You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
we should have consistent behaviour between start-up time and running time. And preferably, firefly shouldn't crash due to one dependency throwing an error as other things could be in a usable state
better error surfacing for the errors
The text was updated successfully, but these errors were encountered:
Firefly crashes when the configured connector is not available on start up but not when running
This is true, and is working as designed. Now, we may want to change that design. The intention was that if FireFly wasn't configured correctly it would not start, forcing someone to intervene and reconfigure it correctly. As opposed to, if it was configured correctly and had been running for some time, but the connector suddenly became unavailable, it should not crash, but simply wait until it comes back. On startup, there is no way to know the difference between an improper config, or a connector that just isn't there temporarily, so we've taken the "fail fast" approach for now.
Thanks for providing the background. It does make sense for users to be aware of faulty configuration at an early stage so they can act on it.
I'm not sure whether multi-namespace support and dynamic configuration reload were factored in for this design. It feels the additions of those two critical features are worth us rethinking on that decision so that we have better isolations logic between namespaces (e.g. I don't have to fix all namespaces just to start Firefly UI to use the only one I care).
There are two aspects of this:
The text was updated successfully, but these errors were encountered: