I use a combination of monit process monitoring, simple icmp checks, and a lot of log checking to monitor most of our extension.org systems infrastructure. It isn’t fancy, and it’s not quite the visibility I want, but it’s good enough for where things are at right now. Most of proper monitoring isn’t about getting data about things as they are, but things that have changed from what they just were prior.
C is for cookie, and icmp’s good enough for me, or something like that.
Anyway, ping checks to the hosted server last night from the NC State monitoring box started failing last night – which is pretty much unheard of in normal operation. And in shelling into the box (a shell into NC State from Road Runner and then over to the hosted server) – showed the box was just fine itself, but the shell session was either dropping packets, or the latency was really, really bad because it was next to unusable. HTTP seemed ok. But I didn’t spend a lot of time browsing to see if it was affected all that much.
I broke out traceroute, not because it’s all that great of a troubleshooting tool, but it at least gives a relative glimpse if the echo times go up somewhere in the middle. From NC State to the hosted server – things looked, well, a little odd (some times left out to fit it in my content div for the blog)
I’ve never actually seen multiple servers on a single line (lines #8 and #9) – but it’s not like I live in traceroute. Combined with line #10 – something looks a little weird at the edge between level3 and above.net. echo times go up 10 fold, and there’s some oddities there.
Things get really weird on the return trip:
So if I’m reading those hostnames right – and they are an accurate reflection of geographic location – it sure ain’t Paris Texas. Traceroute got routed from New York to the UK to France, and back to D.C.
Last night, I thought this was above.net’s problem, today I’m not so sure, I’m thinking it was Level3’s. I wonder if stuff like this actually gets reported anywhere.
Today, things look like you’d expect.