One the first write bringing the unit file in existence we can just
start it, after that we need to tell systemd that we want to actively
reload it.
While this is slightly shaky due to the fact that we do not check all
paths where such a unit could reside, it is something we can do
because earlier one couldn't have a unit/overwrite anyway (from
procfs mountinfo generated one do not support that) and does adding
such override ones from now on should work.
Also note that we can only get here in the "user does no weird stuff"
case when "cephfs_is_mounted" actively tells that there is no cephfs
mounted at the $mountpoint - at which time we can safely re-write the
potential updated unit file, reload and mount again.
So let's make our life a bit easier here until a user actually
complains about a rational issue for this, maybe we have PVE 7.0 then
and can get rid of that anyway :)
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
This fixes a potential races where fuse get's unmouted to late in the
shutdown process, i.e., at a time where network was down and it could
not talk to any MDS or monitor anymore.
We could fix it the same way we did once with the kernel based mount,
i.e., adding _netdev, but doing so would require to switch over from
"ceph-fuse" to "mount.fuse.ceph" which has better compatibility with
the common mount tool API.
As that helper exists we can reuse the newer systemd_netmount
ephemeral unit generator, only some options differ in name between
fuse and kernel variant.
So besides solving a potential issue we get a more unified handling
of those two cases.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
commit 54e0b0034b introduced the
"_netdev" option, for PVE 5.3. The systemd generator then correctly
resolved that in the following resulting order-dependencies:
> Wants=network-online.target
> Before=umount.target remote-fs.target
> After=remote-fs-pre.target system.slice network.target network-online.target -.mount
This worked well and all were happy. With the current systemd in 6.0
we sometimes get the local-fs ones there generated too. This is a
fallout from a try to better handling nested mount hierachies, where
a .mount unit needs to be mounter or unmounted, before or after,
respectively, the parent mount was processed. It seems that sometime
that glitches and thus a "RequireMountFor=/mnt/pve" gets thrown in
and result sometimes in the local-fs order constraints being added.
The issue now is, that one must not have ordering depends to all,
local-fs, local-fs-pre, remote-fs, remote-fs-pre, as that gets you a
ordering cycle. Systemd tries to solve that cycle by randomly
dropping one constraint and retrying. By luck this is a not so
important unit, and all goes on well. Most of the time one isn't that
lucky and something important gets dropped, for example:
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Found ordering cycle on systemd-timesyncd.service/stop
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Found dependency on systemd-tmpfiles-setup.service/stop
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Found dependency on local-fs.target/stop
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Found dependency on mnt-pve-cephfs.mount/stop
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Found dependency on remote-fs-pre.target/stop
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Found dependency on rbdmap.service/stop
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Found dependency on sysinit.target/stop
> Jan 24 18:43:05 prod1 systemd[1]: sysinit.target: Job remote-fs-pre.target/stop deleted to break ordering cycle starting with sysinit.target/stop
Then, most of the time the host reboot hangs for ~10 minutes, often
showing scapegoat units like the pve-ha-lrm being the cause of the
hang (even if no HA is configure >.<).
This behavior is fixed with newer systemd versions, e.g., the v244
from buster-backports, but that is not a real option for us for now.
So until 7.0 we generate the unit with the correct dependencies
directly in the ephemeral /run/ tmpfs backed systemd/system path and
start it.
While FUSE gets only the local-fs ordering constraint, it seems to cope
very well regarding such symptoms. But it _is_ racy and probably only
works due to systemd stopping it early as it has not much ordering
constraints at all.. It should be moved in the future nonetheless, as
there's a mount.fuse.ceph helper that should be not an issue.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
This is but a hack, but we have no general helper/tools module here
and I do not want to do versioned dependencies for this fast-tracked
bugfix to pve-common, so I'll have to live with the shame for now.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
checking '$server:$subdir' is too strict to work in all cirumcstances,
e.g. adding/removing a monitor would mean that it is not the same
anymore, same if one is adding/removing the ports from the config
check only if the subdir is the same and if it is a cephfs
this way, it still returns true if someone changes the config
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Since ceph-fuse is called directly in the CephFS storage plugin, which
can not process the _netdev option, mounting the CephFS storage fails
when fuse is set in the storage.cfg.
This patch moves the _netdev option into the else part of the if fuse is
set statement. _netdev is only added if the CephFS kernel client mounts
the storage.
It seems _netdev is not needed anyway for the fuse mount, as the
connection is closed, once the fuse process gets killed on shutdown.
Signed-off-by: Alwin Antreich <a.antreich@proxmox.com>
it is not a storage plugin, and it makes more sense to have it
top-level, but there we cannot name it CephTools because of the
existing ones in pve-manager
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
As we mount this manually and thus systemd doesn't know about any
dependency for cephFS mounts, this got umounted only at the last
stage of shutdown, where network wasn't active anymore.
But, CephFS needs to be connected to an active MDS for a clean
unmount so without network this mount would delay shutdown for quite
a bit, until after some minutes systemd gave up and forced unmount.
So tell systemd that this mount requires network, which can be done
with the '_netdev'[0] mount option, that lucky for us can be also
passed to a mount call and isn't only available for fstab.
with this a mount gets, among others:
> Wants=network-online.target
> Before=umount.target remote-fs.target
> After=remote-fs-pre.target system.slice network.target network-online.target -.mount
Which does the trick for us.
[0]: https://www.freedesktop.org/software/systemd/man/systemd.mount.html#_netdev
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
- ability to mount through kernel and fuse client
- allow mount options
- get MONs from ceph config if not in storage.cfg
- allow the use of ceph config with fuse client
- Delete secret on cephfs storage creation
Signed-off-by: Alwin Antreich <a.antreich@proxmox.com>
