles outside of X. If you are in X, CTRL-ALT-x (where X is the console identifier) is required to switch out of X and back to the text console. Aluve, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] [OT] dust clumps
Matt wrote: Forgive me if I am wrongand I might be, but hasn't the hard drive industry standardized on like a 1 year warranty for IDE while most SCSI is still 3 to 5 years? Yes, but for recent purchases only. If you purchased an IDE drive 2 years ago with a 3 year warranty, that drive is still under warranty, irregardless of the current warranties that drives come with. ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] re: non-changeable server variables
m0gely wrote: Let me try again. By *just* enabling metamod, the hlds process shows some 10~12% higher cpu usage, using top then without metamod. There are no metamod plugins installed, only metamod itself. meta list shows nothing. Out of curiosity, have you tried recompiling metamod from sources against the SDK with processor specific compiler optimization flags? I have been pondering going through the list of things I use with Half-Life and recompiling everything I can get the source to, and attempting to determine if that helps with the load or not. Then again, if somebody has already done this, and determined it to be crap, I won't bother. :) Sincerely, Warren ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] Half-Life Server Graphing
Prodigy wrote: As many of you know MRTG is pretty cool, but I think this blows its socks off :) http://traffic.lanvancouver.com Looks excellent. Good choice with RRDTool. Do we get to guess at the source and write our own, or might it be possible to release this lovely tool to the world at large so that I might start also enjoying being a statwhore on my own server usage. ;) Aluve, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] Linux Router
Josh wrote: But if you ever get it working as a router, you'll have to post back to this mailing list and let me know :) > >>Thanks :) but its a linux box who is the router, what i have to do in >>order to forward port 27015? > >> > Is there somebody who managed to host a HLDS behind a linux >> > router?? How he done that? :D (=> commands with iptables?) Actually, it's not acting as a router if the server is behind the linux box, but rather a NAT/PAT device. The following IPTABLES commands assume the following : a) Your internal network is 192.168.0.1/24 on eth0 b) Your external IP is static at is 1.1.1.1 on eth1 c) Your CS server is at 192.167.0.100 on port 27015 d) You have the nat, filter and mangle iptable modules either static compiled into your kernel, or available as loadable modules. The commands assume that this is cut/paste into a file, and iptables-restore < filename is used to read in the rules. -- Beginning *nat # Redirect incoming UDP port 27015 traffic to the CS server. -A PREROUTING -i eth1 -p udp -m udp --dport 27015 -j DNAT --to-destination 192.168.0.100:27015 # Setup the source nat so that computers on the internal # network have internet visibility. -A POSTROUTING -o eth1 -j SNAT --to-source 1.1.1.1 COMMIT *mangle # The following is for setting the priority on the HL packets # both in and out. It's not the same as a decent QoS setup, # but it does help by about 20-30ms, depending on the speed # of the machine. -A PREROUTING -i eth1 -j TOS -p UDP -m udp --sport 27015--set-tos Minimize-Delay -A PREROUTING -i eth1 -j TOS -p UDP -m udp --dport 27015 --set-tos Minimize-Delay -A POSTROUTING -o eth1 -j TOS -p UDP -m udp --dport 27015 --set-tos Maximize-Throughput -A POSTROUTING -o eth1 -j TOS -p UDP -m udp --sport 27015 --set-tos Maximize-Throughput COMMIT *filter # This rule returns an immediate failure for the IDENT port, speeding # up connections to places that still check IDENT responses, IRC # servers being one. -A INPUT -d 1.1.1.1 -i eth1 -p tcp -m tcp --dport 113 -j REJECT --reject-with tcp-reset # And this rule drops everything that's not implicitly accepted by the # NAT or filter rules. IE, your network becomes a blackhole on the # 'net. -A INPUT -i eth1 -m state --state INVALID,NEW -j DROP -- END Due to wrapping, the rules may end up a bit confused. All lines start with either a -A or a #, so if you see something like : -A foo --barbaz The actual line is : -A foo --barbaz I hope this answers your questions. Aluve, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] POST YOUR SERVER CONFIG - NOW ;)
Tyler "Overkill" Schwend wrote: > Sounds like an attempt to get some RCONs to me. exec sensitive_crap.cfg Then you can happily pass around your server.cfg without having to remember to out those passwords. :) Aluve, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] won down again?
Jay Anstiss wrote: > Haspers - I believe it was Eric at Valve that said it could take upto 48 > hours for the DNS changes to take effect. I'm sure this will be cleared up > before we know it. 48 hours for DNS propigation throughout the 'net to take place. Authoritative nameservers should have the updated records immiedately, because they are the root of authority for the zone record. If they are still reporting the old IP, then either they haven't been changed, or it's time to invest in new DNS software. :) ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] Increase priority of hlds??
