Details
-
Type:
Sub-task
-
Status: Resolved
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: Auto failover (HDFS-3042)
-
Fix Version/s: Auto failover (HDFS-3042)
-
Component/s: auto-failover, documentation, ha
-
Labels:None
-
Target Version/s:
-
Hadoop Flags:Reviewed
Description
We should document how to configure, set up, and monitor an automatic failover setup. This will require adding the new configs to the *-default.xml and adding prose to the "apt" docs as well.
Activity
- All
- Comments
- Work Log
- History
- Activity
- Transitions
Looks good Todd, minor comments:
- Instructions say dfs.ha.automatic-failover.enabled is in core-site.xml, should be hdfs-site.xml
- "notifying the other machines" / "notifying the other NameNode"
- "active node" / "active NameNode"
- "acts as a client of ZooKeeper" -> "is a ZooKeeper client"
- "has a ZKFC running on the same machine" / "has a ZKFC that runs on the same machine as the NameNode"
- "special health-check command" / "health-check command"
- Use "local NameNode" instead of "local node" most places, eg to avoid confusion with node in the ZK sense (eg "lock node")
- "users may refer" / "refer"
- "within 10-20 seconds", worth mentioning "ha.zookeeper.session-timeout.ms" instead of a raw value
It's not introduced by your change, but you might as well fix it here:
"Prior to Hadoop 0.23.2, the NameNode was a single point of failure (SPOF) in an HDFS cluster." - this actually won't appear in 0.23.2, and instead will first show up in 2.0.0.
Attached new patch addresses the above feedback. I also attached a delta for your reviewing pleasure.
+1 looks great
Committed to branch, thanks for reviews
Attaching patch as well as the rendered HTML file.