1 I've had that problem multiple times with Comcast, where the only thing that actually went out was the internal DNS servers, which they seem to have a strangely hard time keeping running.

It's worth your while to try to find other DNS servers you can use. (Consider a local university.) Then, when Comcast DNS is down but your IP connection is still up, you can really see what a cable connection can do, since the traffic on the segment drops to nearly nil...

1 Hmm.  I need to change the comment form so that it automatically comes up with a "You must register to comment." message if the user isn't already logged in.

Let's make sure that the feature at least works...

2 Okay, it does work (I was pretty sure I'd tested it previously) and it will block the spammers, but I need to clean up the messages a bit.

3 I'm not sure I've logged in from this computer. Still, I think it's a good move on your part, Steven.

Update: Wasn't logged in, got "%s have been disabled." I'm sure Pixy will straighten that out!

4 Yep.  I've fixed a couple of messages, and I'm making a change to the comment form so that it shows a message instead of the edit box if you can't comment.

5 The registration process never sent the confirmation email it claimed was necessary, but it worked, so I can't complain.

-j

6 Yeah, I turned off the confirmation step because too many messages got lost in spam filters or whatever.

7 Just testing to see if this works.  And frankly, I'm amazed the spammers took that long. . .

8 testing.....

9 Methinks sending you to a page to log in might be more seamless.

It looks like it will let me put in my real blog URL ... good.

10

Methinks sending you to a page to log in might be more seamless.

Or provide a login form in place of the comment form, and then bring up the comment form once you log in.  Yup.

11 Oh well, I was looking for somewhere to store a little data anyhow. Maybe I'll even post something, someday.

12 Commenting to verify registration.

13 Let's see here, will it let me change my name from the registered name? And can I post my real blog? It looks like I can. Nice

14

Pixy: THAT's interesting.

I took a look in the administration comments page, and Mark's post is listed as coming from "Mark". Fact is, I don't have the slightest idea what login name he used, and it doesn't appear there's any way for me to find out.

I certainly don't want to do so for Mark, but supposing someone was annoying and I wanted to ban them. If they've changed their posting name, then how am I supposed to determine their login name in order to add them to the ban list?

15

Pixy: In the administration "Comments" page, the posting name column should change somewhat. Right now you've got it so that they link to the same things they link to here.

That should change to this: if a user is not logged in, the name should not be a link. If the user is logged in, the name should be a link to the mee.nu info page for the login that was used. So for the entry above, "Mark" links to Kaedrin in this thread, but in the administration "Comments" page, "Mark" should link to the real login ID he used.

16 You do need to be able to see the link even if they're not logged in, because just the link might be spam.  But yeah, to manage users you kind of need to be able to identify the user...

I'll do some work on this over the next couple of days, but it will probably take a few iterations to reach something that has all the information needed but is still manageable.

17 If they've linked to spam, I can see that in the comment thread and the sidebar. I don't need to see it in the administration frame.

18 Definitely agree with comment 10.

19 Steven - yeah... ish.  It's best if you can see all the information needed for moderating comments in the moderation screen itself.  But I won't hold up improvements while trying for perfection; I'll have an update soon.

Jim - I was just thinking about that some more, and the problem there is that the templates are completely user-controlled, and that opens things up for phishing scams.  So it's best if users are redirected to mee.nu itself for the login, then redirected back to the comment page.

20 Testing...

21 Looks like you finally got it to work.

1 You should put height and width attributes on the img tag for the top picture. Otherwise, the page might not render until the connection to the offline server times out.

2 The size of the top div is already hard-wired.

3 I'm in the same area as you (in my case, Vancouver...) and my wife thought I was crazy to get an AC unit a few years back.  She has appreciated it more and more as she gets older, though...

I grew up in AZ, so I know the value of a good AC unit.  She's grown up around here, and wasn't exposed to home AC that much until she met me.  Now, she doesn't want to live without it.

You might want to think about getting one this fall (when they start clearancing them out).

4

Then again, I might not want to.

Here in Portland it doesn't get beastly hot all that often. Last summer there were only two days when I boiled. Probably there will be more this year, but I doubt it will be more than five or six. It isn't worth buying air conditioning for that -- especially since I live in an apartment.

Waking up this morning, in that marvelous lucid moment between dream and reality, a vision came to me. I saw a bag of ice in my kitchen sink, with a fan blowing air over it. So I just went out and bought a fan and a couple of bags of ice. In an hour or two, when it gets worse, I'm going to give it a try. $35 for the fan, and $2.50 for the ice -- and if it works, buying more ice is cheap.

It has to work at least somewhat; the question is whether it will work well enough to be worthwhile. I think it will, but I'll let you know.

5

Engineers say that, "A stupid idea that works isn't stupid." This one is working a lot better than I expected it to. After just 15 minutes the temperature in here is noticeably lower than before I turned on the fan.

Unfortunately, I don't have a room thermometer so I can't provide a quantitative measurement, but it certain feels a lot better, and not because the air is moving. (It isn't, where I'm sitting, a long way away from the fan and 90 degrees off its direction.)

6

It's the height of the heat of the day out right now, and my ersatz air conditioning is working so well that I've stopped sweating.

Maybe a third of the ice has melted by now. The rest of it will certainly last long enough to get me through the rest of the day.

7 Ah, apartment - that makes a difference...

Glad to hear the ice trick worked - when I was in an apartment, I personally would sit with a sprayer (like on the cleaning bottles), and about every 10 minutes or so would spray water in front of the fan.  Your way is much more efficient.

1 On it.  Brickmuppet mentioned something about a broken template, but didn't specify which one.  I'll see if I can work it out.

Your bottom-bar is a new one; that shouldn't be possible the way the CSS is set up.  Some days - most days, lately - I think we should just shoot CSS in the head and go back to tables.

2 Uh, weird.  I was able to check Brickmuppet's site and leave a comment just fine.

3 Not sure what's going on, but if you don't mind too much I'm going to post another... six comments and see if it goes away.  Feel free to delete them later.

4 If it does go away, that points the finger at the display of that last comment of yours in the sidebar, though I can't see how that could be happening.  If not, it's something in the main section.

5 Two...

6 Three...

7 Four...

8 Five...

9 And six.

10 Darn.

11 Another bad thing: one of my spoilers wasn't being censored on the sidebar when I just looked at it. How bizarre.

12 I think there's an extra </div> being generated somehow, and that's bouncing the sidebar into the outer darkness so that it doesn't float to the right any more.  Haven't found it yet, though.

I have to disappear for a couple of hours, but I'll be back and get it fixed for you.

13 Oh, and I have confirmed that Brickmuppet's blog is fine in Firefox and broken in IE.  There are tags in Minx that will let you do that, but I don't think I've documented them anywhere, and I don't know why Brickmuppet would use them anyway.

I'm going to grab a copy of both your blogs to the test environment as soon as I get back and get things sorted out.

14

I wonder if this is a result of the particularly large number of spoiler tags in use right now. It happened after I added one.

I took a look at this page using FireFox. It's wrong there, too, but not the same. In IE7 the sidebar lands below all the normal stuff. In FireFox it lays on top of the bottom two entries.

Speaking of spoiler tags, there's also a problem with IE with a comment thread that has a spoiler in the last comment and where the comment entry box goes. I think I've mentioned it before, but it's still a problem.

When the spoiler gets opened, the comment entry box doesn't move, so text lays under it. If I click inside the comment entry box, it instantly blinks down to where it ought to be, but the "post" button doesn't move, so that's in the wrong place. The only way to really use it properly is to close the spoiler.

Anyway, you're on the case right now and I'll leave you to it. If you need to delete comments as an experiment, go ahead and do so.

15 Pixy, do you have a snapshot of this page that you can use for debugging? I want to make a couple of new posts, but I don't want to automagically "fix" what happened and make it difficult for you to trace.

16 My guess is that you have a post (or comment) which had an extraneous opening (but un-closed) div (I'm assuming that you're using divs for layout). It looks like your sidebar is part of your main div now, which is a normal result of an unclosed div.

17

There are two DIVs with id=main, and it looks like there are two extra close tags for DIVs as well. The spoiler tag also creates a DIV inside of a P sometimes, which will kick most browsers into quirks mode, if they're not there already.

Yes, the W3C validator pitches a fit on this page.

-j

18 This problem is definitely related to an unclosed div tag somewhere.  I haven't figured out where, but adding a div at the end of the main content (but before the #side div) fixes the problem.

