* John Goerzen <jgoer...@complete.org> [200512 22:45]: > Sorry about that, Chris. I have tried just now and, rather to my > surprise, have been unable to duplicate this in order to gather more > information. I had assumed it would be fairly readily reproducible; > that is my mistake.
Right. I've asked around, and someone insightfully suggested: systemd might have a mnt-foo.mount unit loaded (after reading /etc/fstab for the first time), and if it somehow disagrees with your changes to /etc/fstab, it -might- unmount /mnt/foo immediately after you manually mounting it. I was also pointed at https://github.com/systemd/systemd/issues/1741 - maybe anything there sound familar? > I can report to you that the kernel messages looked as if it had been > mounted; for instance: > > May 11 20:48:46 tinwhistle kernel: [10011.916687] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: (null) > May 11 20:50:19 tinwhistle kernel: [10104.961413] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: errors=remount-ro > May 11 21:06:21 tinwhistle kernel: [11067.149192] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: errors=remount-ro > May 11 21:06:30 tinwhistle kernel: [11076.165431] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: errors=remount-ro > May 11 21:06:47 tinwhistle kernel: [11093.047456] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: (null) > May 11 21:07:14 tinwhistle kernel: [11119.680451] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: (null) > May 11 21:07:30 tinwhistle kernel: [11135.623084] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: errors=remount-ro > May 11 21:07:56 tinwhistle kernel: [11161.524673] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: (null) > May 11 21:08:22 tinwhistle kernel: [11187.655296] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: (null) > May 11 21:09:05 tinwhistle kernel: [11231.262182] EXT4-fs (md0): mounted > filesystem with ordered data mode. Opts: errors=remount-ro > > I believed at the time that it was being -- at least to the kernel -- > mounted briefly and then immediately unmounted again. Okay, that would match the suggestion. In this case I'd think this is more a systemd issue than a mount issue - mount passes your mount request to the kernel, and the kernel mounts it. Then something else (probably systemd) unmounts it immediately. The differing mount opts in your dmesg output might also be relevant. Maybe adding some words to fstab(5) might really be the best thing to do :-/ Any thoughts? Thanks, Chris
