Nomad plan output does not show as expected

[caichao1103]

Nomad version

Nomad v1.7.7
BuildDate 2024-04-16T19:26:43Z
Revision 0f34c85

Issue

when run nomad job plan, the output is not as expected.

Reproduction steps

We have a nomad job, the main job hcl is as below.

group "sentry-beeper" {
        
    scaling {
      min = 2
      max = 3
    }

    network {
      port "http" {}
    }

    task "sentry-beeper" {

      env {
        GIN_MODE          = var.gin_mode
        OTEL_SERVICE_NAME = "${NOMAD_JOB_NAME}"
      }

      driver = "docker"

      kill_timeout = "30s"

      template {
        data        = var.config_yaml
        destination = "local/config.yaml"
      }

      config {
        image        = var.image
        network_mode = "host"
        args         = ["--config=/local/config.yaml"]
      }

      resources {
        cpu    = var.cpu
        memory = var.memory
      }

step1: did a nomad run with docker image = sentry-beeper:main-b142-20240509, successful. And got two alloc running
step2:
change docker image = sentry-beeper:main-b141-20240509 and then did a docker job plan

Expected Result

The plan output should display 2 create/destroy update, as my understand. However, it actually displayed 1 create/destroy update, 1 ignore

Actual Result

+/- Job: "sentry-beeper"
+/- Task Group: "sentry-beeper" (1 create/destroy update, 1 ignore)
  +/- Scaling {

      }
  +/- Task: "sentry-beeper" (forces create/destroy update)
    +/- Config {
          args[0]:      "--config=/local/config.yaml"
      +/- image:        "216927468640.dkr.ecr.cn-northwest-1.amazonaws.com.cn/infra/sentry-beeper:main-b142-20240509" => "216927468640.dkr.ecr.cn-northwest-1.amazonaws.com.cn/infra/sentry-beeper:main-b141-20240509"
          network_mode: "host"
        }

Scheduler dry-run:
- All tasks successfully allocated.

Job Modify Index: 2135838

Job file (if appropriate)

Nomad Server logs (if appropriate)

Nomad Client logs (if appropriate)

[jrasell]

Hi @caichao1103 and thanks for raising this issue. I have been able to reproduce this locally with a minimal jobspec as detailed below. It does look like when the run is triggered, both allocations go through the create/destroy cycle.

Job Spec:

variable "image" { default = "redis:7" }

job "example" {
  group "cache" {
    count = 2
    task "redis" {
      driver = "docker"
      config {
        image = var.image
      }
    }
  }
}

Plan diff when running nomad plan -var='image=redis:3.2' example.nomad.hcl

+/- Job: "example"
+/- Task Group: "cache" (1 create/destroy update, 1 ignore)
  +/- Task: "redis" (forces create/destroy update)
    +/- Config {
      +/- image: "redis:7" => "redis.3.2"
        }

Scheduler dry-run:
- All tasks successfully allocated.

Job Modify Index: 13
To submit the job with version verification run:

nomad job run -check-index 13 -var="image=redis:3.2" example.nomad.hcl

When running the job with the check-index flag, the job will only be run if the
job modify index given matches the server-side version. If the index has
changed, another user has modified the job and the plan's results are
potentially invalid.