Skip to content

Troubleshooting

Chris Caron edited this page Aug 13, 2019 · 38 revisions

General Troubleshooting

The best thing you can do when troubleshooting problems with your notification is to use the apprise object itself from the command line and add up the verbosity of the command (the more v's you specify, the more verbose it gets):

# In the below example, I am trying to figure out why my mailto:// line
# isn't working:
python apprise -vvv -t "test title" -b "test body" \
    mailto://user:[email protected]

The output can help you pinpoint what is wrong with your URL.

If the output appears cryptic, or you feel that you've exhausted all avenues, Don't be afraid to open a ticket and ask here. It greatly helps if you share the output received from your debug response. It might be just a simple tweak to your URL that is needed, otherwise we might have a good bug we need to solve.

Just be cautious as the debugging information can potentially expose personal information (such as your password and/or private access tokens) to the screen. Please remember to erase this or swap it with some random characters before posting such a thing publicly.

To Much Data and Overflow Directive

Out of the box, apprise passes everything you give it right along to the source. Some sources can handle a large surplus of data; others might not deal with it as well and just not post anything at all.

Apprise has a somewhat non-elegant way of handling this situation; but it's much better than nothing. You simply need to tack on the overflow switch somewhere in your Apprise URL; for example:

  • schema://path/?overflow=split
  • schema://path/?overflow=truncate
  • schema://path/?overflow=upstream
  • schema://path/?other=options&more=settings&overflow=split

The possible overflow= options are defined as:

Variable Description
split This will break the message body (and title if required) into as many smaller chunks (based on the confines of the service itself) and send out multiple messages in order to deliver everything.

For example, Twitter restricts public tweets to 280 characters. If your Apprise/Twitter URL was updated to look like twitter://<auth data>/?overflow=split, A message of say 1000 characters would be broken (and sent) via 4 smaller messages (280 + 280 + 280 + 160).
truncate This just ensures that regardless of how much content you're sending to an remote notification service, all of it's contents will be truncated to fit within a single message. This means that any data that exceeds the maximum message side will just be removed for the message itself.
upstream Let the upstream notification service handle all of the data passed to it. Apprise will not mangle/change it's content in any way.
Note: This is the default configuration option used when the overflow= directive is not set.

Please note that the overflow= option isn't a perfect solution:

  • It can fail for services like Telegram which can take content in the format of both HTML and Markdown. If you're using HTML, then there is a very strong possibility that both the upstream and/or truncate option could cut your message in the middle of an un-closed HTML tag. Telegram doesn't fair to well to this and in the past (at the time of writing this wiki entry) would error and not display the data.
  • It doesn't elegantly wrap/truncate messages the end of a word (near the message limits). It just makes a cut right at the limit itself. Similarly with ?overflow=split this hard cut happens throughout all of the messages needed to still send all of the content needed.
  • The overflow=split can work against you too, especially if you're using apprise to send thousands of log entries accidentally to you via a SMS message. Be prepared to get hundreds of text messages to re-construct all of the data you asked it to deliver! 😉

So while this is a viable solution for most services; consider that it may not work perfectly for all.

Special Characters and URL Conflicts

Apprise is built around URLs. Unfortunately URLs have pre-reserved characters it uses as delimiters that help distinguish one argument/setting from another.

For example, in a URL, the &, /, and % all have extremely different meanings and if they also reside in your password or user-name, they can cause quite a troubleshooting mess as to why your notifications aren't working.

Now there is a workaround: you can replace these characters with special %XX character-set (encoded) values. These encoded characters won't cause the URL to be mis-interpreted allowing you to send notifications at will.

Below is a chart of special characters and the value you should set them:

Character Escape Code Description
% %25 The percent sign itself is the starting value for defining the %XX character sets.
& %26 The ampersand sign is how a URL knows to stop reading the current variable and move onto the next. If this existed within a password or username, it would only read 'up' to this character. You'll need to escape it if you make use of it.
(a space) %20 While most URLs will work with the space, it's a good idea to escape it so that it can be clearly read from the URL.
/ %2F The slash is the most commonly used delimiter that exists in a URL and helps define a path and/or location.
@ %40 The at symbol is what divides the user and/or password from hostname in a URL. if your username and/or password contains an '@' symbol, it can cause the url parser to get confused.
+ %2B By default a addition/plus symbol (+) is interpreted as a space when specified in the URL. It must be escaped if you actually want the character to be represented as a +.
, %2C A comma only needs to be escaped in extremely rare circumstances where one exists at the very end of a specific URL that has been chained with others using a comma. See PR#104 for more details as to why you may need this.
: %3A A colon will never need to be escaped unless it is found as part of a user/pass combination. Hence in a url http://user:pass@host you can see that a colon is already used to separate the username from the password. Thus if your {user} actually has a colon in it, it can confuse the parser into treating what follows as a password (and not the remaining of your username). This is a very rare case as most systems don't allow a colon in a username field.
Clone this wiki locally