Denied by policy
A denied connection is usually correct — default deny is the product working. The question is which rule is missing.
1 2 | |
no_matching_rule#
Nothing allows this pair. The common case on a first attempt. Check what the peer's identity actually is before writing the rule — the mistake is usually writing a rule about the identity you assumed:
1 | |
port_not_allowed#
A rule matches the identities but not the port. There is no "all ports" value, by design. Add the port to the existing rule rather than creating a second overlapping rule.
service_unknown#
The name does not resolve to a published service. Either a typo, or the service was deleted and the rule is now dangling. Dangling rules are kept visible rather than silently removed:
1 2 | |
identity_expired#
The peer's certificate aged out and renewal has been failing — probably for hours. Fix renewal; the expiry is the symptom:
1 2 3 | |
Alert on renewal failure, not on expiry. By the time you see identity_expired
you have been broken for a while.
identity_revoked#
Someone revoked this identity. Check who and why in the audit history before re-enrolling — if it was revoked in response to an incident, re-enrolling the same host is not the fix.
connector_offline#
The destination connector is not attached to any edge, so the service has no endpoint. See Handshake failed.
A deny that should be an allow, and is not#
If the rule exists and the identities look right, check the realm. A rule in
staging does not apply in prod, and wildcards do not cross realms:
1 2 | |