Anyone else getting slow/sporadic service? September 27, 2015 8:32 AM   Subscribe

For the last 2-3 days I've been having pages that are reallllly slow to load, and sometimes partially load and then hang. Is it just me?

I think it started on Friday. It doesn't happen every time, maybe one out of every three pages. Initially I noticed it on the "Recent Activity" page and the Boehner resignation page, but it's been happening on the front page too.

I did the log-out, clear cookies, log-in routine, enabled all Javascript hosts I could (in NoScript), and disabled all MeFi GreaseMonkey scripts. I still had the same problem.

I'm on Windows 7, FireFox. I use the old traditional theme, but I've had the problem while logged-out and getting the Modern theme.

I snapped an image of the problem happening under FireBug. The thread was only partially loaded after 2m 51s and the slow request was coming from metafilter.com.

(Posting in Meta as opposed to just contacting mods as I figure it's worth seeing how wide-spread the problem is.)
posted by benito.strauss to Uptime at 8:32 AM (73 comments total)

Yep, same here. I was going to make a MetaTalk post about it but you beat me to it. However now I'm wondering if there is a geographical correlation...
posted by overeducated_alligator at 8:37 AM on September 27, 2015


I don't know of any major issues we've been having server side, but obviously there could be all kinds of downstream stuff, yeah. We haven't been getting any reports about it in the contact form so I'm guessing it's not a widespread issue.
posted by cortex (staff) at 8:38 AM on September 27, 2015


This has happened to me a few times this weekend. Each time, it cleared up after waiting a minute or two and trying again.
posted by double block and bleed at 8:41 AM on September 27, 2015


OSX Yosemite, current version of chrome, new dark theme, several extensions including five mefi tampermonkey scripts. I didn't go through my full troubleshooting routine since it's been so sporadic. I've opted in for ssl.
posted by double block and bleed at 8:46 AM on September 27, 2015


Hmmm, overeducated_alligator and I are both near Boston. Maybe it is geographical.
posted by benito.strauss at 8:47 AM on September 27, 2015


Maybe not. I'm in Indianapolis.
posted by double block and bleed at 8:50 AM on September 27, 2015


It's been happening to me occasionally when I try to access the site at work (shhh, don't tell!), but I've been assuming it's our internet access that's the problem, because it's very very slow even on good days.
posted by jaguar at 8:56 AM on September 27, 2015


Here's my timeline with an acceptable page load. I'll try to record another one if it happens again and I can catch it.
posted by double block and bleed at 8:56 AM on September 27, 2015


but I've been assuming it's our internet access that's the problem, because it's very very slow even on good days.

By that, I meant "the entire internet has been very very slow," not just MetaFilter.
posted by jaguar at 8:57 AM on September 27, 2015


I would normally follow this page and try to see what's up and fix it, but I've been unable to get to my profile page (to check if I've got that SSL thing enabled) for the last 10 minutes. So I'm going to wait to see if it's still a problem on Monday when pb will be back, or maybe there's some Boston-area weekend Internet routng thing that might be cleaned up indedpendently.
posted by benito.strauss at 8:57 AM on September 27, 2015


I am just outside Boston, and I've been having this problem on my phone, on three different laptops, and on my iPad.
posted by Elementary Penguin at 9:02 AM on September 27, 2015


Benito: you can check if ssl is enabled by looking at your address bar. It will start with https: instead of http: (maybe highlighted in green, depending on your browser.)

