A few ideas what might be wrong (my guess is No. 3):
Is http://app:8000 the correct server address? It should be if you use the standard docker compose setup.
Are both containers in the same docker network?
Have you configured DNS servers for your docker daemon? I always had these „Temporary failure in name resolution“ in my docker containers when the daemon did not have dns servers configured. See for example here: Fix Docker's networking DNS config
There is no mayan app container. I think this because of the big changes in the mayan docker-compose.yml that have happened recently. Where it’s using inheritance. I think the container name to use now would be
frontend
Here’s a list of all my Mayan containers.
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
110eab445cb6 mayanedms/mayanedms:s4.5 "/usr/local/bin/entr…" 23 hours ago Up 23 hours 8000/tcp mayan-worker_e-1
ff0ac3d0c8cb mayanedms/mayanedms:s4.5 "/usr/local/bin/entr…" 23 hours ago Up 23 hours 8000/tcp mayan-worker_c-1
c51835acde46 mayanedms/mayanedms:s4.5 "/usr/local/bin/entr…" 23 hours ago Up 23 hours 8000/tcp mayan-frontend-1
b6e8c9684380 mayanedms/mayanedms:s4.5 "/usr/local/bin/entr…" 23 hours ago Up 23 hours 8000/tcp mayan-worker_d-1
a961f172faa5 mayanedms/mayanedms:s4.5 "/usr/local/bin/entr…" 23 hours ago Up 23 hours 8000/tcp mayan-worker_b-1
e10f788a9d1a mayanedms/mayanedms:s4.5 "/usr/local/bin/entr…" 23 hours ago Up 23 hours 8000/tcp mayan-worker_a-1
568b6dec7e99 mayanedms/mayanedms:s4.5 "/usr/local/bin/entr…" 23 hours ago Up 23 hours 8000/tcp mayan-celery_beat-1
ea88cd42c93c redis:7.0.10-alpine "docker-entrypoint.s…" 23 hours ago Up 23 hours mayan-redis-1
1238fb1cc435 rabbitmq:3.11.13-management-alpine "docker-entrypoint.s…" 23 hours ago Up 23 hours mayan-rabbitmq-1
172fff33b355 drrsatzteil/mayan-mindee-web:latest "/entrypoint.sh web" 23 hours ago Up 23 hours mayan-mayan-mindee-web-1
33742aa3df30 elasticsearch:7.17.9 "/bin/tini -- /usr/l…" 23 hours ago Up 23 hours mayan-elasticsearch-1
48e8a0bc3fce drrsatzteil/mayan-mindee-worker:latest "/entrypoint.sh rq" 23 hours ago Up 23 hours mayan-mayan-mindee-worker-1
12e3da0caa92 postgres:13.10-alpine "docker-entrypoint.s…" 23 hours ago Up 23 hours mayan-postgresql-1
Yeah you are right it should be „frontend“ now. I have made this change on my instance as well but in my repo it unfortunately still says „app“.
I edited mayan’s .env file with this change
MAYAN_URL=http://frontend:8000/api/v4/
Then ran the curl test again, and still get an error message about trying to find the app container.
No you need to edit your docker-compose file: change the MAYAN_URL environment variable for the mindee worker container. Don’t forget to recreate the container afterwards.
That worked for the curl test.
Now I need to figure out why I’m not getting it to work in the workflow on Mayan.
I can see from the event viewer that it’s getting triggered.
Should I be using GET or POST for the HTTP method?
Either would work. Make sure that your url is correct:
You can also use conditions here like:
{% if entry_log.workflow_instance.document.document_type.label == “Bezügemitteilung” %}http://mayan-mindee-web:8000/payroll/{{ entry_log.workflow_instance.document.pk }}{% elif entry_log.workflow_instance.document.document_type.label == “Rechnung” %}http://mayan-mindee-web:8000/invoice/{{ entry_log.workflow_instance.document.pk }}{% elif entry_log.workflow_instance.document.document_type.label == “Vertragsunterlagen” %}http://mayan-mindee-web:8000/general/{{ entry_log.workflow_instance.document.pk }}{% elif entry_log.workflow_instance.document.document_type.label == “Befund” %}http://mayan-mindee-web:8000/general/{{ entry_log.workflow_instance.document.pk }}{% else %}http://mayan-mam-web:8000/{{ entry_log.workflow_instance.document.pk }}{% endif %}
Thinking about it this would also work of course as long you specified that the .env file is used by your mindee worker container.
I’ve finally made some progress, after fixing some silly mistakes on my part.
I can get Mayan to send the HTTP request, and I can see both mindee containers processing the document in their logs. When I use a standard API.
But when I try to use our custom API, I don’t even see the web container receiving the document in it’s log.
I’m sure it’s because I did something wrong in the api.json or service.py files. I have them both listed below. Can you tell me what I did wrong?
Thanks
api.json
"custom": [
{
"documenttype": "Mindee",
"account": "MyCompany",
"endpoint": "invoice-12-demo",
"pagelimit": 1,
"storeocr": "mindee",
"metadata": {
"BillToCustomer": {
"fieldname": "sold_to",
"overwrite": true
},
"InvoiceNumber": {
"fieldname": "order_number",
"overwrite": true
},
"OrderDate":{
"fieldname": "date",
"overwrite": true
},
"OrderPONumber": {
"fieldname": "customer_po_number",
"overwrite": true
},
"SerialNumbers": {
"fieldname": "serial_number",
"overwrite": true
},
"ShipToCustomer": {
"fieldname": "ship_to",
"overwrite": true
},
"description": {
"fieldname": "description",
"overwrite": true
}
}
}
],
service.py
@app.route("/invoice-12-demo/<int:document_id>", methods=["GET", "POST"])
def trigger_invoice-12-demo(document_id):
q.enqueue(process_custom, document_id, "invoice-12-demo")
return "OK"
I think I might have led you on a wrong track here when I gave you the advice to mount the changed service.py file. I realised that I install the package without the -e flash when I’m building the container. Could you please try the following just to validate that I’m on the right track here. Please exec in your web container with your changed file mounted:
python -m pip install /app
If you can use your endpoint after that we can think about a more permanent solution.
Unfortunately the web container log does not show any documents being processed after following your above directions.
Could you please stop and start the container? Don’t use “compose down” though, just stop and start please 
Just to double check. Do you want me to use this command?
docker compose restart mayan-mindee-web
Exactly 
Just restart the container after the install of your changed files so that the changes should hopefully take effect 
Restarted the container, then added a new document.
Nothing shows up in the log. 
I just added a generic api endpoint:
/custom/<api_name>/int:document_id
Give it a few minutes to build than you can pull the new image and try that. I haven’t tested it though but I hope it should work 
Should I disable my mount for service.py?
It should not really matter but you might as well disable it. Make sure that you use the charged url to trigger (you might want to try to use curl first)
Edit: hmm, the latest built was apparently too long ago… let me check…
Are you making this custom api, to match the api.json I provided above?
Because I redacted that for public display.
No it should be usable for any custom api which makes more sense anyway. So this was something on my list anyway.
The build is finished btw. You may pull the new image now.