fix: wait out a passing outage while following a deployment

This commit is contained in:
Eric Wendland 2026-10-10 22:52:30 +02:00
commit 082b7d14b8
3 changed files with 78 additions and 11 deletions

View file

@ -10,9 +10,11 @@ deploys a pushed container image digest through the Gandalf gateway.
3. It records the deployment and notifies the node the app is placed on.
4. The node applies it: backup gate, pull, restart, health check, and rollback on
failure.
5. This action follows the deployment and prints the node's progress.
5. This action follows the deployment and prints the node's progress. While it waits, it
rides out a forge or gateway that can't be reached or answers 429/5xx (with one warning
per outage), since the deployment goes on without it.
6. It fails the job when the deployment is rolled back, fails, or gets no result within
`timeout`.
`timeout`. A refused request (4xx) fails it at once.
A newer deployment of the same app that supersedes this one ends the job with a notice.