[httperf] Delay after 302 responses
martin.arlitt at hp.com
Thu Apr 26 13:22:13 PDT 2007
I tried issuing a single request using httperf/0.8 to a server that issues a 302 response for a particular url, and found no problems.
If you could capture a full packet trace using tcpdump of the test you described, I will take a look and see if I can spot the root cause.
Also, what is the keep-alive timeout value on your server?
> -----Original Message-----
> From: httperf-bounces at napali.hpl.hp.com [mailto:httperf-
> bounces at napali.hpl.hp.com] On Behalf Of Claude Brisson
> Sent: Wednesday, April 25, 2007 4:04 PM
> To: httperf at napali.hpl.hp.com
> Subject: [httperf] Delay after 302 responses
> Hi there.
> I'm doing perf & load testing on a page that sends back a HTTP 302
> response code (redirection). No problem, it's not the redirected page I
> want to benchmark, but the cgi issuing the redirection. I'm not asking
> httperf to follow the redirection.
> But the connection stays open around ten seconds after the redirect has
> been received, which spoils my statistics.
> Here is a screenshot of my session:
> # httperf --server=... --num-con=1 --hog --uri=... --print-reply
> httperf --print-reply --hog --client=0/1 --server=... --port=80
> --uri=... --send-buffer=4096 --recv-buffer=16384
> --add-header='Authorization: Basic Ym94dGFsZTppcm9xdW9pcw==\n'
> --num-conns=1 --num-calls=1
> RH0:HTTP/1.1 302 Déplacé Temporairement
> RH0:Date: Wed, 25 Apr 2007 18:08:31 GMT
> RH0:Server: Apache/2.2.3 (Debian) DAV/2 SVN/1.4.2 mod_jk/1.2.18
> PHP/5.2.0-8+etch1 mod_ssl/2.2.3 OpenSSL/0.9.8c
> RH0:Set-Cookie: JSESSIONID=8F33EE5344FE480768791B062895CEF2; Path=/
> RH0:Location: http://...
> RH0:Content-Length: 0
> RH0:Content-Type: text/html; charset=ISO-8859-1
> <<<< HERE a big pause of around 10 seconds >>>>
> RS0: header 373 content 0 footer 0
> Maximum connect burst length: 0
> Total: connections 1 requests 1 replies 1 test-duration 18.553 s
> Connection rate: 0.1 conn/s (18553.4 ms/conn, <=1 concurrent
> Connection time [ms]: min 18553.4 avg 18553.4 max 18553.4 median 18553.5
> stddev 0.0
> Connection time [ms]: connect 4.5
> Connection length [replies/conn]: 1.000
> Request rate: 0.1 req/s (18553.4 ms/req)
> Request size [B]: 478.0
> Reply rate [replies/s]: min 0.0 avg 0.0 max 0.0 stddev 0.0 (3 samples)
> Reply time [ms]: response 3561.5 transfer 14987.4
> Reply size [B]: header 373.0 content 0.0 footer 0.0 (total 373.0)
> Reply status: 1xx=0 2xx=0 3xx=1 4xx=0 5xx=0
> CPU time [s]: user 3.08 system 12.76 (user 16.6% system 68.8% total
> Net I/O: 0.0 KB/s (0.0*10^6 bps)
> Errors: total 0 client-timo 0 socket-timo 0 connrefused 0 connreset 0
> Errors: fd-unavail 0 addrunavail 0 ftab-full 0 other 0
> Do you know where does this behaviour come from?
> According to the HTTP/1.1 specification, when the response does not
> contain any body, the response should be considered over at the first
> empty line (which we see in the print-reply). So I'm pretty sure that on
> the server side, everything is ok.
> Did someone else experiment the same behaviour with 30x redirection
> responses? Is it a known issue?
> # httperf -V
> httperf: httperf-0.8 compiled Mar 17 2006 without DEBUG without
> Debian Etch on a 686.
> ps: moderator: I reposted the topic bkoz I didn't use the right sender
> adress the first time. Please discard the awaiting mail since I cannot
> do it myself - http://linux.hpl.hp.com/cgi-bin/mailman is not responding
> (or muted by the HP proxy...).
> httperf mailing list
> httperf at linux.hpl.hp.com
More information about the httperf