Unreliable notifications / notification does not retry when a network error occurs

[davidjirovec]

I use notifications via email and sometimes, although rarely, it happens that notification sending fails, for example:

2023-11-20 18:03:51 DEBUG:apprise:Delivery: smtp.gmail.com:587
2023-11-20 18:03:51 DEBUG:apprise:Connecting to remote SMTP server...
2023-11-20 18:03:51 DEBUG:apprise:Securing connection with STARTTLS...
2023-11-20 18:03:51 DEBUG:apprise:Applying user credentials...
2023-11-20 18:04:19 WARNING:apprise:Sending email to "[email protected]" failed. Reason: Connection unexpectedly closed: The read operation timed out
2023-11-20 18:04:19 ERROR:apprise:Unhandled Notification Exception
2023-11-20 18:04:19 Traceback (most recent call last):
2023-11-20 18:04:19   File "/usr/local/apprise/Apprise.py", line 565, in _notify_sequential
2023-11-20 18:04:19     result = server.notify(**kwargs)
2023-11-20 18:04:19              ^^^^^^^^^^^^^^^^^^^^^^^
2023-11-20 18:04:19   File "/usr/local/apprise/plugins/NotifyBase.py", line 297, in notify
2023-11-20 18:04:19     the_calls = [self.send(**kwargs2) for kwargs2 in send_calls]
2023-11-20 18:04:19                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2023-11-20 18:04:19   File "/usr/local/apprise/plugins/NotifyBase.py", line 297, in <listcomp>
2023-11-20 18:04:19     the_calls = [self.send(**kwargs2) for kwargs2 in send_calls]
2023-11-20 18:04:19                  ^^^^^^^^^^^^^^^^^^^^
2023-11-20 18:04:19   File "/usr/local/apprise/plugins/NotifyEmail.py", line 829, in send
2023-11-20 18:04:19     return self.submit(messages)
2023-11-20 18:04:19            ^^^^^^^^^^^^^^^^^^^^^
2023-11-20 18:04:19   File "/usr/local/apprise/plugins/NotifyEmail.py", line 895, in submit
2023-11-20 18:04:19     socket.quit()
2023-11-20 18:04:19   File "/usr/local/lib/python3.11/smtplib.py", line 1004, in quit
2023-11-20 18:04:19     res = self.docmd("quit")
2023-11-20 18:04:19           ^^^^^^^^^^^^^^^^^^
2023-11-20 18:04:19   File "/usr/local/lib/python3.11/smtplib.py", line 431, in docmd
2023-11-20 18:04:19     self.putcmd(cmd, args)
2023-11-20 18:04:19   File "/usr/local/lib/python3.11/smtplib.py", line 378, in putcmd
2023-11-20 18:04:19     self.send(f'{s}{CRLF}')
2023-11-20 18:04:19   File "/usr/local/lib/python3.11/smtplib.py", line 365, in send
2023-11-20 18:04:19     raise SMTPServerDisconnected('please run connect() first')
2023-11-20 18:04:19 smtplib.SMTPServerDisconnected: please run connect() first

Whatever, the connection can fail. Problem is that the error is ignored and I never get the message.

Yes I could periodically check the changedetection ui to not to miss anything, but I really prefer to get notified instead.

Any better solution to this?

[dgtlmoon]

It looks like your network is failing to connect to gmail from time to time

[davidjirovec]

Yes, I agree. But I don't think it should be silently ignored. How about storing "notification should be sent" data along with other change detected data in one transaction. Then send the notification and if successful, mark as sent. If sending failed, try again later.

This approach would also make the batching mentioned in #1885 possible.

[dgtlmoon]

It should go into a queue system, if its stored in memory you will lose it if the app restarted

I was thinking there could be in the data-dir sub-directory called notifications with a list of JSON files, each JSON file is the notification, only when its sent without-exception that it will remove that JSON file

When a watch needs to make a notification, it will drop the notification into that directory

It will scan the directory for new notifications say every 5 seconds or something