View previous topic :: View next topic |
Author |
Message |
JarrettBillingsley
Joined: 20 Jun 2006 Posts: 457 Location: Pennsylvania!
|
Posted: Mon Aug 14, 2006 9:14 am Post subject: Help, I am drowning in molasses |
|
|
I love dsource, and I can definitely see it becoming the site for D developers.
But IT IS THE SINGLE SLOWEST SITE I HAVE EVER EXPERIENCED.
Lars said it was something to do with Trac not liking PostgreSQL. Is this true? Can anything be done about it?
It's been really bad lately. I mean, waiting upwards of 60 seconds for a single page to load. It's getting to be a pain to manage my project.
Is there anything that we can do to help? Is it just a matter of manpower, or are there not many options to fix this?
Please, help! I do not like drowning in a syrupy goo. |
|
Back to top |
|
|
StringCheesian
Joined: 31 Mar 2005 Posts: 17
|
Posted: Mon Aug 14, 2006 2:14 pm Post subject: |
|
|
Not quite as bad as 60 sec here, but it's pretty bad for me too. |
|
Back to top |
|
|
Derek Parnell
Joined: 22 Apr 2004 Posts: 408 Location: Melbourne, Australia
|
Posted: Mon Aug 14, 2006 5:13 pm Post subject: Re: Help, I am drowning in molasses |
|
|
JarrettBillingsley wrote: | I love dsource, and I can definitely see it becoming the site for D developers.
But IT IS THE SINGLE SLOWEST SITE I HAVE EVER EXPERIENCED.
|
Wow!? I'm here in Australia and it takes about 3-5 seconds to load any dsource page (I use Opera 9 and cable modem) . _________________ --
Derek
skype name: derek.j.parnell |
|
Back to top |
|
|
JarrettBillingsley
Joined: 20 Jun 2006 Posts: 457 Location: Pennsylvania!
|
Posted: Tue Aug 15, 2006 8:43 am Post subject: |
|
|
Absolutely bizzare. whois says the server's in Plano, Texas - much closer to me than you! I'm also on a cable modem.
It's net non-neutrality! Aaaa! |
|
Back to top |
|
|
pragma
Joined: 28 May 2004 Posts: 607 Location: Washington, DC
|
Posted: Tue Aug 15, 2006 9:54 am Post subject: |
|
|
JarrettBillingsley wrote: | Absolutely bizzare. whois says the server's in Plano, Texas - much closer to me than you! I'm also on a cable modem.
It's net non-neutrality! Aaaa! |
Jarret, have you tried performing a traceroute to dsource.org? Maybe your request is being routed halfway across the country before it gets to Plano?
I'm no network guy, but its possible that there's some bad routing going on, or you and dsource are just on opposite ends of a particular chunk of the internet.
If you ask me, it looks like it's time to switch providers - you're not getting what you're paying for*. Maybe you can make it worthwhile: grab a business DSL account and resell wireless access to your neighbors (provided you don't live in the stix that is)?
(*This goes automatically if you're with Comcast like I am)
Edit
Just found this on digg - it should help: http://www.speedtest.net/ _________________ -- !Eric.t.Anderton at gmail |
|
Back to top |
|
|
StringCheesian
Joined: 31 Mar 2005 Posts: 17
|
Posted: Tue Aug 15, 2006 12:00 pm Post subject: |
|
|
I have a 3 megabit connection and tests show it's as fast as it should be.
The page downloads very quickly, but before it downloads there's a delay of about 5-10 seconds while the status bar says "Waiting for www.dsource.org...". After I keep using the site for a while the delay seems to shrink to 2-3 seconds. No other site does that to me. I'm used to delays of more like 1/5 to 1 second.
Generally I see that sort of thing when the server is misconfigured or poorly coded so that PHP or some SQL query is running slow. I note that dsource.org/forums is a lot slower than dsource.org...
But it looks like there is also a routing issue: Code: | sterling sterling # tracepath www.dsource.org
[snip]
8: ded1-g8-3-0.hstntx.sbcglobal.net (151.164.241.101) asymm 9 60.279ms
9: Mirus-Information-Technology-Services-1062215.cust-rtr.swbell.net (68.92.42.114) asymm 10 78.023ms
10: no reply
11: no reply |
It just keeps printing "no reply" as long as I let it. How come all my browsers can find it but tracepath can't?
EDIT: traceroute does the same thing, but tcptracepath gets farther than Mirus: Code: | 9 ded1-g8-3-0.hstntx.sbcglobal.net (151.164.241.101) 41.823 ms 42.024 ms 43.411 ms
10 Mirus-Information-Technology-Services-1062215.cust-rtr.swbell.net (68.92.42.114) 49.622 ms 44.842 ms 47.219 ms
11 68-91-106-142.ded.swbell.net (68.91.106.142) [open] 52.724 ms * [unknown, ACK] 99.517 ms |
It stops there as if it has found it, but 68.91.106.142 and 68.91.106.142/forum both give a 404. |
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Tue Aug 15, 2006 7:28 pm Post subject: |
|
|
Guys,
I am aware the server is slow. The connection leak that I experienced was "fixed" when I switched from psycopg2 to pypgsql (postgres drivers for Python). All I believe it really did was to move the leak into the FastCGI processes, which begin to take up massive amounts of memory. Then the server begins to swap and it gets uglier from there.
I think that Derek just got lucky, the server is in Houston, it doesn't respond to pings, so your traceroutes make sense, the server should be transferred soon to a much larger server with a larger pipe, and I was out of town for the past four days and wasn't really watching it. Phew - more later as I begin the transfer. Sorry for the delays.
BA |
|
Back to top |
|
|
JoeCoder
Joined: 29 Oct 2005 Posts: 294
|
Posted: Thu Nov 02, 2006 2:14 pm Post subject: |
|
|
The site is still very slow for me, both at home and at the university 25 miles away. Page load times are sometimes a few seconds but often a few minutes. Here's the tracert from univ. As you can see, it timed out.
Code: | Tracing route to dsource.org [68.91.106.142]
over a maximum of 30 hops:
[snip]
6 16 ms 24 ms 6 ms sd-ul.indiana.gigapop.net [192.12.206.245]
7 8 ms 10 ms 13 ms sl-gw36-chi-6-3.sprintlink.net [144.228.154.165]
8 9 ms 13 ms 9 ms sl-bb20-chi-5-0.sprintlink.net [144.232.26.69]
9 45 ms 56 ms 10 ms sl-bb22-chi-14-0.sprintlink.net [144.232.26.6]
10 12 ms 9 ms 11 ms sl-st21-chi-11-0.sprintlink.net [144.232.20.21]
11 11 ms 9 ms 8 ms ggr3-p330.cgcil.ip.att.net [192.205.33.157]
12 31 ms 33 ms 38 ms tbr2033901.cgcil.ip.att.net [12.123.4.250]
13 39 ms 34 ms 43 ms tbr2-cl7.sl9mo.ip.att.net [12.122.10.46]
14 37 ms 34 ms 59 ms tbr1-cl24.sl9mo.ip.att.net [12.122.9.141]
15 36 ms 40 ms 34 ms tbr2-cl6.dlstx.ip.att.net [12.122.10.90]
16 33 ms 31 ms 32 ms ar5-a3120s3.dlstx.ip.att.net [12.123.16.221]
17 31 ms 34 ms 64 ms 12.122.79.66
18 95 ms 41 ms 44 ms ded1-g1-3-0.hstntx.sbcglobal.net [151.164.241.99]
19 55 ms 58 ms 55 ms mirus-information-technology-services-1062215.cust-rtr.swbell.net [6
8.92.42.114]
20 * * * Request timed out.
21 * * * Request timed out.
22 * * * Request timed out.
23 * * * Request timed out.
24 * * * Request timed out.
25 * * * Request timed out.
26 * * * Request timed out.
27 * * * Request timed out.
28 * * * Request timed out.
29 * * * Request timed out.
30 * * * Request timed out. |
|
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Thu Nov 02, 2006 2:25 pm Post subject: |
|
|
JoeCoder wrote: | The site is still very slow for me, both at home and at the university 25 miles away. Page load times are sometimes a few seconds but often a few minutes. Here's the tracert from univ. As you can see, it timed out.
Code: | Tracing route to dsource.org [68.91.106.142]
over a maximum of 30 hops:
[snip]
6 16 ms 24 ms 6 ms sd-ul.indiana.gigapop.net [192.12.206.245]
7 8 ms 10 ms 13 ms sl-gw36-chi-6-3.sprintlink.net [144.228.154.165]
8 9 ms 13 ms 9 ms sl-bb20-chi-5-0.sprintlink.net [144.232.26.69]
9 45 ms 56 ms 10 ms sl-bb22-chi-14-0.sprintlink.net [144.232.26.6]
10 12 ms 9 ms 11 ms sl-st21-chi-11-0.sprintlink.net [144.232.20.21]
11 11 ms 9 ms 8 ms ggr3-p330.cgcil.ip.att.net [192.205.33.157]
12 31 ms 33 ms 38 ms tbr2033901.cgcil.ip.att.net [12.123.4.250]
13 39 ms 34 ms 43 ms tbr2-cl7.sl9mo.ip.att.net [12.122.10.46]
14 37 ms 34 ms 59 ms tbr1-cl24.sl9mo.ip.att.net [12.122.9.141]
15 36 ms 40 ms 34 ms tbr2-cl6.dlstx.ip.att.net [12.122.10.90]
16 33 ms 31 ms 32 ms ar5-a3120s3.dlstx.ip.att.net [12.123.16.221]
17 31 ms 34 ms 64 ms 12.122.79.66
18 95 ms 41 ms 44 ms ded1-g1-3-0.hstntx.sbcglobal.net [151.164.241.99]
19 55 ms 58 ms 55 ms mirus-information-technology-services-1062215.cust-rtr.swbell.net [6
8.92.42.114]
20 * * * Request timed out.
