This is an archive of the discontinued LLVM Phabricator instance.

[zorg] Add linaro-d05-* builders
ClosedPublic

Authored by maxim-kuvyrkov on Nov 29 2017, 8:24 AM.

Details

Summary

... which are hosted by 64-core D05 machine.
Remove explicit parallelism settings and rely on ninja
to use correct parallelism values for 8-core APMs and
64-core D05s.

Diff Detail

Repository
rL LLVM

Event Timeline

maxim-kuvyrkov created this revision.Nov 29 2017, 8:24 AM
maxim-kuvyrkov planned changes to this revision.Nov 29 2017, 10:44 AM
maxim-kuvyrkov added inline comments.
buildbot/osuosl/master/config/builders.py
308 ↗(On Diff #124758)

After a bit more testing, it appears that lit testsuite won't be parallelized without --threads=N and --build-threads=N settings. I'll rework this patch and resubmit.

In this version jobs= settings are removed (so that they are picked up from slave setting). testsuite_flags are left with 8-thread parallelism for now, and we are going to switch them to 64 once APM hardware is decommissioned.

The overall plan is to

  1. Continue running APM hardware on the normal buildmaster.
  2. Connect D05 slaves to silent buildmaster and monitor them.
  3. Swap hardware: connect D05 slaves to normal buildmaster and connect APM slaves to silent buildmaster.
  4. Switch to 64-thread parallelism for the testsuite and remove APM slaves.
  5. Decommission APMs.
rovka edited edge metadata.Dec 1 2017, 2:45 AM

This looks fine, but I think it would be better if the slave names weren't tied to what is currently running on them, but rather to the machine configurations. E.g. if there's no difference in the environment of linaro-d05-01-quick and linaro-d05-global-isel, they should just be linaro-d05-01 and linaro-d05-02. But for instance if the machines running the libcxx bots need something special, then they can be linaro-d05-libcxx-01 and linaro-d05-libcxx-02. That way, if we want to repurpose a slave to run a different builder, it can keep its name without looking awkward, and it's also easy for us to tell which machines have different environments. What do you think?

@rovka , slaves linaro-d05-01-* will run in different containers on the same machine (linaro-d05-01). In this context there is no need to repurpose a slave.
I plan to add a second machine linaro-d05-02 that would host the same set of slaves/builders for redundancy.

rovka accepted this revision.Dec 4 2017, 1:13 AM

@rovka , slaves linaro-d05-01-* will run in different containers on the same machine (linaro-d05-01). In this context there is no need to repurpose a slave.
I plan to add a second machine linaro-d05-02 that would host the same set of slaves/builders for redundancy.

True, but that's an implementation detail on our side, not something that is visible in zorg.
Anyway, that's just a nitpick, LGTM.

This revision is now accepted and ready to land.Dec 4 2017, 1:13 AM
This revision was automatically updated to reflect the committed changes.