Nikolay Borisov writes: > <snip> > > > > I'll have to rebuild the kernel with debug symbols. Do I have to be > > booted into the kernel for that command to be useful? > > Actually I think you are hitting the issue fixed by the following patch: > https://github.com/kdave/btrfs-devel/commit/db0d10b02620b83ee592f6fc023ae146d72c5f65 > > > The patch went into 4.18, yet your initial report said the hang occurs on 4.17. Could you try running 4.19 with e73e81b6d011 ("btrfs: balance dirty metadata pages in btrfs_finish_ordered_io") > reverted. > <snip> As an update.... Yesterday after a hang up, I upgraded the firmware on the raid card. There were items in the changelog that seemed like they might apply. Unfortunately, later in the day the system hung up again. It was also unfortunate that I didn't have the debug kernel installed at the time. I am now rebuilding a debug kernel with the patch you mention above. FWIW i do see from fs/inode.c /* once for the tree */ btrfs_put_ordered_extent(ordered_extent); /* Try to release some metadata so we don't get an OOM but don't wait */ btrfs_btree_balance_dirty_nodelay(fs_info); around the same place in the un-patched linux-4.17.14 source tree that I built from, so I might have been hitting this in the previous kernel also. Cheers and thank you for all the help. sb. Scott Blomquist
