-
Notifications
You must be signed in to change notification settings - Fork 62
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Should health checks expose /app/health ? #45
Comments
I would rather go the other way around and have /health and /health/* This way less "top level" urls are used. Also /health is sort of a standard for health check purposes. Not offering this, leads to more work for the admin running the MP server. |
FYI, this issue is semi-related to Issue #35 as well. I agree that we need to provide the single /health endpoint initially and then grow it from there (/health/*). |
@pilhuhn I think /app/health is better as this endpoint will work in any environment. It works even in the environment with multiple app installed. If we just expose one root /health, how do we gather the health status for multiple app environment. |
There is a conceptual mismatch between So far we've been driving this from the assumption that a single failed health check leads to the termination of the computing node. This would tear down all applications on that runtime on that computing node. From this perspective a single If you want to support runtimes with multiple applications installed this would challenge the baseline assumption behind this spec. I am not saying we cannot do this, but I want that we are all on the same page with respect to assumption and constraints of the current design. |
From @jmesnil (carried over from #29)
The spec should allow to provide further inspection through the health endpoint to check healthiness of a given component.
This has the advantage of keeping the deployment endpoint space clean. What do you think? |
Understood, Heiko. I think @jmesnil is close to what I am thinking. |
@Emily-Jiang I am fine with supporting
When this is guaranteed we can treat multiple |
@heiko-braun agreed. |
Thinking further about response composition and URL namespaces, maybe we should guarantee that the composition level is reflected in the URL namespace, i.e:
This way we could provide some leeway for implementors to support composition levels best suited for their runtime. I.e @jmesnil could still do |
First I think we are talking Microprofile here with an emphasis on Microservices, so I wonder if the case of one app-server hosting many apps AND running Microprofile is really common. This is why I am advocating to only have one top level |
@pilhuhn I do agree on less top level namespaces. I think my latest proposal here reflects that:
|
@heiko-braun I guessed it from the subsystem example, but wanted to make my point explicit :) |
On the last call we've decided to require a single |
Should the context be configurable? What do you think? |
@Emily-Jiang after more than 1 year what is your opinion on this ticket ? |
In the design, it exposes /app/health and /health. Can we just expose /app/health? In this way, the health check endpoint can be accessible without any confusion. It even work in multiple app environment. This will simplify the spec. @jmesnil what do you think?
The text was updated successfully, but these errors were encountered: