I actually played with an active/passive HLDS cluster last month. It's incredibly easy and worked perfectly the first time. But as mentioned, it does take a hit during a failover event. This is where I stopped, and decided a compute utility or grid scenario would be more appropriate... just haven't had the urge to check it out.
-----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of kama Sent: Friday, September 08, 2006 11:00 AM To: [email protected] Subject: Re: [hlds_linux] HLDS Cluster Isnt it easier to "calculate" if the hlds fits on a server before enabling the service? Or have a predifined sollution and depending on what the theoretical max usage will be, open up on a machine that have enough power? Since hlds is not suited for a cluster sollution, why try to force it to do so? A HA sollution might be interesting just so the service is always up and running. but then you have to take into account that it will be a small downtime when you roll up the service on the other node. /Bjorn On Thu, 7 Sep 2006, Marcel wrote: > Hi, > > I tested this one or two years ago and it caused lagg while > transfering the process to another node. After this "transfer" everything was fine. > > So you should manually do the transfers to another server. > > > - Marcel > > Jonathan schrieb: > > Hi, I've been using a cluster-like setup to start counter-strike > > servers on demand for a while, but it's never really worked that > > well. At the moment i use sun grid engine 6, but i was thinking of > > using something like openmosix instead. Any thoughts on this? If i > > use (open)mosix i need to use the 2.4 kernel, which i heard doesn't perform as well as 2.6 for hlds. > > > > Is there any other alternative available? The key idea is the launch > > a server immediately as someone presses a button on a website, > > without having to care about load on the boxes. Right now i have 22 > > diskless server nodes in the cluster. > > > > OM also seems like a good idea since i will be able to expand it > > while it's running without downtime or problems. > > > > One thing i'm not sure about OM is whether the load-sharing thing is > > good or not. Ie, if a node gets overloaded, it moves processes to a > > node with less load, if one exists. A move like this when the > > process is hlds will probably cause a lot of lag for the players. > > > > > > _______________________________________________ > > To unsubscribe, edit your list preferences, or view the list archives, please visit: > > http://list.valvesoftware.com/mailman/listinfo/hlds_linux > > > > > > _______________________________________________ > To unsubscribe, edit your list preferences, or view the list archives, please visit: > http://list.valvesoftware.com/mailman/listinfo/hlds_linux > _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
