On 08/11/17 17:59, Jani Nikula wrote: > On Wed, 08 Nov 2017, Tomi Sarvela <tomi.p.sarvela@xxxxxxxxx> wrote: >> On 08/11/17 17:27, Jani Nikula wrote: >>> On Wed, 08 Nov 2017, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote: >>>> Quoting Jani Nikula (2017-11-08 13:06:39) >>>>> >>>>> Please use trybot (i.e. send patches to >>>>> intel-gfx-trybot@xxxxxxxxxxxxxxxxxxxxx) for hacks that aren't intended >>>>> for upstream. Then they can be prioritized appropriately. >>>> >>>> We don't have excess to the extended test run on trybot yet... >>> >>> Marta, Martin, Tomi - this. >>> >>> Though I believe Chris wants access not excess. ;) >> >> Sssooo... shard-runs for Trybot? Only on success, or also on warning as >> with intel-gfx and igt? Same as the rest. >> >> The very recent change to run also f-feedback warnings on shards, and >> fixed vetting handling, will cause much more shardruns. >> >> After KBL DMC update it could be taken to all the runs, instead of >> avoiding it because it's 20 minutes slower (and GLK is 30 minutes more >> slower) >> >> We'll see if there's enough time in a day. > > IIUC there currently just isn't. The problem is, people are now running > their hack patches on the regular CI to get full runs, and we can't even > prioritize the stuff aimed at upstream over hacks. Agreed. Developers are being nice to the system by pushing stuff to the trybot rather than intel-gfx, so we should provide the same service, albeit one with higher latency than intel-gfx@... Let's however wait to know the impact of the changes Tomi made today before increasing the workload even more (Run shards on WARNING BAT results, and fixing the suppressed tests/machines handling) :s As far as I can see, shards are not a problem right now though: https://intel-gfx-ci.01.org/cibuglog/metrics.html I propose waiting until next week before enabling shards on the trybot. Martin _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
