Hacker News new | past | comments | ask | show | jobs | submit login

Kinda feel like a jerk on this one. Wasn't really expecting to be able to nuke it so easily.

    ab -c 30 -n 3000 http://www.blinklink.me/b/4XIxfoeF4C0

    Concurrency Level:      30
    Time taken for tests:   93.802 seconds
    Complete requests:      3000
    Failed requests:        2137
     (Connect: 0, Receive: 0, Length: 2137, Exceptions: 0)
    Write errors:           0
    Requests per second:    31.98 [#/sec] (mean)
    Time per request:       938.024 [ms] (mean)
    Time per request:       31.267 [ms] (mean, across all     concurrent requests)
    Transfer rate:          182.47 [Kbytes/sec] received

    Connection Times (ms)
                  min  mean[+/-sd] median   max
    Connect:      116  165 165.9    122    1276
    Processing:   137  771 624.6    625    6827
    Waiting:      132  733 595.7    601    6823
    Total:        254  936 644.7    789    7062
Perhaps should have "view" throttling per IP. Quite a few mechanisms could have solved my abuse.



Are you in the habit of running load tests on other people's websites without warning? Remind me to never tell you when I'm working on something.


In this specific case it's not exactly the same as running a generic load test against some website, given the advertised feature of this specific website.

You're intentionally making it too generic to portray the guy as an asshole, when that's not how things went down. Very politician-like. Yuck.


I'm not condoning this, but if your website can be taken down by one machine with ab, you're doing it wrong.


Funny, I did the same thing before I read this. I think given it's back up that someone might be bumping up the number of views.

I'm betting that version 2 will solve this problem.

(Edit: I added another 1000 views by tweeting, so it didn't affect it permanently)


What is ab?


Don't use it for actual benchmarking though, it has lots of known problems. Use something like: https://github.com/wg/wrk


What sort of issues does ab have?


Well the best thing would be to try it yourself. I have found:

   * ab has more results variation between runs
   * ab will almost always report lower performance than wrk
   * If you have two implementations being benchmarked, A and B and B is always faster than A. wrk will report a greater degree of performance separation between A and B.
These results are less noticeable the lower performance the site being benchmarked is.


ab takes more CPU time than a decent web server too.


wrk looks interesting. Have used http://tsung.erlang-projects.org/


How does it compare to Siege?


Thanks for sharing!


The difference between benchmarking and DoS is intent. It could be taken the wrong way and now it's a felony.

"Three Felonies a Day" via https://mailman.stanford.edu/pipermail/liberationtech/2013-J...


apache bench

It lets you simulate hitting your web server to benchmark the speed. You can do cool things like test concurrent and sequential connections.


I didn't know either. Per "man ab":

> ab - Apache HTTP server benchmarking tool


Without having to open the man page:

    $ whatis ab
    ab (1)               - Apache HTTP server benchmarking tool
Or with bsd-games installed:

    $ wtf ab
    ab: ab (1)               - Apache HTTP server benchmarking tool


Is there a similar tool to see what a certain option in a command does, without opening the man page?


If you just type the command in you should see a list of options and what they do.


Probably best not to try this on any systems you care about if the command is completely unknown. killall on Solaris might have some unintended consequences.


You probably know, but try -h or --help behind the command.


apache bench


A friend said it came back "Retry later"




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: