Uptime has never really been Heroku's strong suit. I wonder why we've never heard a meta-analysis from them as far as what they could do to improve it.
To clarify, this incident did not impact running applications. It only affected the ability to make changes to running apps. AKA, this incident did not affect application uptime.
It affected my ability to run commands on production applications. The fact that the status was degraded only for 'development' was, in my mind, disingenuous. Production should should have been orange, not green.
I interpreted your original comment, "Uptime has never really been Heroku's strong suit", to refer to application uptime. I guess you're actually referring to something different.
AFAIK, when any other provider speaks about "uptime", they're referring to whether or not the service they're selling is up and running. For AWS, that'd be whether your instance or ELB is up and running as expected. For Heroku, it's whether your app is up and running.
Issues in the control plane that affect your ability to make changes to your resources are generally outside the scope of any "service uptime" numbers you see published (unless that uptime is specific to the API/control plane).
Edit: We do our best to publish info for both the service, and for its control plane. I would agree that our status site categories ("production" vs. "development") might not be the best way to label this split. Previously we used "app operations" and "tools". We may yet change it again.
Agreed, this was confusing to me as well, especially since there is no real concept of "development" on the heroku platform. you can't separate your dev/test apps from your production instances in heroku, they are all just treated as "production". might be helpful to clarify this on the status page at some point.
12 comments
[ 3.7 ms ] story [ 33.6 ms ] threadhttps://status.heroku.com/incidents/543
Feel free to view Heroku's 12-month historical uptime data here: https://status.heroku.com/uptime
AFAIK, when any other provider speaks about "uptime", they're referring to whether or not the service they're selling is up and running. For AWS, that'd be whether your instance or ELB is up and running as expected. For Heroku, it's whether your app is up and running.
Issues in the control plane that affect your ability to make changes to your resources are generally outside the scope of any "service uptime" numbers you see published (unless that uptime is specific to the API/control plane).
Edit: We do our best to publish info for both the service, and for its control plane. I would agree that our status site categories ("production" vs. "development") might not be the best way to label this split. Previously we used "app operations" and "tools". We may yet change it again.
https://status.heroku.com/incidents/543
Nobody's app stopped running because of this.
Edit: Thanks for correcting it!