Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.
I'm pretty sure this is the same bug that Ben Webster reported earlier. I was hoping somebody might be able to help figure out what's going on.
In the computer lab in 744 Evans (at Berkeley), individual question pages are painfully slow, especially with scrolling and inputing any text (answers or comments). The home page is perfectly responsive when I scroll through it (well, maybe not perfectly responsive, but pretty good). Entering text is perfectly responsive on the "ask a question" page and when editing an answer. The question pages on meta.SE are perfectly responsive. (Edit: scrolling and typing answers is responsive on meta.SE, but typing comments is painful)
I thought the problem might be that MO loads a bit of extra javascript in the footer, so I tried disabling javascript, but it didn't help. In fact, it made the MO home page and meta.SE slow. The ask page was still responsive.
Any ideas for what's going on or how we might be able to fix it?
The problem could be throttling by the IIS webserver.
All the machines here are thin clients that go through the same machine. In particular, everybody browsing from 744 shares an external IP, and that IP may be throttled. Looking at the data I have, it doesn't look like that many visits come from that IP. Also, that wouldn't explain the way in which the pages are slow (scrolling, commenting, and answering all happen entirely on the client side until you hit submit). I'll try to get in touch with the sysadmin here to see if he has any ideas.
@Anton. All the machines might not have the same IP address. More connections would obviously be going from the computer lab.
The commenting etc., must be slowing because of the throttling of the AJAX connections. The AJAX connections are alive even when answering, for instance if some answer is submitted while you are typing, you are notified. I do not however understand the scrolling. Perhaps there is some AJAX script running there too(whatever for, I have absolutely no idea).
I do not feel it can be fixed locally. If you have access to a server where MO is running, then you could proxy all connections to the IIS server of SE through thttpd(configured without throttling). But then you also will have to look at the thttpd log instead, to catch trolls and all. Best might be to email the SE guys explaining the problem and hinting that the reason might be throttling at the server end.
I think the problems are more likely to be due to the software/computers in 744 than anything else. I sometimes have similar problems on firefox on those thin clients when I visit other websites that are probably ajax/java/whatever script intensive, like google reader, google documents, gmail, espn, facebook, etc.
@Kevin: but that doesn't explain the behavior
I thought the problem might be that MO loads a bit of extra javascript in the footer, so I tried disabling javascript, but it didn't help. In fact, it made the MO home page and meta.SE slow. The ask page was still responsive.
Another guess: I recall(from long, long ago) that HTTP 4.0 had a specification like, at the most 4 or so simultaneous connections from the same client. This combined with AJAX keepalive connections, could slow things down.
California budget cuts.
Or it could be many other client-specific things, like the way jsMath fonts are rendered by your X server.
@Tom- I fail to see how California budget cuts would cause the same problem on the computer in my office at MIT. I think there's some other unifying factor.
Does it happen if you bring a laptop from home and plug it into the wall?
@Alberto GarciaRaboso.
The office machines could have a fixed IP address, and when you connect the laptop, it could be actually be part of a different network, obtaining its IP address from a DHCP server. This is definitely the case if you use wlan.
And these two IPs could be routed through different proxies in the department, and there could be less traffic in the route which goes from the laptop. So what I said could still be the case.
Was this issue solved? If not, what about asking at meta.stackexchange?
Hey! I'm in 744 right now, and scrolling/typing seems to be working just fine! (Or at least, way better than the last time I used these machines.) What's up?
I haven't changed anything that should have any effect. Must be that either the math department sysadmins have done something or Fog Creek did something.
I've posted a question on meta.SE
If two people vote it up, I will be able to vote on meta.SE
@Ben Webster.
I have taken upon myself the honor of knighting you up to 51 reputation there. Have a nice time voting!
Yeah, and now someone who moderates meta.SE is going to a very suspicious pattern, as I just voted up about 15 of Anton's posts.
Yup, it looks like you voted up about 12 of my posts in a span of 2 minutes, and you've only cast 15 votes. I'll bet money that those votes will be automatically thrown out. They probably won't notice because they haven't implemented this request of mine yet (which, incidentally, you didn't vote up).
btw, if there's anyone with meta.SO reputation (that's StackOverflow, not StackExchange), could you vote up my lame answer there? I'd like to be able to vote but have never found much to say there.
1 to 23 of 23