mod_jk 1.2.0 / tomcats benchs&problems : Was : jk 1.2 freese/snap time in JTC ?

[GOMEZ Henri <hg...@slib.fr>]

>^^^ Are we having problem with 2.0? 
>
>Ones that count are the non-2xx responses, cause this should never
>happened.
>Sockets perhaps?

We have also problems (Failed requests) :

Tomcat 4.0.4 + Apache 1.3/2.0
Tomcat 3.3.1 + Apache 1.3/2.0

No problem with TCs using their own http connectors.

The Apaches servers setup was to have up to 250 clients.
The Tomcats server allowed up to 250 ajp threads.

That results are really bad, since they show us that under
heavy load mod_jk 1.2.0/Tomcat may miss some requests.

I continue my tests but comments and ideas are welcomed

---

BTW, I've got the bench for Resin 2.1.2 on same platform 
with Resin http connector (max 250 threads) :

Server Software:        Resin/2.1.2                                        
Server Hostname:        localhost
Server Port:            8280

Document Path:          /examples/basic/date.jsp
Document Length:        456 bytes

Concurrency Level:      127
Time taken for tests:   10.515 seconds
Complete requests:      10000
Failed requests:        0
Broken pipe errors:     0
Keep-Alive requests:    0
Total transferred:      5796369 bytes
HTML transferred:       4565016 bytes
Requests per second:    951.02 [#/sec] (mean)
Time per request:       133.54 [ms] (mean)
Time per request:       1.05 [ms] (mean, across all concurrent requests)
Transfer rate:          551.25 [Kbytes/sec] received

Connnection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0    36   35.8     30   359
Processing:    26    97  194.1     40  2894
Waiting:        7    96  194.2     39  2893
Total:         26   133  197.1     73  2906

Percentage of the requests served within a certain time (ms)
  50%     73
  66%     86
  75%    104
  80%    117
  90%    224
  95%    415
  98%    934
  99%   1194
 100%   2906 (last request)

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>