Understanding storage for OpenShift Elasticsearch

Recommended Storage Type for OpenShift Elasticsearch:

• Only block storage is recommended with OpenShift Elasticsearch.
• Use of NFS type (including Gluster or samba or CIFS) storage is not supported as Lucene relies on file system behavior that NFS does not supply. Data corruption and other problems can occur with NFS type storage.

How much storage is required for OpenShift Elasticsearch ?

• This totally depends on the logs generated by workloads running in the cluster. In ideal scenarios, it is recommended to allocated some storage to Elasticsearch to observe the storage utilized by Elasticsearch over certain period of time (say 24 hours). Based on that, the storage allocated to Elasticsearch can be expanded as per the requirements of log retention.
• Furthermore, once Elasticsearch exceeds low disk watermark (i-e. 85%), it stops allocation shards to elasticsearch nodes and marks the indices as read-only once the storage utilization crosses 90%. Also, it will be better to look forward for capacity planning as well. For eg. cluster expansion (horizontal scaling) or scaling of application workload will generate more logs as well. Keeping that in consideration, it will be better to have some extra storage in buffer and then allocate the storage size according. Storage size will vary from cluster to cluster depending on the workload running in the cluster and the amount of logs generated by that workload.