When should cleaning start in Kassandra?

Cassandra nodetool has a command called cleanup :

cleanup [keyspace] [cf_name]

Starts immediate cleaning of keys that no longer belong to this node. This affects the node approximately the same, that the main compaction occurs in terms of temporarily increasing disk space usage and increasing disk I / O. Optionally accepts a list of column family names.

My questions:

  • When does a node have keys that do not belong to it?
  • When should I clean?
  • Should I clean regularly (e.g. once a week)?
+4
source share
1 answer

When does a node have keys that do not belong to it?

When you added new nodes to the cluster, reduced the replication rate or transferred tokens.

When should I clean?

After performing one of the above operations, if you need to save disk space. There is no harm in delaying its launch - there is a performance impact, and the only reason is to save disk space.

Should I clean regularly (e.g. once a week)?

No, only if you need to save space after one of the above operations.

+7
source

Source: https://habr.com/ru/post/1497333/


All Articles