Hello Dremio Team,
Is it possible to have a shared Cloud Cache storage between nodes ?
What is the main difference between Cloud Cache and the disk volume used by each node ?
Thanks a lot.
@imar Cloud Cache are the actual representation of the Parquet data on disk, cloud cache will reside on the disk which resides on the disk volume
Hello @balaji.ramaswamy,
the question of this post was on the cloud cache specifically but more generally, what is the recommended setup regarding executor nodes volumes.
Indeed, we reused the Dremio chart for Kubernetes and it attributes a separate volume for each executor. This means that in the case of a technical issue on an executor (Eg. a Kubernetes worker went completly down) the parts of the Reflections that were on this executor are now gone and the corresponding Reflections are unavailable in Dremio.
I also see that the recommendation for HA requires a shared volume for Master nodes:
So is it possible to have a shared volume for executor nodes to avoid the problem above ?
Is it a working setup for Dremio ?
@allCag Ah ! You wanted for reflections, this is defined as part of the dist:// parameter in your dremio.conf and should be set to a data lake source like S3/HDFS/Azure Storage/GCS as when an executor goes down, the reflections are 100% available, see below documentation on this
1 Like