On Wed, Jun 25, 2025 at 09:12:29AM +0200, Anton Khirnov wrote: > For backups, I want the fs to be static. Back when I wrote this code, it > seemed reasonable to just make the snapshot read-only, so nothing > running on the system could modify it accidentally. As it has always > worked fine until now, I had no reason to dwell on it any further. I > suppose I drop the '--permission r' option and mount the snapshot -oro > instead, but the fact that any change is needed at all still strikes me > as not quite right.
Apologies for not having time to look into this further until this week, but thanks to your observations, I was able to find the root cause of the problem, which is a kernel bug. It's a corner case that only happens if (a) you have an unlinked file which is still has an open file descriptor (which is why we have the orphan inode list in the first place), (b) during that time, you freeze the file system while creating a snapshot, and (c) the snapshot is created read-only (as opposed to just mounting the file system with the ro mount option). The fix is here[1], and should be in the upstream kernel in the next week or so, and then it will get backported to the LTS kernels, and from there, hopefully it will end up in the Debian kernel. [1] https://lore.kernel.org/all/[email protected]/ BTW, which version of the Debian kernel were you using? Thanks, - Ted
