[wildfly-dev] The art, science, and torture of debugging test suite hangs

[Richard Opalka]

Welcome to the WildFly debugging hell David.

The situation is even more complicated than it seems at first sight.

There are tests (in domain test suite for example) when

there are started multiple host controllers

and servers there, each of them binding

to different port. The 'new build profile' proposal should also

address different debugging ports for

all those 'domain controller', 'host controller' and 'server' processes.


Richard


On 04/14/2017 04:56 PM, David M. Lloyd wrote:
> It's really a pain when you are developing a multifaceted change to the
> application server and it hangs.  Sometimes, if you're lucky, you can
> stick jconsole on there and figure out at least a general idea of what
> is going on, or reproduce the problem by running a test, a test class,
> or a test module in isolation with debugging turned on.  Sometimes
> you're not so lucky, or sometimes the hang happens in the manualmode
> test suite where you have to attach a debugger 9,000 times only to find
> out that the hang doesn't happen when you do that.
>
> I think we should introduce a new build profile which activates the
> remote debugging port, but with suspend=n.  This would allow entire full
> test suite runs to be done while allowing debugger to be attached for
> exploratory surgery in the event that something goes wrong.  While it's
> possible that having debugging activated might actually prevent the hang
> you care about, at least there's a chance that something can be done.
>