I don't know why your profile isn't loading. Mine does.
posted by double block and bleed at 9:02 AM on September 27, 2015


This might be one of those things where pb says "I had our provider reboot our app/db server, should be all fixed now." or somesuch. It's often impossible to debug beyond educated guesses from the client side.
posted by double block and bleed at 9:06 AM on September 27, 2015


I'm just outside of Boston and haven't had any issues...
posted by NotMyselfRightNow at 9:07 AM on September 27, 2015


I'm on Windows 7, FireFox. I use the old traditional theme

Me too and it's all been running perfectly for me. I'm in Germany, except I've been using a VPN based in Long Beach most of this weekend.
posted by shelleycat at 9:14 AM on September 27, 2015


I'm on Linux and in the Greater Boston area and this has not been happening for me.
posted by XMLicious at 9:18 AM on September 27, 2015


According to our monitors on this end everything is running at full speed. There doesn't seem to be any obvious reason for slowness from our end of things so I'm guessing this is a problem between the client and server somewhere. Sorry about that—I wish there was an easy fix we could make for it.
posted by pb (staff) at 9:28 AM on September 27, 2015


I've been noticing a general internet slowness this weekend, too. It seems worst when I'm directing traffic through the west coast (but I might be imagining that.) My Amazon AWS based near Portland was almost unusable for a while Friday night/Saturday morning. So, anecdotal, and maybe an imagined connection, but isn't MeFi in that area?
posted by ctmf at 9:42 AM on September 27, 2015


Yep, we're on AWS. Sounds like MeFi is in the same place. There aren't any notices of AWS problems for that location.
posted by pb (staff) at 9:45 AM on September 27, 2015


I've been noticing mefi slowness too.

It might be chartbeat. As a random test, I just set ping.chartbeat.net to poing to 127.0.0.1 and I haven't had a problem since, but that might just be a sampling issue.
posted by rmd1023 at 9:52 AM on September 27, 2015


In the timeline snapshot benito.strauss shared it's definitely the MetaFilter server that he's having trouble connecting with. So I'm not saying chartbeat isn't causing problems for anyone, just that in benito.strauss' case that wasn't the issue.
posted by pb (staff) at 9:53 AM on September 27, 2015


Also, chartbeat loads after the bulk of the page content. So a slow connection to chartbeat might make the page load a little longer, but it wouldn't block any content from loading. So I don't think that's going to be a primary reason for perceived slowness.
posted by pb (staff) at 9:55 AM on September 27, 2015


China is in town and the Vatican. Maybe it is more general overload and super surveillance. Maybe it is just showing Mefi the threadbare spots as a side effect. But then, what do I know?
posted by Oyéah at 9:58 AM on September 27, 2015 [2 favorites]


Ah, okay. I hadn't followed that link to the image.

I'm next to Boston and my path from rcn.net over to mefi looks to be going via NTT, and I'm seeing a bit of packet loss there, but latency is pretty good so I don't know if it's indicative of a problem or just the network gear being busy. (A lot of modern network gear will let things like ping packets drop while they're busy doing more important things like moving important traffic.) Nothing big showing up on NANOG (North American Network Operations Group), though, and if there were something substantial happening I'd expect to see discussion of it.
posted by rmd1023 at 10:02 AM on September 27, 2015


Aaaaand definitely still seeing it with charbeat out of the mix.
posted by rmd1023 at 10:04 AM on September 27, 2015


I was just about to ping the mods about this. MeFi has been very very slow for me the last day or two. I'm in Brookline viewing the site on my iPhone with Peace installed. Other sites load fine.
posted by alms at 10:24 AM on September 27, 2015


Another one here who's been having problems. I'm in NYC, using an iPad.
posted by holborne at 10:41 AM on September 27, 2015


Now I'm getting page load times between 1.5 to 3.5 seconds, so ¯\_(ツ)_/¯
posted by double block and bleed at 10:44 AM on September 27, 2015


Here in Maine I saw some strange delays yesterday evening, but everything is fine now. It was definitely exclusive to Metafilter.
posted by selfnoise at 11:05 AM on September 27, 2015


When pages do load for me they load at their regular quick speed (1-2 seconds), but when they hang they hang for a long time. I can have one tab stuck in the middle of loading one page, and another tab can load the same page in the usual 1-2 second.