19 Steven, go ahead and post whatever you want - I have my snapshot.

20 And thanks guys for the div-hunting. 

I've since fixed the duplicate main thingy, but Steven is running on a modified copy of the bugy version of the template.  I'll try fixing it.

21 The problem with the W3C validator is that it pitches a fit about everything.  I eventually gave up trying to make it all validate, but that now means that I can't use that tool to find problems like this.

22 The problem seems to be somewhere around the "Shingu 20: The summary" - Perhaps Steven was right about all the spoiler tags, as there are a bunch in that post and the comments...

23

It was my last spoiler post in that thread which made the formatting blow up.

But I'm concerned that all of you who are trying to find the problem will read the spoilers for Shingu. Not good; they give away the ending.

24 Got it! Comment 7 in "Shingu 20" is missing a closing DIV tag around the spoiler.

-j

25 Thanks J!  You're spot on.

Now I just have to fix it...

26 Fixed the page; fixing the code now.  Looks like I need to improve the error-checking in the BBCode module.

27 Doh!  I have the necessary checking in place; it just didn't happen because I assigned "spoiler" to the wrong tag class.

28

Did I forget to put in a close-spoiler tag on that one? Oops... (But not too much of an oops; just as well it happened, since it revealed a problem in Pixy's code.)

Pixy, thanks for your effort on this.

29 Yep.  The bbcode engine is supposed to track opened and closed tags and automatically balance them no matter what you do, so that was a bug that needed fixing.

1 Registered users with mee.nu, or registered to comment on your site specifically? I wasn't going to create a mee.nu account (it's not like I need another email/web addy) but if that's what it takes to be allowed commenting privileges...

2 Thanks Steven.

3 GreyDuck, there are a couple of ways in the system to restrict commenters, but the easiest one is to set it to require a registered mee.nu user.  There's no site-specific registration.

4 Pixy: Thanks for the answer. No worries, then: I'll go ahead and sign up. Maybe I'll find a use for it some day other than giving me the ability to pester Steven with comments. *wry grin*

5

Cool. And we commenters register where?  (Don't say "At the link, dummy," because that's not a user registration, it's just to get a mee.nu website.)

Hmmmm. Wait, I think I get the picture here..... 

6 Wow. Talk about being ninja'd....

7

I tried again, thinking maybe I had misunderstood something about the registration.  No joy.

"Sorry, we are currently only allowing sites with mee.nu addresses. We plan to offer support for personal domains soon." 

So for the time being, to comment on any "restricted" mee.nu blog, I have to have a mee.nu blog?

8 Yes.  The blog is free but mandatory.

I thought of changing that, but there's some code that assumes that the user has a site as well, so for now, yeah, you have to have a mee.nu blog.

9 I'll change the error so that it actually detects a blank site field.

10 I'm in. Maybe I'll even think of something to use the site for...

11 Went ahead and signed up too. Wanted one anyway. Now to figure out how HTML works...

12 Just testing something out. . .

13 A mandatory blog is not unusual. LJ does the same thing. In Pixy's shoes I would not strain myself trying to untie the registration from the domain, profile page, etc.

BTW, I have my doubts about the concept of commenting. Sure, it allows me to make a fool of myself in comments at Chizumatic easily, in fact the commenting facility makes it positively tempting, but is this a good thing?

I am thinking about experimenting with the concept of "Blogosphere is my comment board" by disallowing comments, leaving fewer comments, and placing more links. It seems to work fine for political bloggers like Instapundit, so... In fact the Chizumatic itself ran without comments for years.

14

From my own personal experience, the reason it worked for me when I was hosting it on my own server was that I had direct access to the Apache referer logs. My browser "home" page is a summarized referer log on my own server, and I look(ed) at it quite a lot.

That meant that if someone linked to me, I'd see it and could respond if I felt I wanted to.

Trackbacks would perform the same function -- except that trackbacks are essentially useless now because of spammer abuse.

Running a blog on a hosting service (like this one) without access to trackbacks or any kind of access to referer logs makes it nearly impossible to even notice when someone else has linked to you. So far, my only way of realizing it has been when it's someone whose blog I read, and I happen to visit.

Pixy is promising us a statistics page, but it isn't up yet. (He's been busy with more important things.)

