No Pypilot data into Signalk Server after OS Backup swap

[aitonos]

HI!

Let's see if I can get some help here.

Yesterday I had a NVME failure and fortunately I had a very fresh backup done 2 days ago. So all fine in this sense. Backup had been previously done using SD CARD COPIER. Just as the standard procedure.

System started perfectly but some Derived data sentences don't work,( this is quite typical),and also those about topic of this thread: No incoming IMU data from the dedicated PYpilot CPU by NAVITOP (Great hardware)

I checked all I could in SignalK server, and I had 2 different WS Pypilot connections (had them too when I did the original backup), and had no new access requests; so I deleted both connections with the hope that Pypilot would also request access.  After deleting this, no access requests and no incoming data either.

I left both Pypilot CPU and PI SK Server running overnight and this morning I had an access request from Pypilot; I granted read/write permissions and Pypilot CPU is already getting data from SK server; But the SK Server has no WS connection from Pypilot so not gettinmg data from its IMU. If it serves, I just have a Ws-Openplotter.local:3000 connection (Websocket related)

How can I fix this situation in order to get the IMU data back into my SignalK server?

Thanks!

[seandepagnier]

perhaps posting the pypilot log would offer some hints?

journalctl -u pypilot

[aitonos]

perhaps posting the pypilot log would offer some hints?

journalctl -u pypilot

Thanks, Sean:

In The PI5 running the SK server:

pi@openplotter:~ $ journalctl -u pypilot               
-- No entries --

If I SSH to Pypilot dedicated CPU:

pi@openplotter:~ $ ssh tc@192.168.1.115

tc@192.168.1.115's password:
tinypilot is a Sailboat Autopilot Distribution for raspberry pi

  ( '>')
  /) TC (\  Core is distributed with ABSOLUTELY NO WARRANTY.
(/-_--_-\)          www.tinycorelinux.net

tc@pypilot:~$ journalctl -u pypilot
-sh: journalctl: not found




Just in case it serves IMU in mac Arthur Hat in Pi5 is disabled, and when I did the backup it was getting navigation attitude data from Tinypilot computer.
Pî5 can see Tinypilot and it replies PING
Both have dedicated IP addresses
Tinypilot gets NMEA DATA
Maybe is a Signal K issue as I don't get authorization requests, but in SK server Allow New Device Registration  is marked as YES
MDNS is also active

OK, thanks!

[seandepagnier]

oh on tinypilot the logs are in /var/log/pypilot/current

if you run pypilot_client on the pi5 does it detect and connect automatically using mdns?

[aitonos]

oh on tinypilot the logs are in /var/log/pypilot/current

if you run pypilot_client on the pi5 does it detect and connect automatically using mdns?

Pypilot client, when IMU enabled, shows data in SK server. If I disable it, data disappears.

I think the problem is that it doesnt request access to SK server. This is why i start to thing is SK server related, and should maybe to appeal for asolution in Discord.

Anyway, thanks for your help, this is the log:

192.168.1.115 is Tinypilot. 
Pi5 is on 192.168.1.104 but seems that there is no trace there.

pypilot._tcp.local.', name='pypilot._pypilot._tcp.local.', addresses=[b'\xc0\xa8
\x01s'], port=23322, weight=0, priority=0, server='pypilot._pypilot._tcp.local.'
, properties={'version': '0.56'}, interface_index=None)} 192.168.1.115
2024-09-12_12:51:24.89665 zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
2024-09-12_12:51:26.27146 zeroconf exception:
2024-09-12_12:51:26.27759 e is {'127.0.0.1': ServiceInfo(type='_pypilot._tcp.loc
al.', name='pypilot._pypilot._tcp.local.', addresses=[b'\x7f\x00\x00\x01'], port
=23322, weight=0, priority=0, server='pypilot._pypilot._tcp.local.', properties=
{'version': '0.56'}, interface_index=None), '192.168.1.115': ServiceInfo(type='_
pypilot._tcp.local.', name='pypilot._pypilot._tcp.local.', addresses=[b'\xc0\xa8
\x01s'], port=23322, weight=0, priority=0, server='pypilot._pypilot._tcp.local.'
, properties={'version': '0.56'}, interface_index=None)} 192.168.1.115
2024-09-12_12:52:26.32791 zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
2024-09-12_12:52:27.69399 zeroconf exception:
2024-09-12_12:52:27.69446 e is {'127.0.0.1': ServiceInfo(type='_pypilot._tcp.loc
al.', name='pypilot._pypilot._tcp.local.', addresses=[b'\x7f\x00\x00\x01'], port
=23322, weight=0, priority=0, server='pypilot._pypilot._tcp.local.', properties=
{'version': '0.56'}, interface_index=None), '192.168.1.115': ServiceInfo(type='_
pypilot._tcp.local.', name='pypilot._pypilot._tcp.local.', addresses=[b'\xc0\xa8
\x01s'], port=23322, weight=0, priority=0, server='pypilot._pypilot._tcp.local.'
, properties={'version': '0.56'}, interface_index=None)} 192.168.1.115

~

[seandepagnier]

Ok, hard to say if there is a bug maybe in pypilot how it requests access?

Did you try with an alternate install on the pi5?

Also there is a debug function in pypilot/pypilot/signalk.py. If you uncomment the print there, then:
. pypilot.build

Then when you restart pypilot "sudo sv r pypilot" or reboot there should be much more signalk logging info.

I am not sure what log file you posted is that /var/log/pypilot/current ?

[aitonos]

Ok,  hard to say if there is a bug maybe in pypilot how it requests access?

Did you try with an alternate install on the pi5? 

Also there is a debug function in pypilot/pypilot/signalk.py.  If you uncomment the print there, then:
. pypilot.build

Then when you restart pypilot "sudo sv r pypilot" or reboot there should be much more signalk logging info.

I am not sure what log file you posted is that /var/log/pypilot/current ?

I will try but I think is too much for my knowledge!  

[aitonos]

Ok,  hard to say if there is a bug maybe in pypilot how it requests access?

Did you try with an alternate install on the pi5? 

Also there is a debug function in pypilot/pypilot/signalk.py.  If you uncomment the print there, then:
. pypilot.build

Then when you restart pypilot "sudo sv r pypilot" or reboot there should be much more signalk logging info.

I am not sure what log file you posted is that /var/log/pypilot/current ?

I will try but I think is too much for my knowledge!  

Hi again,  Sean, finally I’ve been investigating this myself (with some help from GPT which made me finally brave , hahaha) and following your advice. I’m sure I may be missing something obvious, so I wanted to share the evidence I’m seeing and ask for your guidance.

---

Network / setup (key point)

• My main Signal K server is 192.168.1.104:3000 (this is the “central” SK I want pypilot to register with).
  
• On the same LAN I also have a Victron/Venus system, which appears on the logs as
  
  Code:

  Victron._http._tcp.local
  via zeroconf. ( IT SEEMS THBIS IS THE PROBLEM AND I MISSED TO TELL YOU)
  
• The problem seems to be that pypilot keeps picking the Victron zeroconf service and never reaches the main Signal K server at 192.168.1.104:3000 to request access.
  

---

What I did
Following your suggestions and continuing on my own:

• Restarted pypilot multiple times and inspected
  
  Code:

  /var/log/pypilot/current
  
  
• Verified the SK web endpoint responds on the Pi5 / main SK server:
  
  Code:

  http://192.168.1.104:3000
  (I get a 302 redirect to
  
  Code:

  /admin/
  , so the server is alive)
  
• Forced Signal K configuration explicitly in
  
  Code:

  pypilot.conf
  :
  

Code:

signalk.enabled=true
signalk.host=192.168.1.104
signalk.port=3000
signalk.uid="pypilot-84025861262"

• Checked the Signal K server (192.168.1.104) and there are never any access requests coming from pypilot.
  

