Community
Participate
Working Groups
Wiki is painfully slow at the moment, almost grinding to a halt. Bugzilla is also slow, but not quite as bad. Please investigate. Eclipse.org services have been less than optimal recently. What's the reason?
Hi Martin, Nothing appears to be wrong on our end. Can you try and post some numbers so I can compare your times with ours? TCP traceroute to port 80 [root@host ~]# traceroute -T -p 80 wiki.eclipse.org traceroute to wiki.eclipse.org (206.191.52.45), 30 hops max, 60 byte packets [snip] 12 wiki.eclipse.org (206.191.52.45) 26.575 ms 28.140 ms 29.208 ms Fetch Main_page [user@host]$ time wget --delete-after --header="Accept-encoding: gzip,deflate" http://wiki.eclipse.org/Main_Page [snip] 2011-03-02 09:56:30 (223 KB/s) - “Main_Page” saved [6716/6716] real 0m0.199s user 0m0.001s sys 0m0.002s As you can see, I can reach the host in under 30ms, and the front page comes in within 200ms.
It's better again now. half an hour ago, I tried to edit a page (approx 15 sec), then click on the "Rich Editor" hyperlink (approx 30 sec). Bugzilla open known bug (approx 4 sec). These times are "felt" by me. I'll try to use the traceroute and wget next time such a thing happens. Right now, I see this: [root@szg-qa-lx2-64 ~]# traceroute -T -p 80 wiki.eclipse.org traceroute to wiki.eclipse.org (206.191.52.45), 30 hops max, 40 byte packets 12 * * rx0sh-mountain-cable.mt.bigpipeinc.com (66.244.223.238) 354.612 ms 16 206.191.0.120 (206.191.0.120) 149.352 ms 150.938 ms 152.779 ms 17 wiki.eclipse.org (206.191.52.45) 152.729 ms 152.821 ms 152.860 ms 2011-03-02 17:05:04 (21.9 KB/s) - `Main_Page' saved [6718/6718] real 0m3.857s user 0m0.002s sys 0m0.004s I still think that 4 sec for a high-profile website like wiki.eclipse.org is not breathtaking.
Martin, Your 152ms round-trip to us is definitely hurting. The 352ms hit to bigpipe.com is definitely a problem. However, from dev.eclipse.org, I can hit that same host in only a few miliseconds: dev1:~ # /usr/sbin/traceroute -T -p 80 rx0sh-mountain-cable.mt.bigpipeinc.com traceroute to rx0sh-mountain-cable.mt.bigpipeinc.com (66.244.223.238), 30 hops max, 40 byte packets using TCP 1 css1 (206.191.52.33) 1.323 ms 1.519 ms 1.293 ms 2 206.191.16.219 (206.191.16.219) 2.224 ms 52.025 ms 51.962 ms 3 206.191.0.65 (206.191.0.65) 50.887 ms 43.147 ms 2.045 ms 4 border1-faste2-0.magma.ca (209.217.64.38) 2.000 ms 1.784 ms 0.837 ms 5 10.201.122.206 (10.201.122.206) 9.445 ms 32.839 ms 31.766 ms 6 10.201.122.201 (10.201.122.201) 30.704 ms 29.640 ms 28.578 ms 7 10.201.120.197 (10.201.120.197) 27.495 ms 26.435 ms 25.368 ms 8 10.201.120.78 (10.201.120.78) 24.299 ms 36.628 ms 11.745 ms 9 rx0sh-mountain-cable.mt.bigpipeinc.com (66.244.223.238) 8.928 ms 9.016 ms 9.099 ms There's nothing at my end that would indicate a problem.
Do 152msec round-trip explain a 3.8 sec load time of wiki.eclipse.org ? This would mean 25 round-trips for loading one page?
> Do 152msec round-trip explain a 3.8 sec load time of wiki.eclipse.org ? > This would mean 25 round-trips for loading one page? Such a large lag time will definitely impact your load time. If you access http://wiki.eclipse.org I count 32 network packets just to get the HTML page. I can repeatedly get that page in under 300 miliseconds. I have the advantage of being geographically close to eclipse.org. Unfortunately, I can't test access to eclipse.org from anywhere else since I don't have shell access.
I've tried a few online tools to measure the website performance, and for the most part, the Wiki main page comes in under one second. I'm going to close this as not eclipse ... if there's anything I can do to help, please let me know.
