What recover-quorum and recovery-hosts are recommended to avoid data loss on member restarts when using persistence? (Doc ID 2142836.1)

Last updated on APRIL 03, 2017

Applies to:

Oracle Coherence - Version 12.2.1.0.0 and later
Information in this document applies to any platform.

Goal

On Coherence 12.2.1.0.0:

A customer setup a cluster with 3 nodes. Started three nodes with different active persistence directory and see that each member is persisting owned partitions in the respective directory. When node1 is shutdown, noticed that backup partitioned getting redistributed. When customer shutdown all nodes and tried to bring the Node 1 backup, then customer does not get any data.

Customer has configured 3 Coherence storage enabled nodes with active persistence and local disk storage and 1 Storage disabled node which are part of the coherence cluster.

Customer did a bulk Put to load the data into the distributed cache having a backup count of 1.


Q1. Why is cache size only "10951"? Why is the entire cache not recovered? Is it because customer is using the local disk storage? And only the partitions which are on the local disk are getting recovered? Also the cache size is not always "10951". It varies if there are multiple storage enabled nodes in the cluster. Sometimes, after the restart the cache size shows as 0. Other times as 2000 etc.. Suppose if customer loads all the 33000 entries into 1 storage node and if he stops and restarts the node, in that case, the size shows as 33000.

 

Q2: To recover the data, do all the members listed in the quorum-recovery need to be started? What if customer wants to use only few nodes and not all the nodes defined in the active persistence?


Q3. If customer has N nodes in a cluster before he does shutdown, Does customer always have to start all N nodes before customer starts the recovery? Does this not bring up a huge restriction to the scaling of cluster? Does this mean that customer will never be able to start half of the nodes and recover full ?

 
Q4: Does this mean that on a given host, all members should be configured to persist data under the same base folder? (for local disk option)
 

Solution

Sign In with your My Oracle Support account

Don't have a My Oracle Support account? Click to get started

My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms