fix frontend 500 error when probing service #79
Conversation
| httpHeaders: | ||
| - name: accept-language | ||
| value: en | ||
| initialDelaySeconds: 30 |
There was a problem hiding this comment.
The initialDelaySeconds of 30 seconds is too long for the frontend's readinessProbe. Next.js starts up very quickly, usually completing within a few seconds. With initialDelaySeconds set to 30, it takes at least 30 seconds for the frontend to become available. reduce the initialDelaySeconds to 3 seconds?
There was a problem hiding this comment.
How about making this configurable? I experienced unavailability for at least 20 seconds
There was a problem hiding this comment.
How about making this configurable? I experienced unavailability for at least 20 seconds
In my case, the langgenius/dify-web container took less than one second to reach readiness after starting.
2024-04-24T06:17:49: PM2 log: Launching in no daemon mode
......
06:17:50 0|dify-web | ✓ Ready in 210msIt's important to note that in Kubernetes, the time taken for container scheduling and image pulling is not included in the health check calculations.
dify frontend needs a header
accept-languageto properly respond, bug: langgenius/dify#3724add more initial delay to prevent unnecessary check before frontend it ready