[HDDS-10065] Create repair tool for OM metrics

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Won't Fix
• Component/s: OM
• Assignee: Ethan Rose
• Reporter: Ethan Rose
• Created: 04/Jan/24 23:58
• Updated: 22/Jan/24 20:12
• Resolved: 22/Jan/24 20:11

Description

Due to the bug in HDDS-10063 or the new metric added in HDDS-10064, some of the OM metrics that are persisted through restarts may not be correct. Since persistence is done to a json file, we can make an offline metrics repair tool that gathers the correct counts from the OM DB and writes them to the json file. The OM would need to be stopped during this process so it does not overwrite the json file with incorrect values. The OM would then learn the correct values on restart.

Issue Links

• is superceded by: HDDS-10110 Use RocksDB key count estimates instead of OM metrics file
• requires: HDDS-8100 Support repair commands from ozone CLI