Thanks for checking things on your end, pb. I had noticed some general Internet slowness at the same time, but those have only been "5 seconds instead of 1 second" type of thing, while MeFi gets into minutes. But the fact that the timeline I grabbed had it cutting off smack in the middle of a comment points to something happening at the packet level and not the database or application level.

I guess we'll have to wait for the network gods to intervene. If it's of any value, I'm also on RCN, though I don't know how to see what servers I go through to reach MetaFilter. ('traceroute' doesn't really work with the modern network, does it?)
posted by benito.strauss at 11:06 AM on September 27, 2015


FWIW I am also using RCN.
posted by Elementary Penguin at 11:45 AM on September 27, 2015


I've also had this problem the last few days, though not this morning. Seattle, iPhone.
posted by thetortoise at 11:54 AM on September 27, 2015


RCN here too. This obviously warrants a sputtering, incoherent phone message left on their machines.

RCN PLEASE HOPE ME
posted by overeducated_alligator at 12:03 PM on September 27, 2015 [2 favorites]


There's a jillion other variables, but…RCN here, too! I've been getting this once out of every four or so page loads.
posted by ignignokt at 12:13 PM on September 27, 2015 [1 favorite]


Having this issue in California, too, both on my iMac and on my iPhone.
posted by Hermione Granger at 1:01 PM on September 27, 2015


I've also had some slow/stuck page loads, but I had figured it was a problem with the slowish hotel internet I am using right now.
posted by Dip Flash at 1:15 PM on September 27, 2015


Just chiming in as another Boston-area RCN customer getting intermittent slow or hanging loads for the last day or so.
posted by firechicago at 1:25 PM on September 27, 2015


Well, if XMLicious isn't on RCN we've got something here. Isn't this the time when people who Twitter (i.e. not me) start tweeting with an @RCNService hashtag and linking to this thread? I've heard that's how customer service works these days.
posted by benito.strauss at 1:49 PM on September 27, 2015


Me RCN also and slow.

Would a trace route be helpful?
posted by alms at 2:12 PM on September 27, 2015


My ISP is Comcast and traceroute metafilter.com only returns nodes ending in comcast.net in the reverse lookup, nothing with rcn.net.

So evidently Comcast doesn't have a problem. I expect this means that hell has frozen over.
posted by XMLicious at 2:14 PM on September 27, 2015 [2 favorites]


I too am on RCN (in Chicago) and have been having trouble all weekend.
posted by colbeagle at 2:23 PM on September 27, 2015 [2 favorites]


OSX Mavericks, Chrome, new theme, no chartbeat, AT&T in TX. I've seen the MeFi front page fail to be found altogether maybe 3 times in the past few days. I wait a minute & retry, it loads fine. Never happened before, no other sites acting this way.
posted by scalefree at 3:09 PM on September 27, 2015


Yep, me too. On RCN in NYC. Mefi pages are the only ones I'm having trouble with.
posted by gaspode at 3:24 PM on September 27, 2015


Good sleuthing so far! Has any RCN customer tried contacting support?
posted by pb (staff) at 3:40 PM on September 27, 2015


Was happening to me last night in Virginia on fios on my iPad. I blamed iOS 9.
posted by 4ster at 4:27 PM on September 27, 2015


No idea if this is any help, but here's a traceroute that seems to be exhibiting the behavior I get roughly half the time when I try to load a metafilter page.
posted by firechicago at 5:02 PM on September 27, 2015


Also, a datapoint: when I posted that last comment I got the hanging behavior, but when I finally did manage to reload the page, the comment had already successfully posted, so it looks like the issue is on the return trip.
posted by firechicago at 5:05 PM on September 27, 2015


Chicago here. RCN has been super spotty all weekend. Site says nothing is amiss, but I beg to differ. I've never had this much lag before, and it's been doing this since Saturday early morning at least. Maybe it's the apocalypse moon.
posted by heyho at 5:36 PM on September 27, 2015


Ooops, I meant I talked to RCN, and they said nothing is amiss in my area.
posted by heyho at 5:44 PM on September 27, 2015


Just chiming in to say that I'm in the Boston area, but I use Comcast, and I haven't had any issues with page loads.

