Qdrant on kubernets

[mm3906078]

Hi everyone,

I hope you're having a great day!

I have some questions about Kubernetes and Qdrant after reading the official Helm chart documentation.

• In the Helm chart values.yaml, the local-path storage class name is used. What other storage options are best suited for Qdrant on Kubernetes?

• We have 5 Kubernetes nodes and 10 Qdrant containers (2 containers on each host). If we use local-path storage and one of the nodes in the Kubernetes cluster, for example, node 4, becomes unavailable, what happens to the Qdrant cluster? As far as I understand, we are using StatefulSet in the Helm chart, and with this configuration, no new container will be spawned because the data is stored on node 4. What should we do if node 4 is completely lost and we have no data from it? How do you think we should manage this situation?

• Is there a need to save data? What happens if one of the containers is down temporarily and then comes back up? Will it sync data from all other nodes? How does the presence or absence of data in the container affect the sync process?

• If we need to scale from 4 Kubernetes nodes to 5 Kubernetes nodes, how should we scale up our Qdrant collection? I think we can easily add a new node and then increase replicas and join the cluster, but can we safely change the shard_number and replication_factor for our collection?