Unit U5.2 · embed after: cluster operation and the fault-tolerance test
HDFS Fault Tolerance - dynamic cluster, replication, racks and failover
Data Science & Big Data Analytics • ITec4121 • Wolkite University • Online-First
A fully parameterised cluster: choose 3–6 DataNodes, 10–30 blocks (slider), a replication factor (1–3) and rack-aware placement. Kill one or half the nodes at random, reboot everything, or toggle nodes by hand and read any block - watch healthy / at-risk / lost blocks and the pending re-replication task count, exactly the terminal drill you perform in U5.2.
Cluster & controls
Nodes:Replica placement & health
Sim14_HDFS_U5-2.html · Simulation 14/14 of 14 · online-first course package · back to top