New build machine in the cluster lldb-amd64-ninja-netbsd8


I'm in process of attaching new build machine lldb-amd64-ninja-netbsd8.

It will run prerelease of NetBSD/amd64 8.0 (as of today 7.99.70) with
the GNU toolchain.

Once the new one setup will be in operation, I will upgrade the old one
to 7.99.70 (& retain the same name), switch the to the staging cluster and turn on execution of tests.

Am I right that in order to turn tests I need to switch "runTest=False"
to "runTest=True" in buildbot/osuosl/master/config/

Once the setup for tests will stabilize I will turn them on in
lldb-amd64-ninja-netbsd8 in the main cluster.

The new build node is hosted by The NetBSD Foundation.


As you are using the "scripted" build factory, you'll also need to create a
test_cfg.json in your scripts folder on the build slave, which describes
the kind of tests you want to run:
For example, the linux build bot has this in the file:
    "test1": "local,clang-3.5,i386",
    "test2": "local,clang-3.5,x86_64",
    "test3": "local,totclang,i386",
    "test4": "local,totclang,x86_64",
    "test5": "local,/lldb-buildbot/bin/gcc-4.9.2/bin/gcc,i386",
    "test6": "local,/lldb-buildbot/bin/gcc-4.9.2/bin/gcc,x86_64"
but a single test line would probably be enough for you.
Then, the master will invoke a script "" with the test config
argument: (e.g., ./ local,clang-3.5,i386) and your script should run
the tests.

let me know if you run into problems,

Thank you, I will give it a try and let you know about the status.

lldb-amd64-ninja-netbsd8 [kernel ABI version 7.99.70] is up and running

lldb-amd64-ninja-netbsd7 has been upgraded to 7.99.70 and restarted

I switched the old machine to the staging buildfarm, set runTest=True in
buildbot/osuosl/master/config/ and added the following

$ cat ./build/build/test_cfg.json
    "test1": "local,gcc,x86_64",

As of now it does not attempt to run the regression tests. Perhaps we
need to restart buildmaster configuration? I will ask Galina for it.

Indeed, master restart is needed every time you update the zorg repository.


I've managed to configure lldb-amd64-ninja-netbsd7 to run the unit
tests. There has been observed a regression as the number of passing
ones dropped recently from 600+/1200+ to 200+. I've reproduced this also

I'm still using google scripts. I think we should drop "upload test
traces" on !Google platforms.

Long term I'm planning to prepare my own set of scripts customized for
NetBSD. I'm evaluation inclusion LLVM and Clang tests in execution to
maximize usage of the new buildbot. Perhaps I will be able to spare my
time on it around September.

For now I will try to mark failing ones with expected failure and file
Bugzilla bugs for them. I'm going to turn on tests on

Yes, those classes need de-googlyfying. If you're going to touch them,
you can also delete the "update NDK" step totally -- it was a bad
idea, and we're not using it.