Server Move Updates — Interesting Occurance
Howdy All,
My server has hit 2.5 days of uptime (I've had it for almost 2 weeks, and that's the most uptime it's had yet). I finally (after fighting with tech support) was able to get them to swap the hardware and I haven't had a segfault yet, even though they still insist it wasn't a hardware problem.
I was rather unimpresed with the quality of support I received (although it may finally be resolved), as I've always had great support from 1and1. This was my first time dealing with 'Dedicated Server' support and I wonder if it's a different group, because it was quite the different experience.
Anyways... not that it's up, I wanted to share something interesting that I'm seeing. I purchased a couple of extra IPs as some of my domains will use SSL and I've set this up using interface aliases on the server.
eth0 is the primary IP, and eth0:0 and eth0:1 are the secondary IPs. Here's where this gets to be interesting. eth0:0 is always used. If I specify an interface... it doesn't really matter because quite often that will be ignored.
Examples:
bind: listen-on was set to eth0, however my AXFR was occuring with eth0:0, I had to specify a transfer-source to solve the problem.
Wolfenstein Enemy Territory: I've specified the listen port as eth0:1, yet the only IP that you can connect to is the one assigned to eth0:0.
irssi: Regardless of the IP I specify with the -n flag, I still end up using eth0:0.
So... has anyone seen this before? I sure haven't and I can't find anything online. I'm wondering if there's a flag/setting somewhere that sets eth0:0 to be the primary interface... but I can't find it.