Repository navigation
resource contention on macOS #7356
FlorinAndrei
started this conversation in
General
Replies: 1 comment
|
Hey @FlorinAndrei , Take a look at the following docs. It should help you investigate this: Once you have some more data and we have something actionable, we can file the bug report on that. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Context:
I have a live repo with stacks, I run the
applycommand at a top of a folder. 33 stack files with dependencies, several units each, invoked from a catalog repo (direct from disk, not via a remote repo URL). All is in place on disk, no pulling of modules, etc - this is a repeat run.Initially, the
applycommand prints out a bunch ofGenerating unit ...lines, then waits for a couple minutes or so. More than half of the CPU cores are used at this point.There is very strong contention for resources during this time. All other apps become very slow. The machine is barely usable, and this is a powerful laptop that can run 27B LLMs at 20+ tok/sec. The browser is slow. Apps are slow to start or stop.
pip installgot stuck for a very long time.And yet there are CPU cores still unused. I believe the contention is over the performance cores. But there are cores sitting there unused, and they should be used, even if they are efficiency cores.
I do not see a lot of disk access during this time. It's not disk I/O that's contended. I run the Stats app in the menu bar, with CPU, GPU, RAM, SSD, and NET metrics.
I'm not saying this is a terragrunt bug. It might be terragrunt. It might be macOS. It might be an interaction issue.
But a powerful laptop should not slow down while
terragrunt stack run -- applyis waiting for some intermediate phase to complete.I'm not sure even where to begin investigating this. I just know the issue exists.
All reactions