wsdl generation not thread safe?

[Arjen Poutsma]

Damn, I fixed it for Jetty, but didn't check Tomcat.

I've reopened the issue, and will fix it for 1.0.2. Hopefully, for good this time.

[Arjen Poutsma]

I cannot reproduce this with either Tomcat 6.0.14 and 5.5.23. I used Apache Bench (ab) to request the WDSL of the echo sample app a 1000 times, using 100 concurrent threads. From that test, I get the following:

>ab -n 1000 -c 100 http://192.168.0.2:8080/echo/echo.wsdl
This is ApacheBench, Version 1.3d <$Revision: 1.73 $> apache-1.3
Copyright (c) 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Copyright (c) 1998-2002 The Apache Software Foundation, http://www.apache.org/

Benchmarking 192.168.0.2 (be patient)
Completed 100 requests
Completed 200 requests
Completed 300 requests
Completed 400 requests
Completed 500 requests
Completed 600 requests
Completed 700 requests
Completed 800 requests
Completed 900 requests
Finished 1000 requests
Server Software:        Apache-Coyote/1.1                                  
Server Hostname:        192.168.0.2
Server Port:            8080

Document Path:          /echo/echo.wsdl
Document Length:        2060 bytes

Concurrency Level:      100
Time taken for tests:   2.647 seconds
Complete requests:      1000
Failed requests:        0
Broken pipe errors:     0
Total transferred:      2188186 bytes
HTML transferred:       2062060 bytes
Requests per second:    377.79 [#/sec] (mean)
Time per request:       264.70 [ms] (mean)
Time per request:       2.65 [ms] (mean, across all concurrent requests)
Transfer rate:          826.67 [Kbytes/sec] received

[kmoore4now]

Thanks Arjen,

I'll give it another try and double check my tests. If they still fail, I'll give it a try with ApaceBench.

Ken

[kmoore4now]

Everything looks good. The last test failure was my fault. I hadn't cleaned everything up before testing the update.

Thank you very much for looking into this, and for resolving the problem so quickly.

[pradeepchennai]

Hi,


I had hit the same issue.

The exact scenario to reproduce it is: Run a JMeter HTTP GET request test (NOT webservices test) for the wsdl url with thread size (users)=10; ramp-up period=0; loop-count=50

As already mentioned in this thread, when there are too many concurrent wsdl file requests, the spring app will die since the operations performed during wsdl file generation are not thread-safe.

The solution is to either cache the wsdl in the client end or better, to do away with the dynamic wsdl generation. The latter one is better if you consider the fact that anyone in your nw could use a jmeter test to down your services :-D

I would dynamically generate the wsdl file for code or environment change, and serve this file from the services. The spring xml config would change like:

<bean id="myservice" class="org.springframework.ws.wsdl.wsdl11.SimpleWsdl11Definition">
        <constructor-arg value="/WEB-INF/myservice.wsdl.wsdl"/>
</bean>

Hope that helps..