argo-server goes OOM when serving large artifacts #4588
Labels
Comments
dcherman
added a commit
to dcherman/argo
that referenced
this issue
Nov 24, 2020
…facts When serving very large artifacts, first loading them into memory can potentially cause the pod to go OOM/crash depending on how much memory is available and what limits have been set. Rather than loading it into memory, we can serve files directly from disk. Fixes argoproj#4588 Signed-off-by: Daniel Herman <dherman@factset.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Summary
When downloading very large artifacts, the
argo-serverpod can go OOM since we load the entire file into memory before serving it to the user. Rather than loading the file into memory, we can just useio.Copyto serve the file from disk to the response directly.Diagnostics
What Kubernetes provider are you using?
k3d
What version of Argo Workflows are you running?
latest/edge
Message from the maintainers:
Impacted by this bug? Give it a 👍. We prioritise the issues with the most 👍.
The text was updated successfully, but these errors were encountered: