Client wont keep Modules up-to-date

[codac]

I’m running @karsten13 Docker on my Synology NAS, which runs smooth and nicely.
As I have configured my MagicMirror to work as client/server, the client runs on a Raspberry Pi feeding the Mirror itself.

I have figured out that the client wont show the most current data of the modules. It is not frozen, the time is running and other modules like “MMM-DWD-WarnWeather” or “MMM-Traffic” are always up-to-date.
But the default calendar module as well as the “MMM-SoccerLiveScore” module are always outdated.

How can that be and what can I do to find the root cause?

[karsten13]

@codac

you are talking about this setup?

I think all modules are running correctly which have no node_helper.js in their directory.

Are there any warnings/errors in the logs (client and/or server)?

[sdetweil]

@codac can u ssh into the mirror?

stop the client and open chromium and point it at your mm server.

this will do the same thing as the client just use a different browser.

when I do this for pi0 devices I use this command line

chromium -noerrdialogs -kiosk -start_maximized  --disable-infobars --app=http://localhost:$port  --ignore-certificate-errors-spki-list --ignore-ssl-errors --ignore-certificate-errors 2>/dev/null

change the localhost and $port to point to your mm server

[codac]

@karsten13
yes and using pm2 to start this command:

DISPLAY=:0 node clientonly --address 192.168.0.13 --port 8080

And yes I have errors, it seems as if there is a problem using the 127.0.0.1 address and port 8080:

0|mm       | [13.07.2021 13:19.37.499] [ERROR] Error: listen EADDRINUSE: address                                                                                                                                           already in use 127.0.0.1:8080
0|mm       |     at Server.setupListenHandle [as _listen2] (net.js:1313:16)
0|mm       |     at listenInCluster (net.js:1361:12)
0|mm       |     at GetAddrInfoReqWrap.doListen [as callback] (net.js:1498:7)
0|mm       |     at GetAddrInfoReqWrap.onlookup [as oncomplete] (dns.js:68:8) {
0|mm       |   code: 'EADDRINUSE',
0|mm       |   errno: 'EADDRINUSE',
0|mm       |   syscall: 'listen',
0|mm       |   address: '127.0.0.1',
0|mm       |   port: 8080
0|mm       | }
0|mm       | [13.07.2021 13:19.37.506] [ERROR] MagicMirror will not quit, but it                                                                                                                                           might be a good idea to check why this happened. Maybe no internet connection?
0|mm       | [13.07.2021 13:19.37.510] [ERROR] If you think this really is an is                                                                                                                                          sue, please open an issue on GitHub: https://github.com/MichMich/MagicMirror/iss                                                                                                                                          ues

and I wonder why there is an error catching the newsfeed, as the stated feed is the one from the client, not the server:

0|mm       | /home/pi/MagicMirror/node_modules/electron/dist/electron exited wit                                                                                                                                          h signal SIGINT
0|mm       | [13.07.2021 19:59.53.918] [ERROR] Newsfeed Error. Could not fetch n                                                                                                                                          ewsfeed:  https://rss.nytimes.com/services/xml/rss/nyt/HomePage.xml FetchError:                                                                                                                                           request to https://rss.nytimes.com/services/xml/rss/nyt/HomePage.xml failed, rea                                                                                                                                          son: getaddrinfo EAI_AGAIN rss.nytimes.com
0|mm       |     at ClientRequest.<anonymous> (/home/pi/MagicMirror/node_modules                                                                                                                                          /node-fetch/lib/index.js:1461:11)
0|mm       |     at ClientRequest.emit (events.js:315:20)
0|mm       |     at TLSSocket.socketErrorListener (_http_client.js:426:9)
0|mm       |     at TLSSocket.emit (events.js:315:20)
0|mm       |     at emitErrorNT (internal/streams/destroy.js:92:8)
0|mm       |     at emitErrorAndCloseNT (internal/streams/destroy.js:60:3)
0|mm       |     at processTicksAndRejections (internal/process/task_queues.js:8                                                                                                                                          4:21) {
0|mm       |   type: 'system',
0|mm       |   errno: 'EAI_AGAIN',
0|mm       |   code: 'EAI_AGAIN'
0|mm       | }
0|mm       | [13.07.2021 19:59.56.921] [ERROR] Calendar Error. Could not fetch c                                                                                                                                          alendar:  http://www.calendarlabs.com/ical-calendar/ics/76/US_Holidays.ics Fetch                                                                                                                                          Error: request to http://www.calendarlabs.com/ical-calendar/ics/76/US_Holidays.i                                                                                                                                          cs failed, reason: getaddrinfo EAI_AGAIN www.calendarlabs.com
0|mm       |     at ClientRequest.<anonymous> (/home/pi/MagicMirror/node_modules                                                                                                                                          /node-fetch/lib/index.js:1461:11)

[sdetweil]

@codac said in Client wont keep Modules up-to-date:

EAI_AGAIN

is a name server problem

the api request could not resolve the name

see this a lot with local pihole servers doing DNS filtering

[karsten13]

@codac

it seems as if there is a problem using the 127.0.0.1 address and port 8080:

yes, port 8080 on your client is already in use.
You can try another port in the client config.js but I don’t know if this works.

I wonder why there is an error catching the newsfeed, as the stated feed is the one from the client, not the server:

then remove this stuff from the client config …

I’m not familiar with the clientonly setup and I think this is not used very often and therefore maybe contains some bugs.

[sdetweil]

@karsten13 said in Client wont keep Modules up-to-date:

I’m not familiar with the clientonly setup

it just launches electron pointing at the server, same as u would do with chromium.

doesn’t use the config.js

[codac]

I figured out that is was due to several reasons.

1. The reported error messages from the client side were old ones from the time when I set up the client and before I switched to clientonly mode. So they are supposed to be ignored.
2. The calendar module was able to receive the events from all calendars except for one calendar that runs on my NAS system where I forgot to add the port number to the calendar url.
3. the module MMM-Tankerkoenig is not up to date as there is an error message regarding the identifier that was not found. As far as I can see the code has an error and therefore the module wont update correctly.
4. the module MMM-Formula1 (wrongly called it MMM-SocerLiveScore) module has a ReloadInterval which was just wrongly set up and far to long.

@karsten13
In the Docker on the server side is there anything adequate to the command pm2 logs --lines 300 mm ?

[sdetweil]

@codac said in Client wont keep Modules up-to-date:

In the Docker on the server side is there anything adequate to the command pm2 logs --lines 300 mm ?

wouldnt you just

 docker exec container_id pm2 logs.....

to execute the command inside the container?

to clear the logs

docker exec container_id pm2 flush

[karsten13]

@sdetweil

there is no pm2 in the container …

The command is docker logs <containername or id>, so if you use my docker-compose setup docker logs mm.

[sdetweil]

@karsten13 so I have to restart the container after making config changes?

[codac]

@sdetweil yes, you should do that.

[codac]

I have still not fixed this.

Servers IP: 192.168.0.23 (local Port 8036, container-port: 8080)

On the client I get the following error:

Unable to connect to server: (Error: Unable to read config from server (http://192.168.0.23:8036/config/ (connect ECONNREFUSED 192.168.0.23:8036)

The servers config is set to:

        address: "0.0.0.0",
	port: 8080,
	basePath: "/", 	
	ipWhitelist: ['0.0.0.0/0', '127.0.0.1', '::ffff:127.0.0.1', '::1', '192.168.0.0/24', '172.17.0.0/24'], 

Any ideas why the client wont get the servers config?

[sdetweil]

@codac first, remove the ipWhitelist , set to []

then try
.and the IP address used is the docker host IP address, right.

never use the containers direct address for any code not running ON the docker host machine

where is client running?

if mm rejected the connection due to whitelist you should have seen it in the mm messages

i see a message like this

Access denied to IP address: 192.168.2.106

my config

	address: "192.168.2.106", 	// Address to listen on, can be:
							// - "localhost", "127.0.0.1", "::1" to listen on loopback interface
							// - another specific IPv4/6 to listen on a specific interface
							// - "0.0.0.0", "::" to listen on any interface
							// Default, when address config is left out or empty, is "localhost"
	port: 8090,
	basePath: "/", 	// The URL path where MagicMirror is hosted. If you are using a Reverse proxy
					// you must set the sub path here. basePath must end with a /
	ipWhitelist: ["127.0.0.1"], 

[karsten13]

@codac

may you must open the additional port in the docker setup and add the line with 8036 (untested) in your docker-compose.yml:

    ports:
      - "8080:8080"
      - "8036:8036"

[sdetweil]

@karsten13 said in Client wont keep Modules up-to-date:

ports:
  - "8080:8080"
  - "8036:8036"

why wouldn’t

     ports:
           "8036:8080"

be enough?

[karsten13]

@sdetweil think you are right …

[sdetweil]

@karsten13 actually I think you were closer…

I know a module that uses the port number in a url it presents…
(MMM-Config in the QR code)

8080 won’t work, but if the mapping was added

so

     ports:
           "8036:8080"
           "8080:8080"

would be needed

and if u need to do that, then 8080 on the docker host needs to be used, so then no need for 8036

     ports:
           "8080:8080"

but can’t run two container instances

[sdetweil]

@sdetweil also, my MM uses 8090 not to collide with another app.

[karsten13]

@codac

Servers IP: 192.168.0.23 (local Port 8036, container-port: 8080)

if I read this correctly your setup is

ports:
  - "8036:8080"

So mm should be reachable under 192.168.0.23:8036.

If not the client ip is may not white listed, I would try ipWhitelist: [], in the server config for testing.

Another question: Is 192.168.0.23:8036 not reachable in general (e.g. in a browser) or only not reachable running the clientonly command?