Failing webbit tests - holding up release

[aslak hellesoy]

Hi everyone,

I recently set up Travis to build Webbit on various JDKs:

https://travis-ci.org/webbit/webbit

As you can see there are a few tests that are failing on some JDKs. I don't want to push a new release before tests are passing.

I don't have all the needed JDKs to reproduce this locally, and it's a little tricky for me to install them. 

My personal laptop is OS X which is a disaster when it comes to installing and switching between Java versions. 

My work box is Ubuntu, but I'm a little worried installing other JDKs there. 

Does anyone have experience installing/switching between multiple JDKs on OS X/Ubuntu?

Help (bugfixes) and suggestions are welcome!

Aslak

[Kushal Pisavadia]

Is it worth building out vagrant boxes that point to the webbit source? That way we can arbitrarily run the equivalent JVMs and spin up boxes locally for each JVM type.

Kind Regards,

Kushal Pisavadia
E: kus...@gmail.com

--
You received this message because you are subscribed to the Google Groups "Webbit" group.
To unsubscribe from this group and stop receiving emails from it, send an email to webbit+un...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Nathan Mische]

I use Vagrant[1] + Chef[2] to build lightweight, disposable test VMs. I generally use Ubuntu as that is the default distro provided by the Vagrant project.

[1] http://www.vagrantup.com/
[2] http://www.opscode.com/chef/

--Nathan

--

[peter hausel]

Hi, I can try to look into why the tests failing on JDK6, part of it it's still the cookie parsing issue[1] which is easy to fix. I am surprised travis is failing on oracle 7 though, since the test suite is passing on my local laptop (os x, oracle jdk, 1.7.0_10). Maybe the issue is related to version differences (1.7.0_10 vs 1.7.0_17)?  

It also looks like the test that is failing on both oracle7 and openjdk6 is ReconnectingWebSocketClientTest 

[1] https://travis-ci.org/webbit/webbit/jobs/6446225#L263

[peter hausel]

Hi Aslak,

here is my attempt to fix the tests:

https://github.com/webbit/webbit/pull/116

assuming this fixes Travis, could you please merge my pending PR-s before the release?

Thanks,

Peter

[aslak hellesoy]

On Sun, Apr 21, 2013 at 2:05 PM, peter hausel <peter....@gmail.com> wrote:

Hi Aslak,

here is my attempt to fix the tests:

https://github.com/webbit/webbit/pull/116

assuming this fixes Travis, could you please merge my pending PR-s before the release?

Thanks Peter.

It didn't fix Travis, and I have now concluded that tests are failing because Netty 3.6.3 leaks like a sieve. The next release will use Netty 3.5.3. I'll do my best to merge the remaining issues and cut a release this week, but no promises.\

https://github.com/netty/netty/issues/1301

https://github.com/webbit/webbit/commit/cebfe9a175728a6be1c77c7e78ed6f621569c7df

Aslak

[peter hausel]

I can confirm that netty 3.5.7[1] introduced this problem. Until 3.5.7 our test suite generated about 100 threads, after upgrading to 3.5.7 and above the same number goes up to 700 (most of them are New I/O worker XXX threads). (Apparently I did not see the thread increase locally due to my 16GB RAM, oh well.)

[1]https://github.com/netty/netty/issues?milestone=26&state=closed

[peter hausel]

Now that it looks like the problematic netty version is 3.5.7, would not make sense to upgrade to 3.5.5[1]  at least?

from 3.5.3 to 3.5.5 there were quite a few important fixes:

- fix for connection leaks in SSL handler (when throwing exceptions)

- cookie fixes

- fix for potential connection leaks in HttpContentEncoder (which is used by webbit via HttpContentCompressor)

- performance degradation in http 1.1[2]

Other than upgrading the libraries there would be only a minor change (ie adding `ssl.setCloseOnSSLException(true);` to the pipeline logic) 

[1] 3.5.6 is not recommended, so that leaves 3.5.5, here is the release info: http://netty.io/news/2012/08/22/3-5-5-final.html
[2] https://github.com/netty/netty/issues/476

I can put together a PR if needed. Thoughts? 

Cheers,

Peter

[Aslak Hellesøy]

On Wednesday, 24 April 2013 at 14:42, peter hausel wrote:

Now that it looks like the problematic netty version is 3.5.7, would not make sense to upgrade to 3.5.5[1]  at least?

from 3.5.3 to 3.5.5 there were quite a few important fixes:

- fix for connection leaks in SSL handler (when throwing exceptions)

- cookie fixes

- fix for potential connection leaks in HttpContentEncoder (which is used by webbit via HttpContentCompressor)

- performance degradation in http 1.1[2]

Other than upgrading the libraries there would be only a minor change (ie adding `ssl.setCloseOnSSLException(true);` to the pipeline logic) 

[1] 3.5.6 is not recommended, so that leaves 3.5.5, here is the release info: http://netty.io/news/2012/08/22/3-5-5-final.html
[2] https://github.com/netty/netty/issues/476

I can put together a PR if needed. Thoughts? 

Please do - I think that makes sense.

Aslak

[peter hausel]

On Wednesday, April 24, 2013 9:52:13 AM UTC-4, Aslak Hellesøy wrote:

On Wednesday, 24 April 2013 at 14:42, peter hausel wrote:

Now that it looks like the problematic netty version is 3.5.7, would not make sense to upgrade to 3.5.5[1]  at least?

from 3.5.3 to 3.5.5 there were quite a few important fixes:

- fix for connection leaks in SSL handler (when throwing exceptions)

- cookie fixes

- fix for potential connection leaks in HttpContentEncoder (which is used by webbit via HttpContentCompressor)

- performance degradation in http 1.1[2]

Other than upgrading the libraries there would be only a minor change (ie adding `ssl.setCloseOnSSLException(true);` to the pipeline logic) 

[1] 3.5.6 is not recommended, so that leaves 3.5.5, here is the release info: http://netty.io/news/2012/08/22/3-5-5-final.html
[2] https://github.com/netty/netty/issues/476

I can put together a PR if needed. Thoughts? 

Please do - I think that makes sense.

Great  - I will put together something today.

[peter hausel]

here you go:

https://github.com/webbit/webbit/pull/118

Thanks,

Peter