On Wed, Mar 06, 2019 at 07:55:45AM -0600, Russell Mosemann wrote: > The four most problematic servers were hanging nearly every day. They each > have a dedicated hard drive and are running 4.19.0-0.bpo.2-amd64. I > installed btrfs-progs 4.20.1 on each and recreated the file system on the > hard drive. There have been no issues since then, but it is too early to > tell. There are hardly any references or files on the drives. It might > take up to a month to know if there is an issue. > Thanks for reformatting with newer -progs; this eliminates the most difficult/impossible to reproduce factor. Which of the five rounds of tests from https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=908216#202 have you started with? It should be possible to accelerate triggering the bug by using the network to send additional VM images from each server to the test subject once a day. eg: if four servers send VM images to a fifth server then ideally (optimistically!) the bug would be triggered 80% faster, and if everything is still all clear after a month then you'd have at least 220 reflinked historical backups and five sparse copies of large VM images. Thank you for your work on making this bug reproducible. This not only accelerates its resolution (reproducible case in core functionality needs to be forwarded upstream), but also provides valuable data to inform our wiki's upcoming btrfs on buster (linux-4.19) recommendations. Cheers, Nicholas
Attachment:
signature.asc
Description: PGP signature