Mike Hodgkinson wrote: > Just wondering if its worth nice'ing hlds higher because each time I do a grep on my >large log files or zip them all up it causes lag on the servers? > Any ideas? I already run my servers nice'd. I give them a -18 nicing when I start them up, so the majority of whatever else goes on with the system generally doesn't affect the HLDS processes. Doens't help with the HLG CPU lag, but that's something completely different.. :) Sincerely, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] Ban list data/UDP port 27011
PiTaGoRaS wrote: > The port 27011 is opened in the remote machine (the ban list server) not in > your server. I suppose the packets are sent from your 27015 (or whatever > port is your server running) to 27011 of ban server and viceversa. So > shouldn't be problems with that. Curious about this, I decided to look at a netstat output, and found out that this is not true. HLDS indeed binds a listen process to 27011, as witnessed below : udp65280 0 0.0.0.0:27011 0.0.0.0:* udp0 0 64.69.87.220:27015 0.0.0.0:* Two things I notice about this. This first is it's binding to 0.0.0.0, instead of the IP address I've specified in the server command line. Even if I got around the incoming issue by binding to multiple IPs, as I do for my clan server, the binding to 0.0.0.0 negates this for the ban information. Secondly, why is there a recieve queue on this? IIRC, recieve queue only builds up if the kernel has received traffic on that port, but the program isn't paying attention to the socket. Somebody forget to actually get the data from the socket when they were coding this? :) (Mystery of the not-working banlist solved? ;) Aluve, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] tranfer speed of new maps
Eric (Deacon) wrote: > Honestly it seems like it'd be a lot quicker and easier to simply make > available to the admins a cvar that they can set to determine what kind > of transfer rates they want to allow for downloading resources in-game. Not really, because to avoid file corruption, you'd have to work into the transfer some sort of control system, it's better off just to use some existing TCP protocol like HTTP or FTP. The other advantage that I could see would be map pack downloads. You want to play on this map, you also need to at the same time, download the other 7 maps in this ZIP file designated in the redirect. Maybe change the mapcycle.txt to read : de_bigbang http://www.somehost.com/somedir/mappack1.zip de_bangbig http://www.somehost.com/somedir/mappack1.zip cs_pinkpig http://www.somehost.com/somedir/mappack1.zip Then you only have to interupt your game once and get all the maps that the server is playing. But that's just my opinion, your milage may vary. Aluve, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] Streamline linux, deticated hlds_l server
Rob Sanders wrote: > When I compile the kernel, are there any steps I should take to insure > that its as trim as possible? I've done some searching, but mostly I've > found guides to writing code for the kernel and not so much I'm new and > stupid guides. Will gentoo walk me thru it? I'm not one to assume that > that and go blindly into the night. Is the reason I cant find a guide is > that its that simple? Curse Redhat for being so easy ;) Essentially, when you are compiling a new kernel, a general rule of thumb would be to compile only those things that you absolutely need into the core kernel, and compile everything that you think you might need as a module that can be loaded/unloaded as needed. So go through the kernel configuration (If you want to do it all by hand, I recommend using the 'make menuconfig' option for configuring your kernel) and turn everything off that you don't need, paying attention to look at everything to make sure that the default isn't on. IIRC, some examples of this would be the SYMBIOS SCSI drivers, and the USB drivers. Unless you have a Symbios controller, you probably don't need that enabled, and if it's a server, you probably don't need the USB drivers either (infact, on my servers i even strip out all mouse support, but that's because I'm probably insane). > What services and programs will help me along the way? This machine > will only be serving multiple instances of hlds, nothing else. I expect > ssh, ftp, but what else has shown to be handy in a pinch? Possibly > installing something like iptraf so I can monitor bandwidth? Firewalls > an issue? I can't see into the future, so any personal recomendations > would be helpfull. I want to keep the entire install as trim and speed orientated > as possible. IMHO, ditch FTP unless you have users who need it. If you need to get files onto your machine, use SCP instead, it's more secure, and it's one daemon running instead of two. For firewalling, use IPTables, which also gives you the benefit of being able to setup IP accounting as part of it, removing the need for additional programs to track your bandwidth. For the trimmest possible machine, nothing should be running but the core system stuff, sshd, and hlds_l, and your kernel should be stripped down to the minimal possible configuration and compiled for your CPU architecture. For the best security, everything should be firewalled but your SSH TCP port, and your HLDS_L's UDP ports. One other recommendation I have that is oft overlooked. Use the nice command when running HLDS_L and adjust the servers priority for CPU attention up. I usually run my servers with a nice adjustment of -15, giving it priority for CPU attention over just about every other piece of software running except the kernel. Your milage may vary. :) Sincerely, TrIaX ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
Re: [hlds_linux] Re: [hlds] Testing beta security modules -Saturday
> Eric Smith wrote: > > 3) We're no longer using port 27013 for the ban list. You will now see > > traffic (in/out) on port 27011. This traffic is to/from the ban list > > server. Make sure you have this port open in your firewall. > > TCP or UDP? Can you provide source/target IP(s) so we can open this port > only to the places it's supposed to go? (I'm paranoid like that) Seconded.. :) Sincerely, Warren ___ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux