1.4.0 Release Candidate 3

[Snoopy8]

A few more general tests.

1. NUC Ubuntu 24.04 install stopped, had to run sudo apt-get update before successfully install.  Pi5 install went through with no problems
   
2. Accidentally installed RC2 first, then installed RC3.  However, noticed that the SO Restore backup was still had RC2 functionality.  Not sure whether this is the correct way to do future updates?
   
3. Updated my Cisco router to use latest Fresh Tomato version.  As a result, NUC got a new ip address, had to give new NFS permissions in Synology setup.  However, still require manual remount of NFS folders on rebooting.  Maybe it is time to stop trying to find root cause of this problem, and create a "hacK" to mount remote folders at the end of SO startup process ???
   

---

I added into /etc/systemd/system/snakeoil-rest.service After=... nmbd.service
This is the slowest service in system-analyze blame.  Voila!  NFS folders are reloaded at boot time!

BUT no idea why a service for Samba is affecting the NFS load???

And confirming fix applies to Pi5

[agent_kith]

I added into /etc/systemd/system/snakeoil-rest.service After=... nmbd.service
This is the slowest service in system-analyze blame.  Voila!  NFS folders are reloaded at boot time!

BUT no idea why a service for Samba is affecting the NFS load???

And confirming fix applies to Pi5

You are right, this doesn't make sense. If you look at the nmbd.service file, it wants network.target, and will start after network and network-online. Which is exactly the same setup as Snakeoil service file in RC-1 (or RC-2). This is really a bug in systemd (Or Ubuntu) but can't really be fixed easily.

Since samba is a dependency of Snakeoil anyway, I guess it's safe to start Snakeoil after nmbd service, which is a really good idea anyway just in case CIFS will fail to mount also. Will add this into official High End release.

[Snoopy8]

Posting this, in case it may useful for understanding and solving root cause.

I was curious to see whether I could fix the same reload NFS problem with 1.3 and Bookworm on Pi5.  And adding a "After=nmbd.service" fixed the problem.  And curiously, nmbd.service was relatively fast here compared with the NUC Ubuntu 24.04.  This is the systemd-analyze blame output on Pi5 Bookworm.
 

Code:

7.153s NetworkManager-wait-online.service
1.261s dev-mmcblk0p2.device
1.201s raspi-config.service
492ms [email protected]
449ms systemd-timesyncd.service
358ms ModemManager.service
349ms nmbd.service
310ms smbd.service
282ms NetworkManager.service
271ms lighttpd.service
266ms systemd-logind.service
263ms samba-ad-dc.service
252ms keyboard-setup.service
241ms bluetooth.service

[agent_kith]

Fixing the network page will require changing to the API backend, which is not feasible as that is going to be one to be updated next. So will disable network configuration for now, but it will come back. After some thinking, I might go with Qt6 (i.e. still C++) instead of GoLang. But will experiment around and find the best option. 

With this out of the way I guess 1.4.0 will be ready to be released. Hopefully I can write up enough information in the post to make sure the transition is smooth....