Why auto_revert doen't work? Nomad v1.0.4

I expected that when my task one from two will failed I got revert. And my failed task will run from previos stable version.
My application use rolling update.

  	# Configure the job to do rolling updates
   update {
    stagger           = "10s"
    max_parallel      = 1
    health_check      = "task_states"
    min_healthy_time  = "40s"
    healthy_deadline  = "15m"
    progress_deadline = "16m"
    auto_revert       = true
  	}

  group "logstash" {
    count = 2

    constraint {
      attribute = "${node.class}"
      value = var.node_
    }

    restart {
        attempts = 2
        interval = "1m"
        delay = "20s"
        mode = "fail"
    }

nomad job inspect elk-logstas

... "Tasks": ...
              ],
                "Update": {
                    "AutoPromote": false,
                    "AutoRevert": true,
                    "Canary": 0,
                    "HealthCheck": "task_states",
                    "HealthyDeadline": 900000000000,
                    "MaxParallel": 1,
                    "MinHealthyTime": 40000000000,
                    "ProgressDeadline": 960000000000,
                    "Stagger": 10000000000
                },
                "Volumes": null
            }
        ],
        "Type": "service",
        "Update": {
            "AutoPromote": false,
            "AutoRevert": false,
            "Canary": 0,
            "HealthCheck": "",
            "HealthyDeadline": 0,
            "MaxParallel": 1,
            "MinHealthyTime": 0,
            "ProgressDeadline": 0,
            "Stagger": 10000000000
        },

But when my first container failed I got only restart this one. More time.
But I can manul do job revert to previos version. After this my failed container become vork fine with stable version.
Where is problem?

This is reason

progress_deadline = "16m"

long time for waiting for revert

2 Likes