DiskPaths is a check that monitors the port combinations und sends an alert, if they are not redundant. In the CLI, one would use the command
storage show disk ‑p to check the list of ports.
FILER-01> storage show disk -p
PRIMARY PORT SECONDARY PORT SHELF BAY
------------------ ---- ------------------ ---- ---------
DSWT9015:6.126L1 B DSWT9014:6.126L27 A 0 0
DSWT9015:6.126L2 B DSWT9014:6.126L28 A 0 1
DSWT9014:6.126L29 A DSWT9015:6.126L3 B 0 2
DSWT9015:6.126L4 B DSWT9014:6.126L30 A 0 3
For hundreds of disks this can be quite a task, one that can not be done regularly. This is why we developed the plugin DiskPaths.
We are now updating this plugin to support the Fabric Metro Cluster, so that 8 paths can be monitored. This is presented in the figure bellow (click on the picture to open a more readable PDF).
What kind of scenarios are imaginable. In a correctly wired, perfectly running FMC there are two A-ports and two B-ports for each disk and node – hence combinations such as ABBA, BABA or AABB and BBAA.
For there to be any deviations of these port combinations, one or more of the following events has to occur:
- wiring error: AAAA or BBBB or AAAB or BBBA
- failure of an initiator port: AAB or BBA
- failure of a switch port, a whole switch or bridge: AA or BB
- two initiator or switch ports have failed: AB, BA (excluding FAS 8020 with reduced backend wiring)
- multiple failures and/or combinations of failures: A or B
- even more failures: no port available
As demonstrated above, this plugin can detect quite a number of different error cases in the cluster.
The 8 path logic will be available in version 3.3.0_01 (currently unstable).
Special thanks to Walter Ring of NetApp, who was willing to share his comprehensive knowledge with me, which allowed me to understand the underlying fundamentals.