fancy-leather-59976
08/24/2023, 1:15 PMcurrent limit set in the platform configuration
errors (one for CPU, one for memory). Yet in our Flyte code we don't see anything particular about those workflows that would be causing that problem.
(1) Is there a comprehensive list of the settings (in Flyte and in k8s) that influence what resources are available?
(2) Is there a way to use flyteadmin
or kubectl
to try to locate the element or layer that's setting the resources too conservatively?echoing-carpenter-92090
08/24/2023, 1:19 PMresources
section in the values file for memory
and cpu
.echoing-carpenter-92090
08/24/2023, 1:20 PMhelm get values flyte -n flyte
Your release and namespace may be different.
helm ls -A
to list all releases across all namespaces.fancy-leather-59976
08/24/2023, 1:21 PMflyte-core
.echoing-carpenter-92090
08/24/2023, 1:23 PMhelm upgrade ...
to apply new values to the resources limits/requests.
The tasks may also need to be configured for greater resources but I think you will get a different error if that's the issue.
@task(
limits=Resources(cpu="500m", mem="500Mi"),
requests=Resources(cpu="200m", mem="400Mi"),
)
def my_task() ->...