First Webserver

First Webserver
Image by Arenamontanus
The world’s first webserver.

Question by GS3Central: How Do I set up a WebServer on my home PC?
I know about these two programs.

Abyss webserver
Apache webserver

I am looking for more eaiser to use programs.
I dont mind the cost, I am looking for the program.

Is it possible to get a program to host my website. With nice tools like Counters and stats and all that stuff.

Please, I know that just renting a server is better, but I am looking for setting up a server on my own computer. Thanks.

Best answer:

Answer by dawagnersjca
Search for a book at Are you running a Unix variant or Linux?

Give your answer to this question below!

Setting up the new webserver
Image by cyberfux

The first webserver ever
Image by polapix
Tim Berners Lees Nextstation at CERN
Polaroid SX-70 / 600 outdated 07/09
27.November 2009

Drop in CPU
Image by jspaw
Webserver CPU…anyone guess what the drop is from? (Flickr Staff, don’t answer :) )

Another hint here.

ANSWER IS: php5. :)

Apache on My iPod Touch
Image by reinvented
The Apache webserver running on my iPod touch.

pacific and Turq
Image by K·FREE
My webserver and my dog outside the bathroom door.

Sent from my iPwn. Apologies for typos.

29 Comments on First Webserver

  1. sronnoC
    November, 23rd 2012 at 7:38 pm


  2. rbreslav
    November, 23rd 2012 at 8:32 pm

    president’s speech?

  3. skorgu
    November, 23rd 2012 at 9:13 pm

    Disabling gzip?

  4. • glub •
    November, 23rd 2012 at 9:26 pm

    more ram?

    *revised*, nah, i’m thinking the obama effect as well.

  5. precipice
    November, 23rd 2012 at 9:58 pm

    Seems unlikely that Obama’s speech would lower traffic over a 16 hour period. He does go on a bit, but…

    Looks like the drop was at 8:15p in whatever time zone the graph reports. I’d assume Pacific (home of Flickr HQ), but it’s showing Wednesday at noon already, and it’s only 7:48a Pacific. So either the graph was earlier than this week, or it’s, say, GMT. Let’s guess GMT. If that’s right, the drop was Tuesday at around 9:15a Pacific time. Not the speech, again.

  6. jspaw
    November, 23rd 2012 at 10:06 pm

    Marc: good math. :)

    Not related to traffic, see the next graph taken from the same server, same time period.

  7. smin
    November, 23rd 2012 at 10:38 pm


  8. precipice
    November, 23rd 2012 at 11:12 pm

    Most of the things that would be big upgrades (PHP5, more RAM) would probably make at least one spike significantly lower than the ones on either side (RAM would be machine off/install/boot, for instance). The graph doesn’t seem to reflect that.

    I’m guessing either an external change (added to blah blah blah) or a settings change that would drop load by about 15% and take effect pretty much instantly.

    I am here to eliminate guesses, apparently. :)

  9. btmspox
    November, 23rd 2012 at 11:54 pm

    everydns fail.

  10. precipice
    November, 24th 2012 at 12:27 am

    Hmm. The spike is a little lower than the next one, though. Looks like the bars are five minute intervals. Maybe a restart that took less than about 5-6 seconds could happen there (assuming the first lowered bar took all of the downtime).

  11. • glub •
    November, 24th 2012 at 1:16 am

    Reduced traffic led to a release of RAM to the OS, which led to an increase in disk caching, which led to lower CPU?

  12. jspaw
    November, 24th 2012 at 1:35 am

    Again: not traffic related. :) Apache hits would have dropped proportionally. I’ll give the answer later on today… :)

  13. dathan
    November, 24th 2012 at 1:39 am

    php5 and or more servers on the front end and or more resources on the backend to reduce front-end wait-io

  14. ohesscue
    November, 24th 2012 at 1:50 am

    My guess is that it’s a new release of the Flickr codebase. From the graph, it looks like the drop-off at 8p is roughly proportional to an increase that begins around 3p. From the apache hits graph, the requests were roughly constant over the period shown. So, that sounds to me like code was pushed around 3 that was, perhaps, less efficient than it could have been — and that a fix was pushed around 8p.

    Hedging my bets, there was also a new stable release of php5 about a month ago — perhaps it contained some performance improvements…


  15. getluky
    November, 24th 2012 at 1:52 am

    Hard to tell without knowing whether this is from one specific box, an average of all boxes worldwide, an average of boxes in a colo, etc. It would be easier to tell if it was the result of a load distribution change, a misconfiguration, or a code change. :)

  16. rcrowley
    November, 24th 2012 at 2:36 am

    This looks like the graph for a single host, if I remember correctly. They push code like every 9 seconds so I doubt a bad change would last that long in production.

    I like the PHP5 theory but I don’t want to go slogging through changelogs to prove/disprove it.

    Did you clean up init_config? :)

  17. treebjen
    November, 24th 2012 at 3:05 am

    Ugh — I keep clicking "here" and it just goes HERE.

  18. rcrowley
    November, 24th 2012 at 4:09 am

    9:15am is early for Flickr, isn’t it?

  19. getluky
    November, 24th 2012 at 4:22 am

    Hmm, this is a longshot, but it looks like there are a couple more pct points of cpu_wait directly following the drop. Perhaps developers added caching for something that was slightly cpu intensive, and the drop indicates the point at which the caching change was deployed. Then the webserver may have gone into cpu_wait while the i/o to store/fetch the new cached copy might have occurred.

    A deployment is probably also consistent with the steep one-pixel-width drop in the other hint – while the codebase is being deployed, i’m betting that you guys rotate the webserver off the vip temporarily during the code switch.

    If this was consistent across all webservers, and the pattern followed the type of peak cpu shown before the drop, then that would look like a more solid case. But it’d be easier just to ask the developers. :)

  20. jspaw
    November, 24th 2012 at 5:20 am

    ANSWER IS: php5. :)

  21. getluky
    November, 24th 2012 at 6:05 am

    Aww, I feel like a tool now. :p That was fun though, can you make it a regular game?

  22. precipice
    November, 24th 2012 at 6:39 am

    Regular game! Regular tool! Uh, I mean…

    And yeah, @rcrowley++. 9:15a upgrades? Go to sleep! 😉

  23. dathan
    November, 24th 2012 at 7:21 am

    You can significantly reduce your memory footprint if Flickr started to use Objects, but that would make Cal cry.

  24. jspaw
    November, 24th 2012 at 8:02 am

    dathan: if memory was an issue, we’ll consider it. :)

    thanks all for playing! :)

  25. smin
    November, 24th 2012 at 8:22 am

    What do I win? :)

  26. shanan
    November, 24th 2012 at 8:30 am

    When can we look forward to php6?

  27. CáH2009
    November, 24th 2012 at 9:01 am

    Di =]

  28. gemedj89
    November, 24th 2012 at 9:40 am


  29. Hari
    November, 24th 2012 at 9:55 am

    that depends on what webserver you want to use.
    it ur using tomcat it is as simple as

    next > next > next > finish

Leave a Reply

You must be logged in to post a comment.

WP Login