You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, the /health endpoint runs on the same port as the webhook.
The webhook doesn't implement any authentication mechanism, which means you must expose the webhook to the rest of the cluster (and possibly outside the cluster, depending on the CNI) just to enable readiness/liveness probes.
It would be ideal to limit it to local connections from external-dns to external-dns-ionos-webhook when running as a side-car (ie: listen on 127.0.0.1). But this currently breaks the readiness/liveness probes.
Proposal:
Run a separate server (in a go routine) specifically for /health which can safely be exposed.
The text was updated successfully, but these errors were encountered:
NickLarsenNZ
changed the title
security: health endpoint should be on a different port than the webhook
SECURITY: health endpoint should be on a different port than the webhook
Jan 21, 2025
NickLarsenNZ
changed the title
SECURITY: health endpoint should be on a different port than the webhook
SECURITY: The /health endpoint should be on a different port than the webhook
Jan 21, 2025
Problem
Currently, the
/health
endpoint runs on the same port as the webhook.The webhook doesn't implement any authentication mechanism, which means you must expose the webhook to the rest of the cluster (and possibly outside the cluster, depending on the CNI) just to enable readiness/liveness probes.
It would be ideal to limit it to local connections from
external-dns
toexternal-dns-ionos-webhook
when running as a side-car (ie: listen on 127.0.0.1). But this currently breaks the readiness/liveness probes.Proposal:
Run a separate server (in a go routine) specifically for
/health
which can safely be exposed.Similar to what is done (opt-in) for metrics: #70
The text was updated successfully, but these errors were encountered: