But why doesn’t it ever empty the swap space? I’ve been using vm.swappiness=10 and I’ve tried vm.vfs_cache_pressure at 100 and 50. Checking ps I’m not seeing any services that would be idling in the background, so I’m not sure why the system thought it needed to put anything in swap. (And FWIW, I run two servers with identical services that I load balance to, but the other machine has barely used any swap space – which adds to my confusion about the differences).

Why would I want to reduce the amount of memory in the server? Isn’t all that cache memory being used to help things run smoother and reduce drive I/O?

And how does cache space figure in to this? I have a server with 64GB of RAM, of which 46GB is being used by system cache, but I only have 450MB of free memory and 140MB of free swap. The only ‘volatile’ service I have running is slapd which can run in bursts of activity, otherwise the only thing of consequence running is webmin and some VMs which collectively can use up to 24GB (though they actually use about half that) but there’s no reason those should hit swap space. I just don’t get why the swap space is being run dry here.

So ldirectord is kind of a front-end for ipvsadm. The tools allow you to set up load-balancing between internal servers. I run each service in a VM, and I have at least two copies of each (on separate physical servers). Ldirectord lets me configure how frequently to verify each machine is up, a list of primary servers, and an optional backup when the others go down. Overall it works pretty smooth.

Shorewall is similarly a front end for iptables, allowing a more structured set of configuration files. I’ve been trying to start using Webmin for the first time because it has some nice management of shorewall, maybe I’ll be able to clean up some of my config, but I’d also like to get traffic shaping configured.

I have a dedicated firewall (just moved to a poweredge R620 last night), a NAS, and two VM systems to run services on… all run from home. I enjoy setting things up to play with, so this has all been built up starting from old desktop machines and expanded over time.

I tried playing around with opensense awhile back. Wasn’t impressed and kept running into things I couldn’t get it to do for me, so I stuck with my existing setup. I use ldirectord for load balancing between servers and shorewall lets me generally balance the traffic between WAN connections. It works pretty well but there’s a lot of moving parts.

Does anyone else use a Linux firewall to manage dual connections? I run Shorewall here, but I haven’t really had much luck with traffic shaping to keep the majority of traffic on my primary connection while allowing low-speed info like email to split up between connections.

If your card has an x4 pinout, then it probably needs the additional bandwidth. Plugging it into an x1 slot (if it was possible) would slow down the network traffic. Get a better motherboard with an x4 slot on it so you can use the hardware you want. or find something else that will fit your computer.

Honestly even the 1Gb quad port card I have requires an x4 slot, although I saw some dual-port 2.5Gb x1 cards on ebay. Maybe you could just use two of those?