Bryce Kahle
Bryce Kahle
> My gut feeling would be to make this `TracePrefix` as you mentioned @lmb Can you clarify exactly what you want? I think I'm confused about which specific variant you...
@lmb updated to just provide the prefix
I think this is failing because https://github.com/DataDog/datadog-lambda-extension/blob/23357637625a986f53bdc218c43c9c52656156d5/scripts/Dockerfile.alpine.build#L10 is still using Go 1.20, which cannot parse `go 1.21.7` from the `go.mod`: `go.mod:3: invalid go version '1.21.7': must match format 1.23`
@paulcacheux I included the manager update as well. Feel free to revert if you don't want it.
Waiting on https://github.com/DataDog/datadog-agent/pull/22908
@usamasaqib do we still need this?
AFAIK the OOM killer uses Working Set Memory for decisions (at least in `k8s`/namespaces), which does include `mmap`-ed regions. Is this being taken into account?
> I can understand why it'd use anonymously-mmap'd memory, but file-backed `mmap`s should morally count as page cache only I don't have extensive knowledge of how it treats the different...
I suppose you can treat my comment more like: "I urge you to profile and test using some experiments, rather than relying upon assumed behavior"