Replicator project: https://github.com/matthenry87/additional-liveness-path-bug
❯ curl localhost:8080/actuator/health/liveness | jq
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 461 0 461 0 0 317 0 --:--:-- 0:00:01 --:--:-- 317
❯ curl localhost:8080/livez | jq
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 461 0 461 0 0 91 0 --:--:-- 0:00:05 --:--:-- 121
Both responses:
{
"status": "DOWN",
"components": {
"diskSpace": {
"status": "UP",
"details": {
"total": 994662584320,
"free": 809366671360,
"threshold": 10485760,
"path": "/redacted/.",
"exists": true
}
},
"livenessState": {
"status": "UP"
},
"mail": {
"status": "DOWN",
"details": {
"location": "localhost:25",
"error": "org.eclipse.angus.mail.util.MailConnectException: Couldn't connect to host, port: localhost:25, 25; timeout -1"
}
},
"ping": {
"status": "UP"
},
"readinessState": {
"status": "UP"
}
}
}
We discovered this when our mail server went down, and all of our pods/apps that use the mail starter started getting bounced over and over.
Our work-around is to pass this via environment to all apps:
MANAGEMENT_ENDPOINT_HEALTH_GROUP_LIVENESS_EXCLUDE db,ping,diskSpace,readinessState,mail,redis
MANAGEMENT_ENDPOINT_HEALTH_VALIDATE-GROUP-MEMBERSHIP false
Comment From: wilkinsona
Duplicates #40268.