Not even sure where to start - some services failing to run on 3.0.64

It does, since Syncthing is designed to be per-user sync like Dropbox is. The service it uses is syncthing@.service, and the username running the process goes after the @.

What I can't figure out is why none of the sysvinit wrapped services appear to be working. I'm also more than slightly surprised that Samba doesn't have a native systemd service file, though I wouldn't be shocked to find that that's Debian being a bit lagged.

I'll give that a whirl later tonight or tomorrow once I have a bit to play with it.

Zitat von ryecoaaron

If you want to try my latest change to using the systemd unit file instead of my init script, you can download the package here.

Okay, first thing's first: It works just fine for me. Uninstalled Syncthing and plugin, manually deleted the config files after backing them up for the users I was using, manually reinstalled syncthing then did a purge of it (just in case), then reinstalled via OMV WebUI and went in.

First I created a dummy one with a different user on different ports from the default. Applied no problem, Syncthing GUI came up right away. Deleted that dummy user, set one up for my original user on the default ports, and for a minute or so I was getting connection refused. Then the UI with mapped folders, Username/Password prompt, and changed theme just like it was before testing popped up. Clearly there's somewhere other than ~/.config/syncthing/ that the user configs are saved.

On a momentary whim, however, I went back to the port that I used for the dummy user, and it was still up and running. Given how heavy scans and syncs can be, that could be bad.

Other than that, seems to be working just fine.

Zitat von ryecoaaron

Thanks for pointing that out. Deleted users are disabled now. If you want to try the latest package, it is available in the same place.

Yep, that took care of it. I note that when I deleted the test user, the regular user's web UI became unavailable for about 30 seconds to a minute. I mention it more as a head's up for when someone says "OMG I DELETED $OLDUSER AND YOU DELETED $CURRENTUSER TOO!!!!11!!eleven1!".

Any ideas as to what could be causing pretty much every SysV-controlled script to be blowing up? I've compared scripts between the two boxes and I come up with nothing really different.

Zitat von ryecoaaron

New version in same location.

Gotcha, will give it a spin shortly.

Random thought, I was able to bring my CPU utilization WAAAAAY down - like, average of 15% to about 0.1% - by installing syncthing-inotify along with it. Only problem is, it pretty well chews through the inotify queue and you have to modify /etc/sysctl.conf to increase it. Not exactly a good idea for a plugin, but if you've got a weaker system using Syncthing, might make a huge difference.

Zitat von ryecoaaron

Maybe try reinstalling systemd-sysv and sysvinit-utils?

Might as well. I'd tried apt-get install --reinstall samba to see if I could get it to rebuild the sysv script-generated unit, but that didn't impact. Might as well do the same thing with those.

Zitat von ryecoaaron

Just need to figure out where to set the max user watches.

It's in /etc/sysctl.conf, needs to have fs.inotify.max_user_watches=204800 to make the change permanent, but to change it for the running system, /proc/sys/fs/inotify/max_user_watches is what needs to be set to 204800. Syncthing's github page tells you to use the command sudo sh -c 'echo 204800 > /proc/sys/fs/inotify/max_user_watches' to set it, and it needs to be done prior to starting inotify.

Haven't had a chance to test yet, I rebooted the remote NAS earlier for something else and it never came back online. When/if I can get a hard reboot and make sure there's power to it, I'll update and test.

Thanks!

Alright, got it rebooted, not sure what exactly happened but the system was running, it just didn't have a network connection somehow. networking.service fails to run, which I'm beginning to think may be the root of the problem altogether. Note that network-online.target is fine, and I'm (now) able to access remotely over SSH and the web UI, but networking.service is down. Here's the status output:

root@mjolnir:~# systemctl status networking
● networking.service - Raise network interfaces
   Loaded: loaded (/lib/systemd/system/networking.service; enabled; vendor preset: enabled)
  Drop-In: /run/systemd/generator/networking.service.d
           └─50-insserv.conf-$network.conf
   Active: failed (Result: exit-code) since Sat 2017-03-11 21:40:26 EST; 14min ago
     Docs: man:interfaces(5)
 Main PID: 6239 (code=exited, status=1/FAILURE)


Mar 11 21:40:25 mjolnir ifup[6239]: RTNETLINK answers: File exists
Mar 11 21:40:25 mjolnir ifup[6239]: bound to 192.168.1.133 -- renewal in 33111 seconds.
Mar 11 21:40:26 mjolnir ifup[6239]: ifquery: recursion detected for interface eth0 in post-up phase
Mar 11 21:40:26 mjolnir ifup[6239]: run-parts: /etc/network/if-up.d/openmediavault-issue exited with return code 1
Mar 11 21:40:26 mjolnir ifup[6239]: Failed to bring up eth0.
Mar 11 21:40:26 mjolnir ifup[6239]: invoke-rc.d: action alarm is unknown, but proceeding anyway.
Mar 11 21:40:26 mjolnir systemd[1]: networking.service: Main process exited, code=exited, status=1/FAILURE
Mar 11 21:40:26 mjolnir systemd[1]: Failed to start Raise network interfaces.
Mar 11 21:40:26 mjolnir systemd[1]: networking.service: Unit entered failed state.
Mar 11 21:40:26 mjolnir systemd[1]: networking.service: Failed with result 'exit-code'.
root@mjolnir:~#

Now, that was the result of me trying to enable IPv6 via the web UI, but I just issued a systemctl start networking.service and the only thing that changed was the timestamps. On the other hand, that might be something from sysvinit and this is an effect of the issue, not a cause.

On the plugin updates for Syncthing, uploaded the new one, I've got the enable switch for inotify now, and it works if I create with it (and doesn't if I don't), shuts off if I disable, comes up when I enable, and all around works as I'd expect it to.

Zitat von ryecoaaron

Latest package in same location has the inotify stuff in it. I think it should be close to ready.

That, I'm going to have to set up a test VM for. Don't want to revert the changes on the live boxes now that I've got it just so. Should be able to do it later tonight.