Details

    • Type: Bug
    • Status: Resolved
    • Priority: High
    • Resolution: Done
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: Health & Readiness
    • Labels:

      Description

      *strong text*It appears that when I have multiple health checks, the deployment of a task will complete as soon as a single health check succeeds rather than waiting for all health checks to succeed. That means the old task will be killed before the new task has reached a healthy state.

      Is this expected behavior? I'd rather it wait for all health checks to succeed.

      I've observed this through the UI, but I can't see it in the logs, because I don't see the log statements from the HealthCheckActor.

      Here is the JSON I am posting to marathon:

      {
          "backoffFactor": 2,
          "backoffSeconds": 10,
          "container": {
              "docker": {
                  "image": "mytestcontainer",
                  "network": "BRIDGE",
                  "portMappings": [
                      {
                          "containerPort": 22925,
                          "hostPort": 0,
                          "protocol": "tcp"
                      },
                      {
                          "containerPort": 13001,
                          "hostPort": 0,
                          "protocol": "tcp"
                      }
                  ]
              },
              "type": "DOCKER"
          },
          "cpus": 1,
          "healthChecks": [
              {
                  "gracePeriodSeconds": 90,
                  "intervalSeconds": 20,
                  "maxConsecutiveFailures": 3,
                  "path": "/status",
                  "portIndex": 0,
                  "protocol": "HTTP"
              },
              {
                  "gracePeriodSeconds": 90,
                  "intervalSeconds": 20,
                  "maxConsecutiveFailures": 3,
                  "portIndex": 1,
                  "protocol": "TCP"
              }
          ],
          "id": "testing",
          "instances": 1,
          "mem": 1024.0
      } 
      

        Attachments

          Issue Links

            Activity

              People

              • Assignee:
                kjeschkies Karsten Jeschkies
                Reporter:
                GitHub_dsiegel David Siegel (Inactive)
                Team:
                Orchestration Team
                Watchers:
                Chmielewski, clems4ever, Ken Sipe, Matthias Eichstedt, tgermain
              • Watchers:
                5 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: