auto build r3327: 2 of 7 builds failed - 2 regression tests failed

[Erik Eliasson]

More general comments on the auto-build emails:

About the two regression tests that fails most of the time and 
makes the autobuild emails large:

> 2012_Session_inviteab_hangupab_packetloss/ NOT THE EXPECTED OUTPUT

The result of this test unfortunately varies quite a lot. It is because
the test runs at the same time as the test compiles, and the machine
usually has a quite high load (~4-6, dual Xeon server). The test has a
nice value of 0, while the compiles have niceness 16, but that does not
seem to fix it.

The test takes close to an hour to complete, so not running it the same
time as the compile would increase the time to get autobuild feedback.

A solution would be to remove the packet drop test cases that are
changing since since if they change all the time no one will look at the
result, and the reports are quite large.

>                   2022_session_valgrind/ NOT THE EXPECTED OUTPUT
It makes sence to use valgrind since it catches a lot of problems.
Unfortunately, the output seems to vary a bit, and we have to do some
work on the output to make it predictable, but we should have ok's on
this test as well.

--Erik


> _______________________________________________
> Minisip-devel mailing list
> Minisip-devel at minisip.org
> http://lists.minisip.org/mailman/listinfo/minisip-devel
-- 
Erik Eliasson <eliasson at it.kth.se>