evcc does not start after update to 0.86

[RTTTC]

When updating from 0.85 to 0.86 using sudo apt upgrade
I got the error

Unpacking evcc (0.86.0) over (0.85.0) ...
dpkg: warning: unable to delete old directory '/usr/local': Directory not empty

After the update EVCC starts manually, but EVCC service does not start anymore. Will have to try and change directories I guess?
That did not help:

  GNU nano 5.4                                                                        /usr/lib/systemd/system/evcc.service                                                                                  
# evcc.service
#

[Unit]
Description=evcc
Requires=network-online.target
After=syslog.target network.target network-online.target
Wants=network-online.target
StartLimitIntervalSec=10
StartLimitBurst=10

[Service]
ExecStart=/usr/bin/evcc
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

and i get this result:

root@picharger:~# sudo systemctl status evcc
● evcc.service - evcc
     Loaded: loaded (/etc/systemd/system/evcc.service; enabled; vendor preset: enabled)
     Active: failed (Result: exit-code) since Wed 2022-03-16 15:30:20 EET; 11min ago
    Process: 832 ExecStart=/usr/local/bin/evcc --log error (code=exited, status=203/EXEC)
   Main PID: 832 (code=exited, status=203/EXEC)
        CPU: 6ms

Mar 16 15:30:20 picharger systemd[1]: evcc.service: Scheduled restart job, restart counter is at 5.
Mar 16 15:30:20 picharger systemd[1]: Stopped evcc.
Mar 16 15:30:20 picharger systemd[1]: evcc.service: Start request repeated too quickly.
Mar 16 15:30:20 picharger systemd[1]: evcc.service: Failed with result 'exit-code'.
Mar 16 15:30:20 picharger systemd[1]: Failed to start evcc.

Also - not sure if this is relevant, but I had a manual EVCC installation, before there was a way to update using APT , then I did not do anything with the existing installation, and just did the steps to have apt https://docs.evcc.io/docs/installation/linux
Could there be two conflicting EVCC installations on my raspberry somehow?

Originally posted by @RTTTC in #2825 (comment)

[Pitillo]

I needed to edit the service file under /etc/systemd/system/evcc.service and change ExecStart=/usr/local/bin/evcc --log error to ExecStart=/usr/bin/evcc --log error. Maybe you have one service file left in /etc/systemd/system/ since it has higher priority than the one located under /usr/lib/systemd/system/. After you altered the file you need to execute systemctl daemon-reload and systemctl restart evcc.

For further debugging: could you post the output of journalctl -u evcc -n 25?

[RTTTC]

Thank you, that was exactly the case!!!

Maybe you have one service file left in /etc/systemd/system/ since it has higher priority than the one located under /usr/lib/systemd/system/

Now everything runs fine.
Should I delete something that is old and not needed? And how to do that?

[Pitillo]

In my opinion you could delete the file /etc/systemd/system/evcc.service using rm /etc/systemd/system/evcc.service if a similar file is inplace under /usr/lib/systemd/system/evcc.service afterwards run systemctl deamon-reload again. You should then be safe for further updates if they change something to service file. Further I would recommend to mark the issue as solved.

[andig]

Ich frage mich immer noch, woher 2 unterschiedliche .service Dateien kommen sollen. Unsere Paketierungsskripte liegen in https://github.com/evcc-io/evcc/tree/master/packaging/scripts, ob da irgerndetwas bei deb-helper dazu führt, dass sich auch der Ort der Servicedatei ändert?

[RTTTC]

In my case - the answer is easy, I was simply following the instructions, when this was the only way to install it #2905

[Pitillo]

The manual documentation under https://docs.evcc.io/docs/installation/manual still mentions to create the file under /etc/systemd/system/evcc.service so if a person installs evcc manually first and then decided to use the package the file under etc persists.

[andig]

Na das ist natürlich Quatsch. Würdest du einen PR machen um die gleichen Pfade zu nutzen? Dazu stell ich dann hier noch ein Pin mit der Info ein.

[RTTTC]

I will have a look wether we can maybe add an option on update to delete the file using a postinst script. Something like ask the user wether he wants to keep the file or he wants it removed.

Something like this could be extremely useful, as more and more people report having this problem again and again.

[gonny8]

Since update to 0.86 evcc is unabel to start

see: #2932

Ich habe die Installationsanleitung befolgt und die Geräte mit dem Assistenten hinzugefügt.
Im Log steht nun immer "evcc.service: Main process exited, code=exited, status=2/INVALIDARGUMENT"

[Pitillo]

This seems to be another problem and not related to the other issue discussed here. I would assume that your config file is invalid. Maybe you can show the output of journalctl -u evcc.service -n 25 --no-pager and your configuration file (without private Infos like login tokens)

[gonny8]

Here is the output:

