4.3.1 Dynamic Value Generation for server.id
- Application instance will get a unique value allocated for server.id internally.
- On a graceful shut down of an application instance, server.id value allocated for this instance will return to pool.
- In case of a crashed instance, the server.id value will not go back to pool immediately. A monitoring job will check and recover this value back to pool. This will regulate the continuous growth of values for server.id.
- The values generated for server.id is in the range of 1 to 99.
- The value for server.id allocated from a pool is not sticky to a node. For example, in a cluster of 5 nodes, if node1 gets server.id values allocated as 1 and node2 gets server.id value allocated as 2; on a complete restart of all the nodes, the previous allocated number may shuffle across the nodes. Coincidentally each node may get back the same value upon restart, but it is not guaranteed.
- The allocation and de-allocation of the server.id values are tracked centrally in a database.
- Possibility of race-condition is anticipated when all the nodes can start together, and this situation is taken care to avoid allocation of duplicate value to multiple nodes.
- Refer to system configuration changes section.
Parent topic: High Availability Changes