On , Qu Wenruo wrote: > [27240.680874] perf: interrupt took too long (3952 > 3942), lowering > kernel.perf_event_max_sample_rate to 50400 > > [30658.875802] BTRFS warning (device dm-12): checksum error at logical > > 37889245122560 on dev /dev/mapper/data7, sector 2743145096, root 23674, > > inode 206751, offset 762638336, length 4096, links 1 (path: > > アニメ/!waiting_for_better_quality/Gate: Jieitai Kanochi nite, Kaku > > Tatakaeri/GATE Jieitai Kanochi nite, Kaku Tatakaeri 05v2.mp4) > > Well, it's several seasons ago, and I think there are better BDrip raws now. > (Yeah, I'm also an Otaku) > > Despite that, it's better to hide such personal info though.
Since downloading stuff from internet is legal in my country I don't usually bother to hide stuff like this, but will do so if it's an issue in this mailing list. > And, did you tried to scrub the corrupted device other than the whole fs? > Btrfs default scrub will start threads to scrub all devices at the same > time, maybe some concurrency caused the false alert. Tbh I had no idea I can scrub just the device and not whole filesystem, running it now (but the scrub on this drive takes like 12 hours so I see tomorrow if it helped). > Also, it could be possible to check/repair it by using btrfs-progs. > Although it's still out-of-tree. > > Could you please try the following branch and use "btrfs scrub start > --offline /dev/mapper/data7" to check if it reports the corruption is > fixable? > https://github.com/gujx2017/btrfs-progs/tree/offline_scrub > > Offline scrub gives us a quite good reference on whether it's fixable, > without the possible hassle in kernel. > So it's worth trying. If scrubbing just the device doesn't help in any way, will give it a try. > (But hey, there is better better BDrip raws already, so I don't think > you're really interested in fixing the corruption) True. Plus since it's RAID1 no data were actually lost and the is working without problem. I'm mainly interested in knowing if it's 1) Issue with HW 2) Some hidden issue with the whole fs and it's going to fall apart soon Thanks for tips so far :) W. -- There are only two hard things in Computer Science: cache invalidation, naming things and off-by-one errors.
signature.asc
Description: PGP signature
