[LU-17216] enable_health_write, health_check improvements

• Type: Improvement
• Resolution: Unresolved
• Priority: Minor
• Severity: 3
• Assignee: Tim Day
• Reporter: Tim Day

Description

enable_health_write should be tunable rather than a compilation option. This allows us to test it more easily and gives admins the option to try it out without having to recompile their Lustre servers. It will still be disabled by default.

 

This health write should be enabled for MDT/MGT also. Especially since DNE means there are many more metadata related disks.

 

Getting more verbose info from health checks would be useful. Lustre should report health by OBD device. It should also tell you what's wrong. To implement this, the health check functions could return a enum indicating the root cause of the health check failure (disk IO, ptlrpc, etc.). Then, the individual check need only return the correct enum.

Issue Links

• is related to: LU-17450 sanity: interop test failures with master+2.15