Signal / Whatsapp integration

[pmatuszy]

... and yest another question:
how to do the integration with Signal / Whatsapp? I can't see any section in the documentation explaining this...
Paul

[cuu508]

Signal: https://healthchecks.io/docs/self_hosted_configuration/#SIGNAL_CLI_SOCKET

WhatsApp: see TWILIO_* settings: https://healthchecks.io/docs/self_hosted_configuration/#TWILIO_ACCOUNT

For Signal, you will first want to get familiar with running signal-cli, and sending messages from command line. I would start with reading signal-cli README, man page and Wiki pages.

Healthchecks uses Twilio to send WhatsApp messages. You specify Twilio API credentials and phone number in the TWILIO_ env variables and off you go. But there's a non-trivial onboarding process you have to go through in your Twilio account to enable WhatsApp message sending. I'm not sure if it's even possible if you're an individual, not a company.

[pmatuszy]

Thank you very much for your answer.

I am pretty familiar with signal-cli. I set it to start on the file socket. I've checked and it works there.
When I add integration I get HTTP 500 ERROR which in debug mode show this:

Django Version:	4.1.2
Exception Type:	TypeError
Exception Value:	
argument of type 'NoneType' is not iterable
Exception Location:	/encrypted/webapps/healthchecks/hc/api/transports.py, line 49, in get_nested
Raised during:	hc.front.views.send_test_notification
Python Executable:	/encrypted/webapps/hc-venv/bin/python
Python Version:	3.10.6
Python Path:	
['/encrypted/webapps/healthchecks',
 '/encrypted/webapps/hc-venv/bin',
 '/usr/lib/python310.zip',
 '/usr/lib/python3.10',
 '/usr/lib/python3.10/lib-dynload',
 '/encrypted/webapps/hc-venv/lib/python3.10/site-packages']
Server time:	Wed, 21 Dec 2022 08:35:02 +0000

integration is done with my cell phone number - I've double-checked it and it looks good

Have you got any idea why I can't get this integration working?

[cuu508]

Not 100% sure but I think what happens is signal-cli RPC returns a message in a format that Healthchecks does not expect, and Healthchecks throws an exception.

I'd like to fix Healthchecks to handle it gracefully, and provide an useful error message to the user. I just created an issue – #763.

Would you be able to add a debug statement in the code, to print out the message that Healthchecks receives from signal-cli? In hc/api/transports.py, around line 904 add print(reply):

before:

        for reply_bytes in self._read_replies(payload_bytes):
            try:
                reply = json.loads(reply_bytes.decode())
            except ValueError:
                raise TransportError("signal-cli call failed (unexpected response)")

            if reply.get("id") == payload["id"]:
                if "error" not in reply:
                    # success!
                    break

after:

        for reply_bytes in self._read_replies(payload_bytes):
            try:
                reply = json.loads(reply_bytes.decode())
            except ValueError:
                raise TransportError("signal-cli call failed (unexpected response)")

            print(reply)  # <--- add this
            if reply.get("id") == payload["id"]:
                if "error" not in reply:
                    # success!
                    break

[pmatuszy]

ok - added.

it fails on line # 49

and my line 49 is below:

will it help you?

[cuu508]

After adding the print statement, and triggering the error again, you should see the RPC message from signal-cli printed in the web server's stdout. Can you post it here?

[pmatuszy]

ok sorted. The problem was that I was starting signal-cli without "-a" parameter.
Lessons learned from the exercise:

1. how to start signal-cli?

signal-cli -a +4xxxaaabbbb daemon --socket /tmp/signal-socket

2. set ACL for that file so the user under which credentials WWW server starts has privs to that file

setfacl -m u:USERNAME:rw /tmp/signal-socket

having this in place integration works

Thank you once again for your great, quick and effective support!

[cuu508]

Awesome, happy to hear that.

I just fixed the Signal integration to handle the missing "-a" scenario better. It will now return an error message that says: "Could not send a test notification. signal-cli call failed (-32602)."

And, semi-related, I added the ADMINS environment variable. If you specify an email address in this variable, Healthchecks will send error reports to this address when DEBUG=False. And so, if you hit a HTTP 500 again, you will not have to turn DEBUG on just to see the error details.

[cuu508]

Started a discussion about making "-a" optional here: AsamK/signal-cli#1132