Skip to content

fix(server): skip endpoint probing when unreachable - #5816

Open
ulischulte wants to merge 2 commits into
masterfrom
feat/stop-fallback-on-transport-errors-5778
Open

ulischulte wants to merge 2 commits into
masterfrom
feat/stop-fallback-on-transport-errors-5778

Conversation

@ulischulte

Copy link
Copy Markdown
Contributor

A transport error on the actuator index query (connection refused,
reset, timeout) made ChainingStrategy fall back to probing, firing one
OPTIONS request per configured endpoint that could never succeed.
The empty result then overwrote previously detected endpoints.

Transport errors now propagate instead of yielding empty, and
EndpointDetector keeps existing endpoints when detection fails.

Closes #5778

A transport error on the actuator index query (connection refused,
reset, timeout) made ChainingStrategy fall back to probing, firing one
OPTIONS request per configured endpoint that could never succeed.
The empty result then overwrote previously detected endpoints.

Transport errors now propagate instead of yielding empty, and
EndpointDetector keeps existing endpoints when detection fails.

Closes #5778
Catching every error in EndpointDetector hid real bugs behind a WARN
with no stack trace. The trigger already logs errors with the full
stack, and nothing is saved on error, so existing endpoints are kept
either way.

Treat any IOException in the cause chain as a transport error, so a
reset during the body read no longer falls back to probing. Codec
errors still fall back because the instance is reachable.
@ulischulte
ulischulte requested a review from a team as a code owner October 23, 2026 09:34

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Feature: Endpoint detection strategy should be retried

1 participant