Summary
The user is testing the flyte-binary v1.12.0 deployment and experiencing high memory usage that results in Kubernetes eviction while submitting the sleep_more_minutes workflow approximately 300 times. They suspect this issue may be linked to a specific GitHub problem and are looking for advice on how to investigate the memory usage. The user also downgraded to 1.11.0, but the memory behavior did not change, with high peaks observed in both versions. They find it interesting that this might be due to the blobstore cache in the single binary as its unified configuration.