如何在 conf/elasticsearch.yml 中定义 datacenter.group 以运行 Elassandra 多数据中心？

Question

I've 2DC

DC1
x.x.x.1 running Elassandra 6.2.3 (seed)
x.x.x.2 running Elassandra 5.5.0 (seed)

DC2
x.x.x.3 running Elassandra 6.2.3 (seed)

Actually I didn't want to create a multi data center, at first I have only two nodes in DC1 but they're unable to connect with each other due to minimum version that allowed connectivity among Elassandra is 5.6.

The thing that stop me from re-install Elassandra from 5.5 to 6.2 is I have an important data on that node. So I came with the multi data center solution.

The solution I've got from Strapdata's guy previously is

1.Create a new Cassandra Datacenter DC2 running version 6.2.3, with a dedicated datacenter group (see https://elassandra.readthedocs.io/en/latest/configuration.html#multi-datacenter-configuration)

2. Re-create your indices in DC2, there is few differences in the elasticsearch mapping between version 5.5 and 6.2, so you have to deal with that manually.
If you have a lot of data to re-index, you can stop the single-thread index build with a nodetool stop -id <compaction_id>, and restart it in multi-threads, see https://elassandra.readthedocs.io/en/latest/operations.html?highlight=--num-threads#create-delete-and-rebuild-index

3. Test your application on DC2 (warning, there is breaking changes in the Elasticsearch API when upgrading)

4. Remove old DCs running version 5.5 when everything is ok on DC2.

I've search all over the internet, there're no mentioned about the datacenter.group in elasticsearch.yaml ( http://doc.elassandra.io/en/latest/configuration.html#multi-datacenter-configuration )

Now I've no idea what should I do with the datacenter.group one

please help Thanks

Answer 1

For guys that might came across the issue,

after a couple of hours, I've figured out how to define the datacenter.group

just simply add datacenter.group:<your desired name> at the bottom of elasticsearch.yml file.

Then restart Cassandra service

systemctl restart cassandra

You're good to go. all the data will automatically transfer to the new node.