21 * * * Request timed out.
22 * * * Request timed out.
23 * * * Request timed out.
24 * * * Request timed out.
25 * * * Request timed out.
26 * * * Request timed out.
27 * * * Request timed out.
28 * * * Request timed out.
29 * * * Request timed out.
30 * * * Request timed out. |
|
The server does not respond to pings. The timeout is to be expected. As to the main problem, I am trying to restart often, when it gets slow. just let me know in IRC if i'm there (boorad) and I'll restart. KirkMcDonald has threatened to look at the offending code and hopefully offer up a solution.
BA |
|
Back to top |
|
|
aldacron
Joined: 05 May 2004 Posts: 1322 Location: Seoul, South Korea
|
Posted: Thu Nov 02, 2006 8:17 pm Post subject: |
|
|
I've gotten into the habit of working in other tabs while waiting for dsource to load in the background. I can usually read through several posts at GameDev.net before a single page loads here! Though, some times, it can be snappy. That must be shortly after you've rebooted. _________________ The One With D | The One With Aldacron | D Bits |
|
Back to top |
|
|
teqdruid
Joined: 11 May 2004 Posts: 390 Location: UMD
|
Posted: Fri Nov 03, 2006 10:19 am Post subject: |
|
|
brad wrote: |
The server does not respond to pings. The timeout is to be expected. As to the main problem, I am trying to restart often, when it gets slow. just let me know in IRC if i'm there (boorad) and I'll restart. KirkMcDonald has threatened to look at the offending code and hopefully offer up a solution.
|
So this is a total hack and clearly not a solution, but have you considered writing a small demon which monitors the mem usage of the offending process(es) and restarts them when they get too large?
~John |
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Fri Nov 03, 2006 10:41 am Post subject: |
|
|
teqdruid wrote: | brad wrote: |
The server does not respond to pings. The timeout is to be expected. As to the main problem, I am trying to restart often, when it gets slow. just let me know in IRC if i'm there (boorad) and I'll restart. KirkMcDonald has threatened to look at the offending code and hopefully offer up a solution.
|
So this is a total hack and clearly not a solution, but have you considered writing a small demon which monitors the mem usage of the offending process(es) and restarts them when they get too large?
~John |
The solution is fixing the code that can't find a cached env, and makes a new one taking up more memory.
The hack is restarting when it gets slow. I have a cron job that restarts every 24 hrs, but memory usage gets bad before that. I'm not sure I know how to do the mem usage detection. Do you have some script to do that?
BA |
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Fri Nov 03, 2006 10:41 am Post subject: |
|
|
teqdruid wrote: | brad wrote: |
The server does not respond to pings. The timeout is to be expected. As to the main problem, I am trying to restart often, when it gets slow. just let me know in IRC if i'm there (boorad) and I'll restart. KirkMcDonald has threatened to look at the offending code and hopefully offer up a solution.
|
So this is a total hack and clearly not a solution, but have you considered writing a small demon which monitors the mem usage of the offending process(es) and restarts them when they get too large?
~John |
The solution is fixing the code that can't find a cached env, and makes a new one taking up more memory.
The hack is restarting when it gets slow. I have a cron job that restarts every 24 hrs, but memory usage gets bad before that. I'm not sure I know how to do the mem usage detection. Do you have some script to do that?
BA |
|
Back to top |
|
|
teqdruid
Joined: 11 May 2004 Posts: 390 Location: UMD
|
Posted: Fri Nov 03, 2006 11:55 am Post subject: |
|
|
brad wrote: |
The solution is fixing the code that can't find a cached env, and makes a new one taking up more memory. |
Or disable the code that puts envs into a cache instead of freeing them.
brad wrote: |
The hack is restarting when it gets slow. I have a cron job that restarts every 24 hrs, but memory usage gets bad before that. I'm not sure I know how to do the mem usage detection. Do you have some script to do that?
|
Sure. Sorta.
Code: | ps u | grep 'PROCESS' | xargs | cut -d ' ' -f 4 - |
Will return the ? mem usage as reported by 'ps u'... Just check that number against a constant and restart if necessary.
For instance, to kill Kate when it eats memory, it'd use the following:
Code: | #!/usr/bin/perl
$z=`ps u | grep 'kate' | xargs | cut -d ' ' -f 4 -`;
if ($z > '1') {
$r = `killall kate`;
if (!$r) {
print "Killed!\n";
} else {
print "Failed kill\n";
}
} |
I love perl! |
|
Back to top |
|
|
lindquist
Joined: 14 Sep 2006 Posts: 55 Location: Copenhagen, Denmark
|
Posted: Wed Nov 08, 2006 9:17 pm Post subject: |
|
|
I'm experiencing fairly serious speed problems as well. I live in Copenhagen Denmark, and it usually takes me about 15-20sec to load a forum page. The project pages are slightly faster.
Most sites give me 0.5-2 sec load time and my download speed is somewhere between 200-800kb/s (depends ... duh!) |
|
Back to top |
|
|
|