All Questions
Cassandra hard

What is the critical operational danger of setting `gc_grace_seconds` to a very low value (e.g., 1 hour) in a cross-datacenter Cassandra cluster?

  1. A The cluster will run out of JVM heap space and enter a constant GC loop.
  2. B Data Resurrection: If a node is down for 2 hours (longer than the grace period), it misses the deletion and its tombstones are purged on the healthy nodes. When it returns, it will treat its 'old' data as valid and replicate it back to the whole cluster.
  3. C Encryption keys will expire, preventing the database from booting.
  4. D Cassandra will stop accepting writes until a full backup is performed.

One system design question a day, with spaced repetition so the ones you miss come back.

Get the app

Related questions