[HBASE-12787] Backport HBASE-12028 (Abort the RegionServer when it's handler threads die) to 0.98

• Type: Task
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.98.10
• Labels: phoenix
• Hadoop Flags: Reviewed

Release Note

This change adds the site configuration property "hbase.regionserver.handler.abort.on.error.percent". This is the percent of region server RPC threads that can fail before a regionserver is aborted, and can have one of the following values:
- -1 Disable aborting (default)
- 0 Abort if even a single handler has died
- 0.x Abort only when this percent of handlers have died
- 1 Abort only all of the handers have died

• Assignee: Alicia Ying Shu
• Reporter: Andrew Kyle Purtell
• Created: 31/Dec/14 00:31
• Updated: 20/Nov/15 11:55
• Resolved: 10/Jan/15 03:23

Description

Backport HBASE-12028 (Abort the RegionServer when it's handler threads die) to 0.98. There are two 0.98-specific changes that should be addressed:

1. The default configuration should preserve current behavior
2. The interface change to RpcSchedulerFactory should be binary compatible for Phoenix if possible, and we can do this by detecting with reflection if an older implementation of the interface is missing the new 'create' method and by invoking the older deprecated 'create' method instead. We can cache the decision after making it once.

Issue Links

• is related to: HBASE-12028 Abort the RegionServer, when it's handler threads die
• is related to: HBASE-12788 Promote Abortable to LimitedPrivate