Thanks for the suggestion. I ran memtest and it found no errors. This is an HP Gen8 microserver and according to the spec sheet uses ECC RAM. On Wed, 4 Mar 2020 at 21:05, RooSoft Ltd @ bluehost <roosoft@xxxxxxxxxxxxxxxxxxxx> wrote: > > On 03/03/2020 15:42, Jonathan H wrote: > > Update: > > > > My most recent scrub just finished. It found a few hundred errors, but > > many files that were not mentioned by the scrub at all are still > > unreadable. I started another scrub and it is finding new errors and > > correcting them, but I aborted it since I do feel like constantly > > scrubbing is making progress. > > > > Much more interestingly, I ran `btrfs rescue chunk-recover` and it > > reported that the majority (2847/3514) of my chunks were > > unrecoverable. The output from the `chunk-recover` is too long to > > include in a pastebin. Is there anything in particular that might be > > of interest? > > > > Also, I take the existence of unrecoverable chunks to mean that the > > filesystem is not salvageable, is that true? > > I would suspect memory corruption at this stage then. Run memtest and if > it finds anything wrong switch to ECC and don't use that memory for > critical anything. > > -- > == > > Don Alexander >
