Beefy Boxes and Bandwidth Generously Provided by pair Networks
Perl: the Markov chain saw
 
PerlMonks  

PerlMonks is *slow*. What do we need to fix?

by rr (Sexton)
on Oct 24, 2002 at 16:54 UTC ( #207778=monkdiscuss: print w/ replies, xml ) Need Help??

Hello

Perlmonks is very slow. I have tested this site from many locations, and although network connection is bad from some locations, it seems that it's the site itself that is slow.

[rebel@hydrus rebel]$ /usr/sbin/traceroute www.perlmonks.org traceroute: Warning: www.perlmonks.org has multiple addresses; using 2 +09.197.123.153 traceroute to perlmonks.org (209.197.123.153), 30 hops max, 38 byte pa +ckets 1 192.168.123.254 (192.168.123.254) 1.181 ms 1.109 ms 1.962 ms 2 10.33.96.1 (10.33.96.1) 9.433 ms 9.131 ms 18.006 ms 3 pos0-1-nycmnyb-rtr2.nyc.rr.com (24.29.97.113) 9.216 ms 8.780 ms + 9.920 ms 4 pos3-0-nycmnyd-rtr1.nyc.rr.com (24.29.97.21) 10.215 ms 10.408 ms + 10.445 ms 5 pos5-3-nycmnya-rtr2.nyc.rr.com (24.29.97.38) 9.291 ms 24.512 ms + 10.025 ms 6 pop2-new-P3-1.atdn.net (66.185.137.29) 10.729 ms 9.779 ms 10.44 +8 ms 7 bb2-new-P0-1.atdn.net (66.185.137.18) 10.051 ms 10.336 ms 10.91 +9 ms 8 bb1-nye-P4-0.atdn.net (66.185.152.196) 209.445 ms 214.379 ms 21 +3.826 ms 9 bb2old-nye-P1-0.atdn.net (66.185.152.87) 12.547 ms 10.887 ms 11 +.259 ms 10 bb2-vie-P8-0.atdn.net (66.185.152.201) 163.221 ms 211.253 ms 22 +7.308 ms 11 bb2-chi-P6-0.atdn.net (66.185.152.214) 35.390 ms 34.344 ms 34.5 +21 ms 12 aol-gw.cgcil.ip.att.net (192.205.32.237) 36.900 ms 36.473 ms 38 +.336 ms 13 tbr1-p013902.cgcil.ip.att.net (12.122.11.205) 37.986 ms 36.807 m +s 40.990 ms 14 * * * 15 gbr5-p30.n54ny.ip.att.net (12.122.11.10) 57.526 ms 56.267 ms 96 +.317 ms 16 gbr4-p100.n54ny.ip.att.net (12.122.5.110) 55.172 ms 60.944 ms 5 +5.466 ms 17 gbr1-p20.phlpa.ip.att.net (12.122.2.18) 58.739 ms 58.249 ms 58. +922 ms 18 gar1-p360.phlpa.ip.att.net (12.123.137.21) 57.630 ms 58.248 ms +58.251 ms 19 12.125.176.98 (12.125.176.98) 67.124 ms 64.718 ms * 20 perlmonks.pair.com (209.197.123.153) 65.605 ms 66.857 ms 82.911 + ms bash-2.05$ traceroute www.perlmonks.org traceroute: Warning: www.perlmonks.org has multiple addresses; using 6 +6.39.54.27 traceroute to perlmonks.org (66.39.54.27), 30 hops max, 40 byte packet +s 1 user.hsrp.w18.nyc.access.net (166.84.1.30) 0.934 ms 0.714 ms 0. +446 ms 2 S8-0-0-T3.l3core.nyc.access.net (166.84.64.14) 1.879 ms 1.554 ms + 1.327 ms 3 panix-r.netaxs.net (207.8.210.94) 3.510 ms 2.344 ms 2.667 ms 4 169.130.235.13 (169.130.235.13) 2.336 ms 3.420 ms 2.740 ms 5 12.125.51.209 (12.125.51.209) 3.441 ms 3.429 ms 3.009 ms 6 gbr2-p70.n54ny.ip.att.net (12.123.1.134) 3.282 ms 2.824 ms 2.78 +6 ms 7 gbr4-p70.n54ny.ip.att.net (12.122.1.177) 2.120 ms 2.973 ms 3.56 +0 ms 8 gbr1-p20.phlpa.ip.att.net (12.122.2.18) 6.507 ms 6.562 ms 6.271 + ms 9 gar1-p360.phlpa.ip.att.net (12.123.137.21) 7.136 ms 4.929 ms 5. +363 ms 10 * * * 11 ads.perlmonks.org (66.39.54.27) 11.623 ms 11.512 ms 11.275 ms
Are the servers over-worked? Is there some problem with database contention? Io contention?

I would like to discuss what we need to improve this situation, perhaps rally some hardware or money.

Perhaps the site admins can help us on this topic.

rr

Comment on PerlMonks is *slow*. What do we need to fix?
Download Code
Re: (nrd) PerlMonks is *slow*. What do we need to fix?
by newrisedesigns (Curate) on Oct 24, 2002 at 17:01 UTC

    What PerlMonks really needs is updates to its code. If you're willing to lend a hand to pmdev, that would be your best bet.

    Turn off nodelets and monk quips/images. Your pages will load faster because you are requesting less from the server.

    See also this node.

    John J Reiser
    newrisedesigns.com

      Really? I saw a few conflicting points on that, but I would not doubt you as sites like this often have organic and messy code bases. I would be glad to help, I didn't see any join pmdev information tho, where is it?

      As far as hardware goes, if some were donated would PAIR allow installation? What hardware is in use now? Is there some documentation about this around?

      I did disable my nodelets, which helps a little bit but not much.

      I am seriously honest when I say I would contribute and use this site more often if it were not so infuriatingly slow. BTW, this is not criticism, it's an observation and I am willing to help (aka not just complain).

      rr

        I am seriously honest when I say I would contribute and use this site more often if it were not so infuriatingly slow. BTW, this is not criticism, it's an observation and I am willing to help (aka not just complain).

        That's wonderful to hear. I don't have much time or know-how to really work on a site this large, so I would feel very hypocritical if I complained about the speed.

        I think it is wonderful that you are willing to help. If you contact the pmdev or gods groups, they would be able to tell you about how to help.

        John J Reiser
        newrisedesigns.com

        As far as hardware goes, if some were donated would PAIR allow installation? What hardware is in use now? Is there some documentation about this around?

        That isn't necessary. We (pair) are already donating all of the hardware and bandwidth that perlmonks is using for free. Presently that includes three of our quickserve dedicated servers. We've also indicated to vroom that we'd gladly add another server if he thought it would make a difference. But right now, I think the plan is to try to improve the code a bit. You can only throw so much hardware at a problem before it doesn't make much of a difference any more. :)

Re: PerlMonks is *slow*. What do we need to fix?
by Ovid (Cardinal) on Oct 24, 2002 at 17:17 UTC

      I'd have thought everyone would be thrilled to learn we're all part of a giant non-sociology experiment. :) Well, at least Perl Monks turned out to be one of my favorable examples...

Re: PerlMonks is *slow*. What do we need to fix?
by rinceWind (Monsignor) on Oct 25, 2002 at 16:24 UTC
    I guess those of us who are using PM every day get used to the delays. However, one comment I have had from friends and colleagues is how slow the response is. At that point, I would explain about the nature of CGI, and make out that the application has to do quite some work before it displays a page ... excuses, excuses.

    I don't think this is doing the reputation of Perl any good, as a web application language - IMO it's time this was addressed.

    Before replying to this post, I decided to download the Everything engine, just to take a look for myself. One of the pieces of advice that monks are meteing out on SoPW regularly, when CGI apps are slow, is to consider ModPerl.

    Has anybody thought about, or even got working, an Everything Engine on ModPerl? From a cursory inspection, it does not look too bad, as the CGI code itself seems to be localized into the module html.pm.

    By the way, I am interested in PM Dev as well. Not sure how much time I can give (possible issues with PHBs and circular tuits).

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: monkdiscuss [id://207778]
Approved by ChemBoy
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others making s'mores by the fire in the courtyard of the Monastery: (8)
As of 2014-07-31 17:38 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    My favorite superfluous repetitious redundant duplicative phrase is:









    Results (249 votes), past polls