Re: [Bacula-users] Bacula and vchanger Cannot Label Volume

Hi,

Ok. Here I am replying to my own post again.

On Thu, May 12, 2016 at 2:32 PM, Pann Tolk <pann.tolk AT gmail DOT com> wrote:

...
...

Howerver, on the other hand. from the /var/log/vchanger/vchanger01.log, I can see that the magazine was initially mounted. Then, by the time vchanger attempts to perform the REFRESH command, the magazine was already dismounted:

May 12 12:15:48: created lockfile for pid 4533
May 12 12:15:48: magazine 0 has 18 volumes on /mnt/vchanger/1fde15be-83a4-4b67-accc-378d5fa10ea7
May 12 12:15:48: update slots needed. magazine 0 has 18 volumes, previously had 0
May 12 12:15:48: magazine 1 is not mounted
May 12 12:15:48: magazine 2 is not mounted
May 12 12:15:48: magazine 3 is not mounted
May 12 12:15:48: 18 volumes on magazine 0 assigned slots 1-18
May 12 12:15:48: saved state of magazine 0
May 12 12:15:48: saved dynamic configuration (max used slot: 37)
May 12 12:15:48: drive 0 previously unloaded
May 12 12:15:48: ==== preforming REFRESH command pid=4533
May 12 12:15:48: SUCCESS pid=4533
May 12 12:15:48: removing lockfile for pid 4533
May 12 12:15:48: created update lockfile for pid 4533
May 12 12:15:48: bconsole: running 'update slots storage="vchanger01"'
May 12 12:15:48: popen: child stdin uses pipe (5 -> 6)
May 12 12:15:48: popen: child stdout uses pipe (7 -> 8)
May 12 12:15:48: popen: forking now
May 12 12:15:48: popen: parent closing pipe ends 5,8,-1 used by child
May 12 12:15:48: popen: parent writes child's stdin to 6
May 12 12:15:48: popen: parent reads child's stdout from 7
May 12 12:15:48: popen: parent returning pid=4534 of child
May 12 12:15:48: popen: child closing pipe ends 6,7,-1 used by parent
May 12 12:15:48: popen: child will read stdin from 5
May 12 12:15:48: popen: child will write stdout to 8
May 12 12:15:48: popen: child executing '/usr/sbin/bconsole'
May 12 12:15:48: created lockfile for pid 4539
May 12 12:15:48: restored state of magazine 0
May 12 12:15:48: magazine 0 is not mounted
May 12 12:15:48: update slots needed. magazine 0 no longer mounted; previous: 18 volumes in slots 1-18 <==== Here magazine 0 was no longer mounted
May 12 12:15:48: magazine 1 is not mounted
May 12 12:15:48: magazine 2 is not mounted
May 12 12:15:48: magazine 3 is not mounted
May 12 12:15:48: saved dynamic configuration (max used slot: 37)
May 12 12:15:48: drive 0 previously unloaded

I don't know what's going on behind udev - the filesystem was mounted, but immediately, it gets unmounted. May be someone can shed some light to this mystery.

I believe I have found the cause of the problem posted earlier.

Apparently, systemd isolates processes launched by udev by putting them into a different mount namespace, which prevents new mounts from propagating back to the main namespace. So, by default mounts done within udev .rules do not propagate back to the host.

To fix this problem, I need to edit the "MountFlags=" in "/usr/lib/systemd/system/systemd-udevd.service". Changing it from "MountFlags=slave" to "MountFlags=shared" solved the problem.

However, I'm not sure if this permanently fixes the problem...ie not sure if the next update will cause the change to revert back to original setting.

Josh,

Not sure if you want to update your Howto documentation to let users of systemd distributions aware of this issue.

Thanks

Pann

------------------------------------------------------------------------------
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j

_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users