HN Uptime Visualized: Lots of mini-outages lately?

http://hnstatus.net/

by chaosmachine • 16 years ago

20 13 16 years ago

hnstatus.net

chaosmachine• 16 years ago

Almost seems to be doing a hockey stick, if you look at the last 6 months.

fossguy• 16 years ago

what tool do you use to generate these stats?

ronnier• 16 years ago

The graphs look like MRTG. Probably running a request every few minutes and looking at the response, if there is one, and plotting it.

http://oss.oetiker.ch/mrtg/

caller9• 16 years ago

You could combine Nagios, Lilac, and PNP4Nagios to get some pretty good monitoring with graphs(PNP4Nagios), notification(Nagios), and the ability to configure new servers from templates(Lilac). Not for the feint of heart, but not outright tedious. With some slight hackery you can even graph and alert on windows drive freespace.

pmjordan• 16 years ago

It's weird, when HN goes down, it seems to accept the HTTP connection, but closes it again once the request is sent. I'm not sure on the exact timing, but I'm definitely never getting "connection refused" or an actual HTTP error code. No other site behaves like that.

derefr• 16 years ago

I'm guessing there's a proxy or load-balancer involved—the backend goes down, but the proxy stays up, so when you send your request it gets forwarded first, and then fails.

wmf• 16 years ago

A working proxy would send a 500 error. The closed connections are a telltale sign of an amateur Web server; probably written in Arc in this case.

Groxx• 16 years ago

I've been noticing it too, was wondering if it was just me ( downforeveryoneorjustme.com said it was just me, though that could be because it came up in the meantime).

swombat• 16 years ago

I think I heard somewhere that HN is auto-rebooted every half hour or so.

pg2• 16 years ago

No, I manually restart it when it gets slow.

rogermugs• 16 years ago

if you manage to hack HN you're going to lose some karma... but lets be honest... you're also gonna get some mad props.

pg2• 16 years ago

I forget exactly why, but attempts to connect get ignored when MzScheme is in the middle of GCing. I'm going to start tossing old stuff back out of memory, which should mitigate the problem, but I've been too busy with YC lately to work on HN.

HN Uptime Visualized: Lots of mini-outages lately?

Related Stories