MetaFilter DNS records goofy?

July 16, 2008 9:55 PM

Are the metafilter.com DNS records configured properly? I'm having to do a DNS lookup of all mefi sites like once a minute, as if the expiration counters in the records are set way off. Even if you were planning a server move, that's way too tight. Seconds vs minutes error?
posted by intermod to Uptime at 9:55 PM (35 comments total)

Could be. I basically have very little knowledge of how DNS server TTL stuff works, I think I just copied settings from an earlier DNS server.

If anyone can do a dig on mefi and tell me what I should change them to, I'd be happy to change the settings.
posted by mathowie (staff) at 9:57 PM on July 16, 2008

It all looks reasonable to me: expiration of 1 day. The nameservers at easydns.com have 1-hour expiration for their own names, which seems a little short to me in the absence of an anticipated move, but shouldn't cause a real problem.
posted by hattifattener at 10:19 PM on July 16, 2008

chucknorris.com's dns records don't expire. They lurk.
posted by blue_beetle at 11:48 PM on July 16, 2008 [2 favorites]

metafilter.com. 86400 IN A 74.53.68.130

That really isn't a problem. It could be doubled or tripled, but it isn't causing trouble.
posted by cmonkey at 3:36 AM on July 17, 2008

Your ISP might be caching DNS records and the TTLs. Though it's not clear to me why they'd make them lower...
posted by Skorgu at 6:04 AM on July 17, 2008

Could be tied to the current DNS problems across the Internet.

Turns out there's a pretty massive exploit which enables poisioning of the DNS caches, those ISP's which are fixing the problem might be lowering the TTL (essentially the trust threshold) to refresh potentially corrupt entries as they patch all their servers to correct for the problem.

Shot in the dark, but sounds like it could fit...
posted by Static Vagabond at 7:07 AM on July 17, 2008

I think the MetaFilter DNA helix may be wound a little tight, too.
posted by It's Raining Florence Henderson at 7:48 AM on July 17, 2008

I think it's the Fetzer Valve. We're gonna need some ball bearings.
posted by spicynuts at 8:33 AM on July 17, 2008 [2 favorites]

I was probing to determine skeletal girth and muscle tone. It's a new technique. We mock what we don't understand.
posted by horsemuth at 8:40 AM on July 17, 2008 [1 favorite]

You could (possibly) harness B2C deployments and leverage indexed action-items.
posted by tellurian at 8:44 AM on July 17, 2008

IM IN YER DNS HOSING YER TTLS
posted by quonsar at 8:44 AM on July 17, 2008 [1 favorite]

spicynuts, I am shocked that you don't know the difference in DNS timeouts affected by Fetzer valves versus polyhootenators. Clearly a polyhootenator problem here.
posted by beelzbubba at 8:45 AM on July 17, 2008

Did anyone check the flux capacitor?
posted by inigo2 at 9:00 AM on July 17, 2008 [1 favorite]

This wouldn't be a problem if you got a hardcopy version delivered to your door like every one else does.
posted by srboisvert at 9:09 AM on July 17, 2008 [1 favorite]

Clearly a polyhootenator problem here.

Ahmm...no. Polyhootenators are web 1.0. They are obsolete and Web2.0 is not backwards compatible with Polyhootenators. Please don't jump into conversations you don't understand.
posted by spicynuts at 9:14 AM on July 17, 2008

I want monohootenators, myself. Polyhootenators lead to increases in cholesterol levels.
posted by Class Goat at 9:55 AM on July 17, 2008

That was blown way out of proportion. It's the Transhootenators that'll kill you.
posted by SpiffyRob at 10:08 AM on July 17, 2008

They're also more than meets the eye, I hear.
posted by SpiffyRob at 10:08 AM on July 17, 2008

Oh! Oh! I see the problem. The R2 Unit has a bad motivator.
posted by The Man from Lardfork at 10:22 AM on July 17, 2008 [2 favorites]

This is all over my head. If anybody wants me I'll be over here, distimming the framingroshes.
posted by Lentrohamsanin at 10:33 AM on July 17, 2008 [2 favorites]

I've got a spare Johnson rod here, if you need it.
posted by Blazecock Pileon at 10:39 AM on July 17, 2008

I'd piss on the damn sparkplugs if I thought it'd do any good.
posted by OmieWise at 10:45 AM on July 17, 2008

have we tried reversing the polarity?
posted by heeeraldo at 10:47 AM on July 17, 2008

Hey, heeeraldo, don't start in with jokes about poles, OK?
posted by beelzbubba at 11:06 AM on July 17, 2008

Did you ensure that your computer is plugged in?
posted by spiderskull at 11:09 AM on July 17, 2008

Also, check to ensure that your proboscis is entwatenated.
posted by spicynuts at 11:12 AM on July 17, 2008 [1 favorite]

Fnord.
posted by Godbert at 12:14 PM on July 17, 2008

Please fix the DNS records so that there is no more goofiness around here!
posted by aubilenon at 2:37 PM on July 17, 2008

This answer would be right at home in this thread.
posted by generichuman at 3:18 PM on July 17, 2008

it's been far too long since my proboscis was entwatenated. sigh.
posted by quonsar at 4:01 PM on July 17, 2008

This answer (Slashdot) was just about the most awesome thing I read all day today.

Thanks all for the answers, well, the early ones. I don't think my ISP would be driving TTLs down to 60 seconds for all users due to cache poisoning concerns. More likely I've got a caching problem, although that seems equally unlikely on my shiny Ubuntu Linux installation. Thanks again.
posted by intermod at 7:36 PM on July 17, 2008

i assume you're on dhcp and using whatever nameserver is being fed to you. you could try changing the config to use a fixed server - perhaps opendns. that would solve things if the problem is a badly configured nameserver provided by your isp.

it's also pretty easy to run bind as a local cache fi you feel like playing with that.
posted by not sure this is a good idea at 7:45 AM on July 18, 2008

Have you tried bending the ballcock?
posted by Sys Rq at 10:15 AM on July 18, 2008

One thing I've noticed is that every pageload of a metafilter page also wants to connect to styles.metafilter.com, whcih I'd have expected the browser to cache. Maybe it's just reading the resource's date to make sure it's not stale, but it takes longer than I'd have expected.
posted by orthogonality at 6:13 PM on July 18, 2008

Does this seem anyway connected to my net problem?
posted by Gyan at 10:49 PM on July 18, 2008