Reaching Zen in ElasticSearch's Coordination
summary
Elasticsearch rewrites its coordination (Zen) for 7.0 from a formal TLA+ model and model checking to fix repeated network-partition bugs; master elections become much faster, and the manual, error-prone minimum_master_nodes setting is gone, replaced by an initial_master_nodes bootstrap list used only on first startup. Publication now requires majority persistence before commit, cluster-state loss bugs are eliminated in tested scenarios, and new logs give clearer debugging of discovery and ping timeouts.
AI-written summary · about
chapters (7)
- 0:00Introduction to cluster coordination
- 3:56What is Zen in Elasticsearch
- 7:06Discovery and master election
- 11:59Initial master nodes setting
- 16:20Cluster state publication
- 17:24Summary and improvements
- 21:19Q&A on node removal and pings
related talks
37mInside Onyx
20mDiving into ElasticSearch Discovery #bbuzz
27mMesos HA Getting Rid of Zookeeper
44mElasticsearch at Scale of Billions of Documents
15mUnconference Session: Dedicated Coordination Nodes - Why? When? How?
22mDesigning Concurrent Distributed Sequence Numbers ...
38mTowards consensus on distributed consensus #bbuzz
22mIntroducing Zelos - Zookeeper API leveraging Delos