SysAdmin

A quick reference for calculating the exact timeout window before Hadoop marks a Datanode as dead, based on the heartbeat and recheck-interval settings.

When hostname -f returns a short name instead of a fully qualified domain name, the issue usually traces back to the HOSTNAME variable in /etc/sysconfig/network or a misconfigured /etc/hosts entry. Understanding the difference between the gethostname and gethostbyname system calls makes the fix straightforward.

Kerberos authentication failures can be tricky to diagnose without detailed logging. This guide walks you through enabling Kerberos event logging on a Windows machine via the registry so you can capture authentication events in the system log.

A quick reference for the most useful JVM debugging commands, including heap inspection, environment introspection, and enabling GC logging on a running Java process.

Specific JDK 1.8.0 builds (u40, u45, u91, u92) are known to break SPNEGO authentication and Hadoop web interfaces. This post documents the affected versions, the symptoms, and the recommended downgrade path.

Enable verbose SSL/TLS handshake logging in a JVM running Hadoop or Cloudera services to diagnose certificate and protocol issues.