Replies: 3 comments 2 replies
-
Ubuntu runners have 7 GB (from your warning seems you've maxed out at 4?). If that's not enough, you can use a macOS runner instead which offers 14 GB. Also, consider setting a concurrency limit. https://github.com/typelevel/cats/blob/14286c9ef89a652fcff182bbd685f6caf930701e/build.sbt#L68 |
Beta Was this translation helpful? Give feedback.
-
Well we're using 4GB for the heap, which leaves 3GB for the entire rest (non-heap and the OS). Using anything more than 4GB for the heap makes the build OOM (on the kernel level) with a 100% certainty ;) We'll try the macos runners, thanks, though that's more of a work-around than a fix. I don't think compiling a bunch of modules using scala-native should require more than 4GB of heap :) Besides, it works for JVM (compiling all modules), JS and native-2, so there's clearly something wrong with native-3. |
Beta Was this translation helpful? Give feedback.
-
Btw. we already have concurrency restrictions enabled globally :) https://github.com/softwaremill/tapir/blob/master/build.sbt#L31 |
Beta Was this translation helpful? Give feedback.
-
Since some time, most of the tapir builds fail with
Error: Process completed with exit code 137.
when building Scala 3 native artifacts.This is unfortunately not deterministic, as about 20% or so builds pass (so retrying the build helps). Also, Scala 2 native builds just fine.
I'm not sure how to debug this - there are some warnings on the console:
But we're already allocating the maximum memory available on a github actions macine.
An example failing build is here: https://github.com/softwaremill/tapir/actions/runs/4751063533/jobs/8439821428
There's a couple of modules built for native, but not that many (core, tests, some integrations, one server interpreter).
Some guidance as to how to debug this would be greatly appreciated :)
Beta Was this translation helpful? Give feedback.
All reactions