Previously with failed update whole replacing release were
deleted which wrongly delete app which weren't modified. Now
simply deleting changed apps. (HandleHealthCheckStep)
When checking health, check all apps in an replacing release
instead of marking release healthy if one return true. (HealthCheckStep)
Polish and add more debug.
Now that failed update works correctly and after that we can
again deploy stream again, one thing is left to another issue which
is an annoying thing that stream is still showing failed as last
status in a history.
Fixes #966
Fixes #964
To test, start skipper with 100s timeout as it defaults to 300s
To test, start skipper with 100s timeout as it defaults to 300s
Do deploy and fail
time
with update by using wrong probe path:After deploy:
During update you see:
After failed update: