Skip to content

NPD v0.7.0 release
Compare
Choose a tag to compare
@Random-Liu Random-Liu released this 26 Jul 00:05
· 713 commits to master since this release
v0.7.0
599ca53
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

Changes: v0.6.4...v0.7.0

NPD Metrics Support

In v0.7.0, @xueweiz added metrics support into NPD and made a major refactoring. The detailed design can be found in the proposal.

  • System stats monitor. A system stats monitor is added into NPD as a new problem daemon. It collects useful node problem related system stats with OpenCensus such as disk/io_time, disk/weighted_io and disk/avg_queue_len.
  • Problems as OpenCensus metrics. Besides node condition and events, problems detected by existing problem daemons are also collected into OpenCensus as metrics: problem_counter{reason="PROBLEM_REASON"} xxx for events and problem_gauge{reason="PROBLEM_REASON",type="PROBLEM_TYPE"} 1 or 0 for conditions.
  • Prometheus endpoint. A Prometheus exporter is added to export all OpenCensus metrics collected by NPD through Prometheus.
  • Plugin system for problem daemons. A plugin system for problem daemons is added. Problem daemons can be disabled at compile time with build tags, such as disable_system_stats_monitor, disable_system_log_monitor and disable_custom_plugin_monitor. New problem daemon plugins can be added in the future.
  • Plugin system for problem exporters. A problem exporter interface is added. The original kubernetes problem reporting logic was moved into k8sexporter. Prometheus support is implemented as prometheusexporter. New problem exporters can be added in the future.

Docker Image: k8s.gcr.io/node-problem-detector:v0.7.0

SHA: 1eea20dff5adb82b26b68781dff7c594b0570fd6
MD5: eb44947c63f4869202ce89474571cded

Assets 3
Loading