fstab: /var/db/squashfs/repos.rw=rw:/var/db/squashfs/repos.ro=ro /var/db/repos fuse.unionfs user,allow_other,use_ino,nonempty,auto_unmount,cow,dev,hide_meta_files,comment=systemd.automount 0 0 Now it seems that if I mount something on top of it (e.g. underlying squashfs), and then 'umount -l' systemd gets in some ugly unmounting loop. journalctl -f: mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. mar 25 10:01:41 pomiot systemd[285]: Failed unmounting /var/db/repos. and so on. Sometimes intermixed with: mar 25 10:01:41 pomiot umount[20859]: umount: /var/db/repos: odmontowanie nie powiodło się: Operacja niedozwolona [EINVAL]
What's interesting, also 'systemd --user' is killing CPU. If I SIGTERM both master systemd and 'systemd --user', the systemd returns to normal.
That is with 219*? Also, does it change when changing "lzo" USE flag value? Also, maybe running "journalctl --verify" will show some corruptions that could explain this problems :/
So more details: it actually happens immediately after mounting /var/db/repos, either directly via 'systemctl start var-db-repos.mount' or via the relevant automount. At this point, systemd starts very fast unmounting attempts... I have: $ cat /etc/systemd/system/var-db-repos.mount.d/req-sqfs.conf [Unit] Requires=var-db-squashfs-repos.ro.mount After=var-db-squashfs-repos.ro.mount which may be relevant. It may also be related to #541402. It's not Journal-related.
It looks to be regression in automounts. For now, I just disabled automounting as a workaround.
Yeah, sorry for the journald comment, I guess I was thinking on other bug/problem that day :S
Guys, I had same problem with mount -loop some.iso isodir, and I found that following patch fixes the problem for me: http://cgit.freedesktop.org/systemd/systemd/patch/src/core/mount.c?id=628c89cc68ab96fce2de7ebba5933725d147aecc Please try it and apply! :)
Should probably grab this too. http://cgit.freedesktop.org/systemd/systemd/patch/src/core/mount.c?id=496068a8288084ab3ecf8b179a8403ecff1a6be8
Please give systemd-219_p112 a try. +*systemd-219_p112 (26 Apr 2015) + + 26 Apr 2015; Mike Gilbert <floppym@gentoo.org> +systemd-219_p112.ebuild: + Add a snapshot from the v219-stable branch upstream.
