Flyte enables production-grade orchestration for machine learning workflows and data processing created to accelerate local workflows to production.

Flyte

image.png

what is the way to find out why certain map task fails?
also is the only way to check what input it ran against is to switch between map_exec and input tabs in deck?

do you have a <https://docs.flyte.org/en/latest/flytesnacks/examples/productionizing/configure_logging_links.html#configure-logging|logging setup> for your cluster? we have a stackdriver logging and each map task elemnets will generate separate log links for you so you can probe why individual task has failed

i do and i see logiles in cloudwatch but there is nothing in console

also when tasks error out - there is nothing in logiles

i see my own application logging in logs but when error happens

this text in the middle - it is nowhere in the logs

i see, i think map task might need different wiring

Might want to look into logging for maptasks, they <https://github.com/flyteorg/flyte/issues/3291|require separate configuration>. However, ArrayNode uses the same logging as regular Flyte tasks.

is there a sample config I can find somewhere?

<https://flyte-org.slack.com/archives/CP2HDHKE1/p1674515380828799>

Yeah, so the logs should be something like:
```plugins:
  k8s-array:
    logs:
      kubernetes-enabled: true
      kubernetes-url: "<http://localhost:30082>"      
  logs:
    kubernetes-enabled: true
    kubernetes-url: "<http://localhost:30082>"```
where the log configuration for the `k8s-array` element should be identical to the regular `logs` configuration.

thx - should this sit on the same level as k8s config? i.e. configuration &gt; inline&gt; plugins?

and i dont have regular logs: i have it under logging:.....