[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [jetty-dev] 8.1.2.v20120308 - rare mixup of requests
|
Thomas Becker skrev:
Hi Per,
Hi
any luck with 8.1.7?
Well as I believe Ive already told you this problem does not reproduce
every time I run my 1hour+ test. But it shows up now and then - about
50% of the times I run this test. I have been running the test today on
8.1.7. Three times until now, and I havnt seen the problem on 8.1.7 yet.
This is a good sign, but I need to run it more times in order to be
close to 100% sure. My highscore :-) on 8.1.2 is running the test 7
times in a row without the problem showing up. Can I run 10 times on
8.1.7 Im ready to conclude that the problem is not in 8.1.7. It will
take a day or two more.
I've tried to reproduce this problem with the test case attached. The
test uses your huge request body and sends it to a jetty server
running a servlet that just verifies that the request body matches the
expected body. I run many clients in parallel and have each of them
send a single request to the server. That way I should be able to
reproduce the problem you have if jetty is at fault. I ran this test
against the lastest jetty-7 revision which is basically identical with
jetty-8 for this case.
Well I think you should be running it against 8.1.2. Thats where I see
the problem.
However I can't reproduce your issue.
Hmmm, why am I not surprised :-)
Please double check that what I'm doing is right and if you think it's
not or not sufficient to reproduce the issue you have, feel free to
modify the test as needed.
Will have a look at it tomorrow, tuesday, and maybe do a little work
myself in order to see if I can reproduce it in a fairly simple test, or
if I can do other stuff in order to collect more information about what
happens.
The test contains some commented code that reads from the request's
InputStream instead from the parameter map. I've tested both ways of
reading the request body without being able to reproduce any mixed up
request body.
Thanks for participating on such a weak foundation! But on the other
hand, it is this kind of hard errors we live to nail, right :-)
Cheers,
Thomas
Cheers, Per Steffensen
