You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
internal/ci: split Linux runners into "small" and "large"
Cheap workflows like "tip triggers" and "trybot dispatch" do not need
as much CPU and memory as expensive trybot workflows, so those always
use the small runner.
The trybot runner still uses "linuxMachine", which now defaults to the
small runner, but can be opted into the large runner for those
repositories which really need it. The majority of them should be
perfectly fine on the smaller machine with half as much CPU and memory.
The "small" size has been chosen in a relatively conservative way,
only being half as large as the "large" machine.
This is because even the public GitHub runners have four CPU cores
and 16 GiB of RAM, so going all the way down to 2 CPU cores and 4 GiB
might be too low.
Thankfully, now that the split is made here, and given that these
runner profiles have generic "small" and "large" names, we can later
resize either of them without having to re-configure all repos.
Signed-off-by: Daniel Martí <[email protected]>
Change-Id: Ib62b2ef79cd7a6d64413b426843abbfdb5627e2a
Reviewed-on: https://cue.gerrithub.io/c/cue-lang/cue/+/1218980
Unity-Result: CUE porcuepine <[email protected]>
Reviewed-by: Paul Jolly <[email protected]>
TryBot-Result: CUEcueckoo <[email protected]>
0 commit comments