On 08/19/2010 10:38 AM, Garrett D'Amore wrote:
On Thu, 2010-08-19 at 10:02 -0700, Erik Trimble wrote:
I happen to have a bunch of spare IBM e325/e326/e326m machines sitting
around. These are all old dual-socket Opteron 200-series machines.
Good, these are still serviceable.
In addition, I should have at least several UltraSPARC IIIi and IV
systems that are abandoned, and I should be able to resurrect them. I
*might* be able to scrounge up a T1000, but that's unlikely right now.
Maybe sometime soon.
For building Illumos or OS/Net the IIIi and IV systems are actually
faster.
And, of course, there's some decent storage that goes with this. Figure
about 1TB or so of fast SAN-based disk, and multi-TB of SATA archival.
Awesome!
I've been using all this stuff at one of my friend's small ISP/developer
shop for about 5 years now, and I'd be happy to set it up to give us
some sort of nightly build and automated test configuration. The sole
issue is that bandwidth in and out is limited, so I don't want to be
pushing big gobs of data back and forth. Running an Hg mirror internally
is the best thing for me, and I would *not* be able to serve out built
binaries to the public. But it would be more than fine to do
build/testing and then upload the results (including build binaries) to
some machine under the IllumOS umbrella.
Any interest in having me start working on an automated build/test
system?
Yes, me for one!
Nexenta has a system called NABAT, but its highly specific to
debian/apt, so we need something more generic.
Upcoming NexentaCore 3.9 is going to be Illumos based and *is* going to
be nightly tested/released by NABAT. While I agree, we need a generic
framework I would suspect that we only need to focus on testing Illumos
components and not its permutations or compositions. Good example would
be ztest utility..
For instance, NABAT is also testing NFS/CIFS/iSCSI against tens of
clients such us Windows/Linux/ESX/Xen/Hyper-V/etc to ensure its
interoperability. And at distro level it is easier and Illumos is not a
distro.
The initial thing we need is nightly builds. The way I see this:
each build system maintains a "clean" clone, via hg pull -u from
http://hg.illumos.org/illumos-gate
For nightly builds, a full clone (can be done simply by zfs clones to
make it really fast) of the local "clean" clone is done, and a build is
run in that workspace (full build, including shadow, lint, etc.)
I want the mail_msg files to be e-mailed to a list, which I guess we
will create for the purpose. Gatekeepers (me at the moment, until I
figure out how to sell the job to richlowe :-) will monitor that mail
and thump developers that break the build.
For test, having another system (maybe a VM?) that could onu the
resulting nightly builds and then boot, and run some basic battery of
tests, would be cool.
Speaking of which, do we happen to have any of the automated tests from
the OpenSolaris distro?
Some of the tests are open crap^wsource. But not all of them.
They might be usable as a starting point though.
- Garrett
_______________________________________________
Discuss mailing list
[email protected]
http://lists.illumos.org/m/listinfo/discuss
_______________________________________________
Discuss mailing list
[email protected]
http://lists.illumos.org/m/listinfo/discuss
