We are currently using version 2.8 with cluster with more than 300 nodes. We encountered a limitation because of the size of the physical address hardcoded to 20 in 2.8. I noticed that 2.10 solved the issue and used the code to fix 2.8. But I realized that since the logical_addr_cache is created inline and not in the init() method, the configuration does not affect its size since it uses the default value of 20.

The cache should be created in the init() method for the configuration properties to be taken into account.