On 16 Feb 2019, Andreas uttered the following: > Thanks for your reply. > > as far as I can tell bcache is caching BtrFS metadata for me, with or > without my patch. Perhaps BtrFS correctly flags it as REQ_PRIO. In my I think that's probably it: XFS is different (and apparently REQ_PRIO is the wrong thing to look for after all). > testing bcache was consistently (and repeatedly) bypassing file contents > just because they were read as a result of a readahead operation, hence > the need for my patch. I also found that, before my patch, > sequential_cutoff and congestion thresholds had close to no effect on > how much would be cached or bypassed. For what it's worth, on XFS that is definitely not true: push up the sequential_cutoff and the amount cached goes up. (I'm pushing it up periodically in RAID-chunk-size units until I find that enough is cached for my metadata reads to disappear.) > The problem you linked to seems to be a separate issue where different > filesystems flag their internal IO inconsistently leading to similar yet > different problems. Yep. > a solution as I feel like my knowledge of bcache and the Linux kernel in > general is far too limited to produce a patch of sufficient quality. But that's true of everyone to start with! Our biggest problem is impostor syndrome. Dive in! (But keep good backups.) -- NULL && (void)