---

Relevant log excerpts (as requested)
From

Code:

/var/log/pypilot/current

, repeated pattern:

Code:

signalk process 11729
signalk failed to read token /home/tc/.pypilot/signalk-token

And repeatedly:

Code:

signalk zeroconf service add Victron._http._tcp.local. _http._tcp.local.

Plus continuous zeroconf activity:

Code:

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

What I do not see in the logs (even after restart):

• Any HTTP attempt to
  
  Code:

  192.168.1.104:3000
  
  
• Any OAuth / access request flow
  
• Any token creation
  
• Any error related to contacting
  
  Code:

  192.168.1.104:3000
  
  

---

Key points (my interpretation — please correct me if I’m wrong)

• Even with
  
  Code:

  signalk.host=192.168.1.104
  set in config, pypilot still runs zeroconf discovery.
  
• It seems to “lock onto” the first relevant
  
  Code:

  _http._tcp.local
  service it sees (the Victron/Venus one).
  
• After that, it never proceeds to contact the configured Signal K host (192.168.1.104:3000), so the main SK server never receives an access request.
  

I’m not claiming this is definitely what’s happening — just describing what the logs suggest from my side.

---

Conclusion
It looks like zeroconf discovery may be taking precedence over the explicitly configured Signal K host, at least in a LAN where a Victron/Venus device advertises

Code:

_http._tcp.local

.
If this is expected behaviour, could you advise how to force/priority-connect to the main SK server at

Code:

192.168.1.104:3000

?
I know this is a quite unique ( or not so unique) setup in a boat, maybe it would be nice to add an option to choose the SK server to point at to request access or leave it automatic ( when empty) like for NMEA.
Thanks a lot for your time and for maintaining pypilot — and apologies if I’m missing something obvious.

---

Full relevant log excerpt (

Code:

/var/log/pypilot/current

) 
signalk process 2316
signalk failed to read token /home/tc/.pypilot/signalk-token

signalk zeroconf service add Victron._http._tcp.local. _http._tcp.local.

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

signalk process 11061
signalk failed to read token /home/tc/.pypilot/signalk-token

signalk zeroconf service add Victron._http._tcp.local. _http._tcp.local.

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

signalk process 11217
signalk failed to read token /home/tc/.pypilot/signalk-token

signalk zeroconf service add Victron._http._tcp.local. _http._tcp.local.

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

signalk process 11485
signalk failed to read token /home/tc/.pypilot/signalk-token

signalk zeroconf service add Victron._http._tcp.local. _http._tcp.local.

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

signalk process 11729
signalk failed to read token /home/tc/.pypilot/signalk-token

signalk zeroconf service add Victron._http._tcp.local. _http._tcp.local.

zeroconf addresses ['127.0.0.1', '192.168.1.115'] 2
zeroconf exception:

---

What is notably absent from the log
(no entries found)
- http://192.168.1.104:3000
- any HTTP request
- any OAuth / access request
- any token creation
- any connection error to 192.168.1.104:3000

(Forced configuration I used)

signalk.enabled=true
signalk.host=192.168.1.104
signalk.port=3000
signalk.uid="pypilot-84025861262"

---

Additional verification (outside pypilot)

Code:

$ wget -S -O - http://192.168.1.104:3000/

HTTP/1.1 302 Found
Location: /admin/

This confirms the main Signal K server is reachable and alive, but pypilot never attempts to connect.

[aitonos]

BTW, I am not in my boat,  I'm doing all this through VNC, but I have no remote acces to the VENUS OS  to disconnect it and see if Then I get the request in the main server.

[aitonos]

Hi again, 

I was intrigued to see if to have 2 running SK servers was the reason and got  atime to come to the boat, and yes, I switched off the VENUS OS running in my PI3 and rebooted the PI5 with main SK server and data came automatically. ( So it was keeping the old token as didn't have to request new access)

So we ( at least me) have learnt a new story!

Is there a way that I can force Tinypilot CPU to connect the real SK server?

Thanks!