If you are happy with SMS and twilio, then use it. Having many options is generally always awesome.
The main reason I made ntfy was because I mumbo jumbo (that's a technically term) on all sorts of servers for work and for fun, and I often do things in a tmux or screen that take forever, like rsync something (ha!) or reseed a MySQL replica. On many of those servers, I cannot or should not install anything, and I certainly don't want to leave an auth token in the bash history.
For those use cases, having a short command to ping me when something is done is FREAKING AWESOME. This has become muscle memory to me by now:
rsync -av bla root@myserver:/bla; curl -d "rsync done" ntfy.sh/philalerts
No auth token. No looong string of things. Just a simple curl.But again. If you're happy with SMS+twilio, then there's no reason to switch. Though I would love it if you could give it a shot.
I am merely an rsync lover like you.
In 2005 I named my company rsync.net because I thought it was descriptive of the service.
Prior to naming it that, I explicitly asked permission from both the original author of 'rsync' and the maintainers at the time. They said they had no problem with it.
I have not authored any piece of the rsync tool.
Some twisted soul could even put sending a message to you via cronrab every minute, and you don't seem to have any way to block it, or filter it from the rest.
I'd say that it's infinitely better to have an opaque an otherwise useless API token in your bash history than to have a URL of a bare and uncontrolled web service in the same bash history, and now on the open internet.
If I were to design a service like that, I'd have an URL like domain.api/username/${token}, where the token can be changed and / or retited, should it become a target of a spam attack. Exporting this token once in the beginning of a shell session should not be hard; SSH even allows you do it with SendEnv.
As for the retirable token, you are right. In the current approach this cannot be done. I have an open ticket for auth (optional) that I'll soon implement. I want to get it right, so if you have thoughts, I'd love input on it: https://github.com/binwiederhier/ntfy/issues/19
It's just like with passwords.
Edit: isn’t this super costly for personal use?
Not really ... I notice twilio deducting USD $9 every 10 or 12 days ? It's a business expense.
I also 'lookup' numbers to check if they are mobile or not:
curl -s -X GET "https://lookups.twilio.com/v1/PhoneNumbers/$number?Type=carrier&Type=caller-name" -u $accountsid:$authtoken | /usr/local/bin/jq '.'
... which is handy. Usually shows their name and carrier ID as well.As you can see, I pipe it through 'jq' because that's what all the cool kids do.
And if you're doing this before you try to SMS them, you can skip the lookup entirely, if it's a landline you'll just get an error back.
(I work at a company that uses Twilio heavily and it's full of this sort of microoptimization)
I have a group SMS with family where people are constantly sharing pictures etc and it is working fine.
My alerting regime is also built on 'curl' commands that generate SMS via twilio.
For example:
... and I am very happy with the ability to send SMS with both a configurable from and to number.However I am interested and open-minded - what are the benefits of using a push notification ?