Comments here have gotten out of control a couple of times, but I've been applying the Giuliani "broken window" approach to enforcement, and it seems to be working pretty well. There's only been one commenter who has been outright abusive and needed to be banned, and I had to ask Pixy to take care of it.

I certainly understand that it's a pain to have to register in order to comment, but with the software here as it currently exists that gives me a degree of control without having to beg Pixy for help in egregious cases. And I don't want to be in a position of being forced to switch as a result of the next abusive commenter that shows up. I won't give him that satisfaction.

I certainly don't intend to be tyrannical about using the ban-hammer, but I will continue to use the "broken window" philosophy to deleting comments and closing threads.

15 I see. Thanks for the explanation.

Perhaps Pixy can cook up something regarding logs in due time, as a part of the further site development. Mere statistics is probably too coarse...

BTW, I found that Google pretty good at finding backlinks. Try this:
http://blogsearch.google.com/blogsearch?hl=en&tab=wb&q=chizumatic&btnG=Search+Blogs

16

It's not really very good. I know of at least two recent links to here that are not on that list.

17 Not a problem ... I'll just handle it like live journal ... a page that consists of nothing but a forward to my blog.

1 I suspect the reason they switched us back was that they needed to flame-test in order to trace the problem.

2 Huh.

That might explain why I'm not seeing any problems now.

They did switch the Global Exchange link back on earlier today, and it was working at that time, but Globlx might have screwed the pooch again.

The routing change has added 25ms to my ping time, but since it's was 200ms already, that's not very noticeable.

Softlayer are busy provisioning a new 10Gb link, which might help sort this stuff out.  Or not, since I believe that is with Globlx too.

3 After half an hour it's still clean. I'm not going to keep posting these updates; I'll let you know if it goes bad again.

4 Thanks.

5 I'm back to 30% lost packets this evening. I don't think there's any point in complaining about it; it's evident that they know what the problem is, and they're working on it. It's also evident that they have a bypass but not a solution, so they're occasionally switching the bypass off again so they can work on the problem further. I'm just going to have to put up with it until they figure out how to fix it for real.

6 Well, it's clean again, but they don't appear to have thrown a switch this time. The fail rate just got lower and lower and now it isn't losing anything. That could be simply because of reduced late-night traffic, of course. I'll leave pingplotter running overnight again, and we'll see what it looks like in the morning when traffic picks up.

7 No updates on Softlayer's forums, but Global Exchange and Comcast aren't necessarily notifying them every time they change a router setting.

8 Steven, I don't know if this is related or not (I doubt it), but sometimes I get your full page load with all the backgrounds, and other times I get it without any formatting beyond the header and some basic text format.  If it were WP, I'd say the main.php wasn't loading one of the templates. Not sure about Minx.

9 That's what happens if the main page loads successfully but the CSS file load is missed. It means you've been having communications problems with the server, too.

10 ubu, how often does it happen?

11

Right now, it's happening every single time. The page loads quickly enough, but I just get the picture, the MEE (heart) NU graphic, four links at the top, and then the rest of the page shows up as otherwise unformatted hypertext: all the links are there, text is bolded, etc. but there is no background, no other graphics (except the comment box), no post formatting.  If it is the CSS file, then I'm probably seeing the browser defaults for each of these.

Other times, I'd get this on the first pass, but a reload would get the correct screen, sometimes with a brief hiccup during loading.  It's been going on since before Steven started complaining, but now it's constant.

12 Correction to the "no other graphics:"  I am seeing the ping plot.

13 Urm.  That's pretty weird.

Try loading the stylesheet itself: http://chizumatic.mee.nu/files/mee.css

See if you get an error or a timeout or something else.

14

Ubu, try clearing your browser cache. I would suspect you have a corrupt CSS file but your browser thinks it's up-to-date.

(By the way, as far as I know this site doesn't have any background.)

15

Well, I'm going to go for the bad cache theory.  It's been too busy for me to take a break and stop back by until now -- and everything's back to normal. (No, you don't have a background graphic, but your posts are on a white background and the rest of the screen is gray.)

1 Yep, it's looking much better now.  Softlayer posted this message:

We have turned down our BGP peer with Global Crossing in response to numerous complaints of connectivity issues. We will continue to monitor this situation closely until Global Crossing has resolved this fully.

