You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@jmeter.apache.org by Felix Schumacher <fe...@internetallee.de> on 2016/11/23 20:57:06 UTC

broken builds on travis ci

Hi all,

with pr #234 I succeeded to get the builds on travis ci back to green. I 
had to change three things on the way:

  1. Start the rmi server in the batch tests with localhost as host 
name, since travis ci runs with the hostname bound to loopback and 
JMeter refuses to start in such a constellation (you might have a look 
at https://bugs.launchpad.net/ubuntu/+source/jakarta-jmeter/+bug/589042, 
where such a problem has been discussed before, alas without travis ci :)

  2. Revert to ant 1.8.2 compatibility as that is the version running on 
travis. My solution to replace "if:set" with "condition" feels a bit clumsy.

  3. Disable two tests HTTP_HTTPS and BUG52310, which are either 
problematic because of some strange dns problem or maybe some firewall 
issue.

I would like to apply these changes to trunk. What do you think?

Felix


Re: broken builds on travis ci

Posted by Felix Schumacher <fe...@internetallee.de>.
Am 23.11.2016 um 22:48 schrieb Philippe Mouawad:
> On Wed, Nov 23, 2016 at 9:57 PM, Felix Schumacher <
> felix.schumacher@internetallee.de> wrote:
>
>> Hi all,
>>
>> with pr #234 I succeeded to get the builds on travis ci back to green. I
>> had to change three things on the way:
>>
> Great
>
>>   1. Start the rmi server in the batch tests with localhost as host name,
>> since travis ci runs with the hostname bound to loopback and JMeter refuses
>> to start in such a constellation (you might have a look at
>> https://bugs.launchpad.net/ubuntu/+source/jakarta-jmeter/+bug/589042,
>> where such a problem has been discussed before, alas without travis ci :)
>>
>>   2. Revert to ant 1.8.2 compatibility as that is the version running on
>> travis. My solution to replace "if:set" with "condition" feels a bit clumsy.
>>
> This may be a problem with Java 8 . I remember I had to upgrade when using
> Java8.
I will try to move to java 8 next, but I haven't seen any problems yet.

>
>>   3. Disable two tests HTTP_HTTPS and BUG52310, which are either
>> problematic because of some strange dns problem or maybe some firewall
>> issue.
>>
> What test exactly.
> 52310 may not be too problematic , but the one with HTTP/ HTTPS is more
> important, I remember one of my commits had broken an important part.
The tests are only disabled on travis, not on jenkins or buildbot and 
definitely not on your local builds.

I would like to re-enable them, but I have no idea, why they fail in the 
first place.

Regards,
  Felix
>
>> I would like to apply these changes to trunk. What do you think?
>>
> +1
>
>> Felix
>>
>>
>


Re: broken builds on travis ci

Posted by Philippe Mouawad <ph...@gmail.com>.
On Wed, Nov 23, 2016 at 9:57 PM, Felix Schumacher <
felix.schumacher@internetallee.de> wrote:

> Hi all,
>
> with pr #234 I succeeded to get the builds on travis ci back to green. I
> had to change three things on the way:
>
Great

>
>  1. Start the rmi server in the batch tests with localhost as host name,
> since travis ci runs with the hostname bound to loopback and JMeter refuses
> to start in such a constellation (you might have a look at
> https://bugs.launchpad.net/ubuntu/+source/jakarta-jmeter/+bug/589042,
> where such a problem has been discussed before, alas without travis ci :)
>
>  2. Revert to ant 1.8.2 compatibility as that is the version running on
> travis. My solution to replace "if:set" with "condition" feels a bit clumsy.
>
This may be a problem with Java 8 . I remember I had to upgrade when using
Java8.

>
>  3. Disable two tests HTTP_HTTPS and BUG52310, which are either
> problematic because of some strange dns problem or maybe some firewall
> issue.
>

What test exactly.
52310 may not be too problematic , but the one with HTTP/ HTTPS is more
important, I remember one of my commits had broken an important part.

>
> I would like to apply these changes to trunk. What do you think?
>
+1

>
> Felix
>
>


-- 
Cordialement.
Philippe Mouawad.