pi@raspberrypi:~ $ journalctl -u evcc.service -n 25 --no-pager
-- Logs begin at Fri 2022-03-18 23:17:01 CET, end at Mon 2022-03-21 09:05:55 CET. --
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/meter/config.go:33 +0x15c
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/evcc-io/evcc/meter.NewMeterFromTemplateConfig(0x344b900)
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/meter/template.go:43 +0x1d8
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/evcc-io/evcc/meter.NewFromConfig({0x3658bd8, 0x8}, 0x344b900)
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/meter/config.go:33 +0x15c
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/evcc-io/evcc/cmd.(*ConfigProvider).configureMeters(0x32036c8, {{0x3658b70, 0xc}, {0x3658bac, 0x4}, {0x30c6160, 0xa9}, 0x0, 0x0, 0x0, ...})
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/cmd/config.go:143 +0xc4
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/evcc-io/evcc/cmd.(*ConfigProvider).configure(0x32036c8, {{0x3658b70, 0xc}, {0x3658bac, 0x4}, {0x30c6160, 0xa9}, 0x0, 0x0, 0x0, ...})
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/cmd/config.go:126 +0x2c
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/evcc-io/evcc/cmd.configureSiteAndLoadpoints({{0x3658b70, 0xc}, {0x3658bac, 0x4}, {0x30c6160, 0xa9}, 0x0, 0x0, 0x0, 0x2540be400, ...})
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/cmd/setup.go:187 +0x30
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/evcc-io/evcc/cmd.run(0x21a1e20, {0x21d21f8, 0x0, 0x0})
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/cmd/root.go:161 +0x514
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/spf13/cobra.(*Command).execute(0x21a1e20, {0x30a4108, 0x0, 0x0})
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/go/pkg/mod/github.com/spf13/[email protected]/command.go:860 +0x644
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/spf13/cobra.(*Command).ExecuteC(0x21a1e20)
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/go/pkg/mod/github.com/spf13/[email protected]/command.go:974 +0x414
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/spf13/cobra.(*Command).Execute(...)
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/go/pkg/mod/github.com/spf13/[email protected]/command.go:902
Mär 21 09:05:55 raspberrypi evcc[5790]: github.com/evcc-io/evcc/cmd.Execute()
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/cmd/root.go:126 +0x20
Mär 21 09:05:55 raspberrypi evcc[5790]: main.main()
Mär 21 09:05:55 raspberrypi evcc[5790]: /home/runner/work/evcc/evcc/main.go:33 +0x30
Mär 21 09:05:55 raspberrypi systemd[1]: evcc.service: Main process exited, code=exited, status=2/INVALIDARGUMENT
Mär 21 09:05:55 raspberrypi systemd[1]: evcc.service: Failed with result 'exit-code'.

evcc.yaml.txt

I had opened a new topic #2932 but this has been closed.

[Pitillo]

I don't know why your issue got closed as it's definitively not the same as this one. From the log you provided here and in the issue #2932 I would say that that's a real big in evcc as a segfault is generated.
@premultiply Could you have a look into the issue again check my assumption and if so reopen the issue #2932? I don't see relations between #2932 and #2923 ...

[andig]

@gonny8 Du hast eine Fehlermeldung beim LGPCS. Lass uns das bitte in #2932 weiter diskutieren. Da gab es zuletzt eine Änderung im Nightly die ggf. Die Ursache sein könnte.

[gonny8]

Ja, gerne.
Ich hatte nur hier reingeschrieben, da mein Topic geschlossen wurde.

[andig]

Prinzipiell ist die manuelle Installation in /usr/local/bin ja richtig daher habe ich jetzt nochmal eine Warnung in der Doku hinzugefügt:

Achtung: Bei Umstellung auf die evcc Releases über apt muss der angelegte Service wieder entfernt werden. Anderenfalls kommt es zu Konflikten und Fehlern beim Start von `evcc.

[andig]

Kann jemand mit Sicherheit sagen, wo die services angelegt werden sollen?

[Pitillo]

Eigene Service Files soll man unter /etc/systemd/system/ ablegen, durch Pakete installierte service Dateien sollen unter /usr/lib/systemd/system/ abgelegt werden. Es spricht aus meiner Sicht jedoch nichts dagegen die Nutzer anzuweisen, die Datei ebenfalls unter /usr/lib/systemd/system abzulegen... Am Ende hat es nur etwas mit der priorisierung der Datei durch systemd zu tun. Dateien unter /etc werden gegenüber solchen in /run bevorzugt, wobei Dateien in /run denen in /lib bevorzugt werden (bzw diese eben von der Priorität her überschreibt).
Ähnlich verhält es sich mit der binary Location: User sollten ihre custom binarys unter /usr/local/bin ablegen, wogegen Pakete ihre binarys unter /usr/bin ablegen sollen. Aber es spricht natürlich nichts dagegen, dass der User sein binary ebenfalls unter /usr/bin ablegt...
Aus meiner Sicht sollten wir einfach die Doku ändern und es einheitlich halten und in dem preinst Script prüfen ob vorher eine manuelle Installation vorhanden war. Wenn dem so ist, sollte der User gefragt werden, ob die Dateien gelöscht und durch die des Pakets ersetzt werden sollen.

[Pitillo]

Hier ist das ganze erklärt:

       /etc
            System-specific configuration. This directory may or may not be read-only. 
            Frequently, this directory is pre-populated with vendor-supplied 
            configuration files, but applications should not make assumptions about 
            this directory being fully populated or populated at all, and should fall 
            back to defaults if configuration is missing.
 (…)
 VENDOR-SUPPLIED OPERATING SYSTEM RESOURCES
       /usr
            Vendor-supplied operating system resources. 
            Usually read-only, but this is not required. Possibly 
            shared between multiple hosts. This directory should not
            be modified by the administrator, except when installing 
            or removing vendor-supplied packages.

https://www.freedesktop.org/software/systemd/man/file-hierarchy.html

Hier noch eine Diskussion zu dem Thema.

[RTTTC]

Wenn dem so ist, sollte der User gefragt werden, ob die Dateien gelöscht und durch die des Pakets ersetzt werden sollen.

Also please let the users know that their existing config (yaml) will not get lost, as loosing configuration could be confusing/scary when faced with question about deleting something during install?

[andig]

I would be happy to take a PR as I can't thoroughly test this on Mac.

[andig]

Wer könnte helfen, #2966 zu testen?

[andig]

ping

[RTTTC]

I can't do it as I barely can manage my pi without help from you guys.
Well actually - I am unable to manage it without the help :D