2 Is there a new virus floating around that has got everything boloxed up? My machine just picked up a Winfixer bug (caught it, thankfully) off 8-bit Theater over at nuklearpower.

3 Will, why are you posting that here? What's it got to do with me?

4 Concern that a big spreading virus might be the cause of the internet troubles?

( though, how exactly can you catch a virus off of a web page? )

5

That's mostly it. If there's a new bug out there that everybody's fighting with, it could casue connection problems. Our IT lady mentioned there was a big critical update coming out for Windows today as well. Hopefully the connection troubles are local and easy to fix.

As for catching a virus from a web page, you do it the same way you get spyware and malware. Some sites are malicious and will fish for hits (beware most bittorrent and warez sites) and send along all kinds of nasty stuff back through your browser. Others are compromised by a third party looking to spread their particular bug.

6

how exactly can you catch a virus off of a web page?

It helps if you run IE.

Anyway, this problem doesn't appear to have anything to do with viruses, but with specific routing problems in the Global Exchange network in Texas.

7

There are known exploits for IE that can force code on you involuntarily and infect your system. Most of those have already been patched by Microsoft, but a lot of people are behind on their patches.

But as Pixy says, the problems I've been having for the last few days have nothing whatever to do with that.

8 Ah, okay.  *hugs copy of Firefox*

9 There are people that still use IE?

10 I've enabled a second web server, so you now show up at two IP addresses.  If you see anything untowards, you know who to yell at.

11 ubo roi- Mostly people who don't know about Firefox, or are technically unsure and unwilling to DL something.  It took until IE7 before I could convince my mom to switch, for example. . . and even that depended on the fact that Firefox is actually *more* like IE6 than IE7 is.

( I don't care what upgrades they have, IE7 has to be the stupidest interface 'upgrade' ever )

1 If people could also add their internet provider and state, that would be very useful in tracking down the problem.

2

Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 44ms, Maximum = 68ms, Average = 46ms

Also Comcast from Richmond, VA around 1 pm EST

3 Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 57ms, Maximum = 62ms, Average = 57ms

twtelecom.net, Seattle WA

4 I seem to be having problems copying from the command prompt, so I'm going to type this out.

Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 43ms, Maximum = 72ms, Average = 47ms

I'm using Fidelity Communications in Missouri at about 12:30 central time.

5 From xo.net, Oakland, CA:

Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 47ms, Maximum =  70ms, Average =  48ms

From covad.net, Oakland, CA (OpenBSD system):

--- mee.nu ping statistics ---
100 packets transmitted, 100 packets received, 0.0% packet loss
round-trip min/avg/max/std-dev = 49.891/59.955/192.932/13.964 ms

6 Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 41ms, Maximum = 57ms, Average = 45ms

Comcast, Monroe, MI

7

Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 46ms, Maximum = 88ms, Average = 53ms

From Phoenix, AZ via XO network

Would traceroutes be useful?

8 OK, thanks; I think that's enough. It looks like this is a Comcast problem, and I'll have to get onto Comcast's tech support about it. I can't access this site directly right now at all; I'm running 70% losses on pings. But I'm able to get in easily using a proxy, which is why my IP hash is different.

9 Or maybe not. Pixy is having problems, too. If you do this and get 100% we don't need to hear about it, but if there's anyone else having problems, please let us know.

10 Well, here's from my server at the Infinity Internet hosting facility (Linux ping, big fat hosting pipe): --- mee.nu ping statistics --- 100 packets transmitted, 100 received, 0% packet loss, time 99082ms rtt min/avg/max/mdev = 67.818/69.800/83.695/2.105 ms, pipe 2 The story's entirely different from my office (Windows ping, Comcast business-class broadband): Ping statistics for 75.126.89.164: Packets: Sent = 100, Received = 29, Lost = 71 (71% loss), Approximate round trip times in milli-seconds: Minimum = 77ms, Maximum = 84ms, Average = 80ms Portland OR, both cases.

11 I was getting 97% packet loss just now, and from the traceroutes the problem was somewhere in Dallas, just outside our hosting company's network.

Because of the problem, I couldn't even reach the support site to log a fault.

Then suddenly everything started working, at least sort of.

