FS#42224 - [connman] shuts down before nfs shares unmounted

Attached to Project: Community Packages
Opened by fiat500 (fiat500) - Friday, 03 October 2014, 00:57 GMT
Last edited by Balló György (City-busz) - Saturday, 27 January 2018, 10:06 GMT
Task Type Bug Report
Category Packages
Status Closed
Assigned To Christian Rebischke (Shibumi)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:
Connman shuts down before network shares (nfs) are unmounted, causing systemd to wait to unmount them, but the network connection is already disappeared.

Additional info:
Using cmst to connect/configure connman backend, but it seems as if connman needs a line in its systemd target file to wait for network shares to umount.

Steps to reproduce:
Running connman without any other network manager, and with network shares mounted upon reboot/shutdown.
This task depends upon

Closed by  Balló György (City-busz)
Saturday, 27 January 2018, 10:06 GMT
Reason for closing:  Fixed
Comment by Daniel Wallace (gtmanfred) - Tuesday, 21 October 2014, 22:07 GMT
This shouldn't be a problem.

The connman service has Before=remote-fs.target so it should be started before remote-fs.target, and should also stopped after the remote-fs.target while shutting down.

Do you have _netdev in your /etc/fstab for that mountpoint?

Can you provide me some more information, or a link to anyone else with this issue?

Also, which version of connman are you on, cause it looks like this issue was fixed with that remote-fs.target addition back in February.

connman@connman.net/msg15117.html"> https://www.mail-archive.com/connman@connman.net/msg15117.html
Comment by fiat500 (fiat500) - Tuesday, 21 October 2014, 22:45 GMT
I'm using connman from community, which is currently 1.25-1. I was running cmst-git on top of it. I was also mounting any nfs shares manually from the commandline given that I have a local partition of the same mountpoint (that's recently, originally I was mounting in a different place, but still same behaviour). I do show that remote-fs.target is enabled probably by dependency. I can mount it under sshfs and do the correct uninterrupted behaviour (given that sshfs gives way on any disconnect). I show nfs-client under remote-fs.target.wants, and connman under multi-user.target.wants under /etc/systemd/. As far as I can tell connman shuts down before remote-fs, as I don't see any After= in any of the target.wants that I mentioned here. I do have something come up on startup about pnfs and that it cancels nfs-server due to a cycle order with rpc-(I'd have to enable it again), but I have disabled nfs-server and shouldn't have any affect as to the mounted nfs-client shares.
Comment by fiat500 (fiat500) - Tuesday, 21 October 2014, 23:52 GMT
Looks like connman updated. Haven't tried to see if it still does nfs mounts yet.
Comment by fiat500 (fiat500) - Friday, 07 November 2014, 15:06 GMT
It looks like the last time I had nfs mounted, it was still tied up with connman closing out first.
Comment by Christian Rebischke (Shibumi) - Saturday, 27 January 2018, 03:00 GMT
Is this bug still an issue? I mean we have 2018 now, otherwise I am going to close this issue.
Comment by fiat500 (fiat500) - Saturday, 27 January 2018, 06:44 GMT
It seems to have rebooted with nfs mounts and connman running so it's probably been committed to repos and/or git aur by now.

Loading...