Long story short, the power went out a few days ago when I was out of town and I've been putting out fires. The latest discovery was that most of the content on one of my disks is missing. I ran a SMART test that came back clean and ran an xfs_repair test that very much did not. Hundreds of errors. And I don't really know how to read that report properly, despite trying to decipher it with the wiki open on another monitor. I'm gonna guess this pretty much sums it up:
bad magic number
bad on-disk superblock 4 - bad magic number
primary/secondary superblock 4 conflict - AG superblock geometry info conflicts with filesystem geometry
would zero unused portion of secondary superblock (AG #4)
would reset bad sb for ag 4
bad uncorrected agheader 4, skipping ag...
sb_icount 19968, counted 11520
sb_ifree 5257, counted 5218
sb_fdblocks 469533046, counted 462796299
As I scroll through the log, it "would junk" a whole lot of entries. And a bunch more would end up in "lost+found".
So with this in mind, I think I am supposed to run the check with a -v instead of a -n tag? But would it be easier to just run a parity rebuild on the disk, since this is the only one with problems at the moment? And how would I run a parity rebuild on a disk that is historically already in the array?