12

I just saw a 2% loss, but I don't know who is the provider here at work.

13 They're not having storms in Dallas again, are they?  Anyway, I have some servers just up the road and those ones are fine.

Since I couldn't reach the hosting company's support site, I posted a message on their forums.  Now I can reach the support site, but can't reach the forums.

Both mee.nu and mu.nu seem to be happy right now, so I'm off to take a short nap.

(It's 4:30 AM here in Sydney.)

14 Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 56ms, Maximum = 60ms, Average = 58ms

Earthlink from San Francisco CA 11:50am

15 My hosting company got back to me - apparently there's a routing issue between Comcast and Global Crossing.  They are poking the engineers involved with pointy objects.  As all you engineers know, that's what they do when you are busy trying to fix something.

16 And now I'm really going to bedzzzzzzzzz

17 Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 19ms, Maximum = 31ms, Average = 20ms

In Houston, Texas.  We're changing from Roadrunner to Comcast and my bills now come from Comcast.  I guess the best thing to tell you is probably that my current IP is 70.114.245.*

Bob Price

18

Hoping it won't be redundant...

Statistiche Ping per 75.126.89.164:
    Pacchetti: Trasmessi = 100, Ricevuti = 99, Persi = 1 (1% persi),
Tempo approssimativo percorsi andata/ritorno in millisecondi:
    Minimo = 161ms, Massimo =  300ms, Medio =  210ms

ISP Fastwbenet.it (Italy), 19:08 UTC

19 Ping statistics for 75.126.89.164: Packets: Sent = 100, Received = 99, Lost = 1 (1% loss), Approximate round trip times in milli-seconds: Minimum = 38ms, Maximum = 82ms, Average = 53ms

20 OK, I don't think we need any more help. Thanks, everyone!

21 --- mee.nu ping statistics ---
100 packets transmitted, 100 received, 0% packet loss, time 106336ms
rtt min/avg/max/mdev = 53.581/55.834/88.960/5.110 ms

AT&T WorldNet, Colorado (my office).

22 oops...I must have been pulling the info off my other window just as you posted the "thanks everyone!" message...

23 The internet is full of weirdness and irony. I've had pingplotter running all morning, and it's been getting progressively worse. For the last hour it's been solid 100% packets lost. Which means that the only way I can get from Oregon to Texas to leave comments here is by using Anonymouse.org, a free anonymizing proxy in Germany.

24 Ping statistics for 75.126.89.164:
    Packets: Sent = 100, Received = 85, Lost = 15 (15% loss),
Approximate round trip times in milli-seconds:
    Minimum = 45ms, Maximum = 48ms, Average = 46ms

Verizon FiOS, PA

(Doesn't look like it's getting any better ...)

25 And I don't have to go through the proxy anymore. Yaay!!!

26 I can't believe this; it's gotten worse again! Come on, you guys! Get your act together! I'm back to 41 out of 100 lost pings.

27 It's gotten gradually better in the last two hours but it's still not ideal. I'm going to leave pingplotter running overnight.

1 Urm.  No, that shouldn't be making things painfully slow.  Let me take a look around.

The elapsed time to create this page was 33ms (server side), so the problem may be elsewhere.

2 The timings I'm seeing look pretty normal.  Queries take slightly longer now because they have to go across the network, but that only adds a few tens of milliseconds even to a large page.

Let me know if things are still slow from where you are, and I'll see if I can track it down.

3

Right at the moment it seems pretty brisk. But earlier today it was taking upwards of a minute for me to load the page. In a couple of cases the browser timed out.

Even when it took 20 or 30 seconds to load, the report at the bottom would say that it only took a fraction of a second. I think thats describing CPU-time, not wall-clock time.

That also applied to the management pages, when I was trying to post.

4 I think it might have been Comcast's problem, not yours. There was another site I couldn't reach at all about two hours ago and now it responds briskly.

5

I think thats describing CPU-time, not wall-clock time.

Yep.  I have a custom setting in that code, and if it's me who's fetching the page I get a more detailed report.  So I can see that this page took 10 milliseconds of CPU time, but 140 milliseconds elapsed time.  (Which is kind of sluggish.  Hmm.)  But clearly you aren't talking about 100ms here or there.

6 If it happens again, I'll do some traceroutes and pings.