I'm pretty sure this is the first time I've ever been grateful to Comcast for anything, ever.
posted by litera scripta manet at 6:10 PM on September 27, 2015


The problem goes away if I access the Internet by tethering to my Verizon iPhone. It comes back if I switch back to my home RCN connection. Using SSL makes no difference.
posted by alms at 6:21 PM on September 27, 2015


I followed up with RCN via twitter - I pointed out that it's not an RCN problem, but it's causing problems for people coming from RCN. It's a tough situation to be in, since there's nothing they can do to fix it (unless they take drastic steps like increasing the routing cost for traffic transiting NTT so it takes another path, but that can be complicated), but I have no idea if they've rattled NTT's cage on it.

I don't know if pb or other modly types complaining to the AWS folks would do any good.
posted by rmd1023 at 8:34 PM on September 27, 2015 [2 favorites]


Loading fast here in Seoul :)
posted by Joseph Gurl at 8:49 PM on September 27, 2015 [2 favorites]


Thanks, rmd1023. Looking at you profile it seems you're likely to understand this better than most anyone else here. So is NTT a top level backbone between RCN and the MeFI AWS servers? Would we expect to see similar problems with other things hosted at the same AWS location?
posted by benito.strauss at 10:27 PM on September 27, 2015


How much crap can be hung off a text-only web page? Metafilter aims to find out.
posted by telstar at 3:53 AM on September 28, 2015 [1 favorite]


I have no idea what any of the technobabble in this thread means, but I can add a data point that I am on RCN in Camberville and Metafilter has been loading slowly for me. I'm at home on the work VPN right now and it's fine going through that (I have no idea about the specifics of how the VPN works).

My wife has not complained about slowness, but she also keeps about forty million tabs open on all of her devices so they all just run slowly in general.
posted by backseatpilot at 8:00 AM on September 28, 2015


Was happening to me this weekend to. As mentioned by others, some pages load fine, some got stuck loading, but a reload seemed to make it go away all but once. IOS, but on my wireless connection which is through Time Warner Cable. Milwaukee, wi.
posted by [insert clever name here] at 8:12 AM on September 28, 2015


benito.strauss: Yeah, NTT is a big network provider. It looks like RCN doesn't have a direct connection to AWS so they send their traffic via NTT. I'd expect to see similar problems for other things in the same area of AWS if my theory is correct.
posted by rmd1023 at 8:33 AM on September 28, 2015 [1 favorite]


I'm on RCN in the Boston area and I've had a few times over the past few days when Metafilter was slow or unable to load. Usually a reload clears it right up but not always.
posted by bondcliff at 8:35 AM on September 28, 2015


In the Boston area on work Internet (which does not go through RCN): things are fine.

(The whole "bad connection to AWS => bad Internet" situation reminds me of how fragile things are. The cloud is just other computers, man.)
posted by ignignokt at 8:50 AM on September 28, 2015 [1 favorite]


FWIW, I'm on Comcast and things have been pretty slow across the board for me today. I'm seeing a lot of "Connecting" and "Looking up" pauses when hitting all websites, not just MeFi.
posted by Thorzdad at 8:51 AM on September 28, 2015


meanwhile here in chile the sun is out, spring is here, amazon just delivered, and pages load like greased lightning :P
posted by andrewcooke at 10:25 AM on September 28, 2015


It's been a little better for me today, so I'm going to hold off on trying to phone RCN. I think it's more likely that it will clear up through some process that's already underway (like some routing tables getting updated somewhere) than me trying to get direct intervention via the phone help line.

