This does not appear to have the debug logging enabled (you have to restart the Dremio coordinator/master after enabling the debug. That being said have you tried rebuilding the reflection but disabling the reflection, saving this change, and then re-enabling?
yes, i did try disable/enable/delete-all/recreate reflection options.
just accessing vd uses reflections, but if the reflection is part of another vd’s definitions substitution error occurs, I got the logs again from both master&executor
The problem with the logging is apparently the appender, as there is no “text” in the k8s deployment. You should be able to see the debug logging with:
The query ran on “2020-06-17 00:37:43” but the logs are only till “2020-06-16 19:40:31,131”
The error you are seeing may be a generic error. If you see the reflection is was matched, which means it had all the information it needs to accelerate the query. The reason why it did not choose could be because of 2 reasons
A) Dremio’s planner estimated cost of using reflection to be more expensive than going back to PDS, to confirm this, turn on “planner.verbose_profile” and rerun the query and send us the profile
B) There was an exception during substitution, for this we need the server.log right after you run the query