[HBASE-25500] Add metric for age of oldest wal in region server replication queue.

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Component/s: metrics, regionserver, Replication
• Assignee: Rushabh Shah
• Reporter: Rushabh Shah
• Created: 12/Jan/21 15:59
• Updated: 21/Jul/21 14:25
• Resolved: 08/Feb/21 19:08

Description

We have seen cases recently where we have wal from 2018 timestamp in our recovered replication queue. We came across this un replicated wal while debugging something else. We need to have metrics for the oldest wal in the replication queue and have alerts if it exceeds some threshold. Clearly 2 years old wal is not desirable.