Man, the Internet is like the weather, isn't it? Big and complicated with so many moving parts that it's hard to see what's causing what, and one teeny piece of it has a huge effect on your life. I wonder how far the analogy can be pushed? Are there Internet weather maps out there? I'm betting NOCs have something like that. Are any publicly accessible?
posted by benito.strauss at 10:32 AM on September 28, 2015 [1 favorite]


The Farmer's Almanac says the internet is gonna be wicked fast next year.
posted by bondcliff at 10:35 AM on September 28, 2015 [11 favorites]


Well, 700 comment threads are now loading in three seconds. Back to nice Internet weather and I guess that means the apple crop is going to be good. I wonder what happened, and thanks to all of you who poked and prodded your parts of the web.

"The Great 2015 Blood Moon Eclipse RCN Internet Slow-Down" t-shirts will be available in my Etsy shop in a week or two.
posted by benito.strauss at 6:05 PM on September 28, 2015 [3 favorites]


Hmm. RCN customer in Chicago here. I had tons of bad internet problems all this past weekend. Some Metafilter, yes, but more commonly Tumblr and anything with HTML5 video. I ran tests on Speedtest but it always said I was getting my full 25mbps as I should be. I thought maybe it was my aging iMac, which I only just upgraded to Yosemite a week ago. This thread makes me think it really was RCN after all. (Things seemed better this morning for the few minutes I was online at home.)
posted by dnash at 10:09 AM on September 29, 2015


Seems like this is mostly resolved, but I thought I'd chime in with a similar ISP-based issue, along with a fix for it that might be helpful in this case or similar ones. I'm on CenturyLink and was having trouble for weeks with any data coming from Apple - App Store, iTunes Store, System Updates, etc. CenturyLink denied up and down that there was any problem on their end for about a week - denied that it was even possible for this to be their problem - until I finally got in touch with a decent support person who said it was a known issue* and that changing the DNS server from the default CenturyLink servers seemed to help, which has worked for me. Not sure if that would be helpful in this case, but just thought I'd mention it.

*Oh, my friends, the validation I felt, you have no idea.
posted by Rock Steady at 6:33 AM on September 30, 2015 [2 favorites]


I've lost trust in the ability of most ISPs to maintain their own Domain Name Service (DNS) servers . Slow DNS performance is the most likely cause of "Looking up..." pauses, and has the effect of making it seem like your intertubes have been filled with molasses.

I've switched many friends and family over to using Google's DNS. It's usually been a positive experience, and never a negative one.

Follow that link that Rock Steady posted above for instructions. It's fairly quick and easy, and is easily reversible if it doesn't work for you.
posted by dws at 10:33 AM on October 1, 2015 [1 favorite]


(on the other hand, that gives google even more information about what you do on the internet).
posted by andrewcooke at 11:38 AM on October 1, 2015


If the first part of a page has been delivered to your browser then the DNS look-up has already succeeded, so I don't think this particular event was DNS related. Nice to have the instructions though.
posted by benito.strauss at 11:56 AM on October 1, 2015


Depending on what you mean by "partically loaded," the FireBug screenshot isn't inconsistent with a DNS lookup taking most of that huge block of time. Since the remaining requests are to the same domain, they'll use the cached DNS lookup result. Next time, expand that initial slow GET (by clicking on the left [+]).
posted by dws at 1:50 PM on October 1, 2015


Noticing improvements in loading latency over the last week.
posted by telstar at 11:55 AM on October 5, 2015


UK here. I'm a bit late to the party but the issue started for me a couple weeks before this thread. It has been intermittent for me, today it's been on again and I finally had time to take a screenshot. It doesn't last long but when one is underway I get perhaps one out of 5 pages to fully load, half the time pages will stop loading in the middle. I've started to copy and paste links from the frontpage in the address bar when the page doesn't load and then it often works (it's really more that one out of several requests eventually works).
posted by yoHighness at 1:15 PM on October 21, 2015


« Older Mildly Irritating Mefite Makes Good   |   A kind word doesn't cost a thing. Newer »

You are not logged in, either login or create an account to post comments