Even if the nomad provider health check is tcp, it wants a GET method.

[Great-Stone]

Nomad version

Nomad v1.4.1 (2aa7e66)

Operating system and Environment details

Linux

Issue

Even if the nomad provider health check is tcp, it wants a GET method.

Reproduction steps

service {
tags = ["api"]
port = "api"
provider = "nomad"
meta {
meta = "Api"
}
check {
type = "tcp"
port = "api"
interval = "10s"
timeout = "2s"
}
}

Expected Result

Success Health check via TCP

Actual Result

Oct 10 03:34:58 raspberrypi nomad[11410]: 2022-10-10T03:34:58.109+0100 [ERROR] http: request failed: method=GET path=/v1/client/allocation/0fbc59e0-8690-9373-1d90-4ec40517dc39/checks error="rpc: can't find method Allocations.Checks" code=500

Job file (if appropriate)

Nomad Server logs (if appropriate)

Nomad Client logs (if appropriate)

Oct 10 03:34:58 raspberrypi nomad[11410]: 2022-10-10T03:34:58.109+0100 [ERROR] http: request failed: method=GET path=/v1/client/allocation/0fbc59e0-8690-9373-1d90-4ec40517dc39/checks error="rpc: can't find method Allocations.Checks" code=500

[jrasell]

Hi @Great-Stone; the client log line you have included is coming from the Nomad HTTP API and not the client internal health checking process. In particular, it is stating that something wishes to view the health checks for allocation 0fbc59e0-8690-9373-1d90-4ec40517dc39 but the client is unable to handle the request. This is potentially because this client is not running Nomad v.1.4.0 or greater, which is the version where native health checking was added. Are you able to confirm all Nomad servers and client are running v.1.4.0 or greater?

I tested locally running Nomad Nomad v1.4.1 (2aa7e66bdb526e25f59883952d74dad7ea9a014e) using nomad agent -dev and registering the following job spec:

job "example" {
  datacenters = ["dc1"]

  group "cache" {
    network {
      port "db" {
        to = 6379
      }
    }

    service {
      port = "db"
      provider = "nomad"
      check {
        type = "tcp"
        port = "db"
        interval = "10s"
        timeout = "2s"
      }
    }

    task "redis" {
      driver = "docker"

      config {
        image          = "redis:7"
        ports          = ["db"]
        auth_soft_fail = true
      }

      resources {
        cpu    = 500
        memory = 256
      }
    }
  }
}

The nomad alloc checks <alloc_id> command returns the expected check results:

Status of 1 Nomad Service Checks

ID         =  d35fb8338075037133b21f123c234d80
Name       =  service: "example-cache" check
Group      =  example.cache[0]
Task       =  (group)
Service    =  example-cache
Status     =  success
Mode       =  healthiness
Timestamp  =  2022-10-10T08:36:22+01:00
Output     =  nomad: tcp ok

[Great-Stone]

Hi @jrasell, It was a mistake in my configuration. As you mentioned the version of Nomad Client was 1.3.5 as opposed to version 1.4.1 of Nomad Server. I checked this log on Nomad Server, so I thought it was server dependent. After upgrading Windows Nomad Client, it works fine.
One suggestion, if users occasionally mix versions of Nomad, it would be nice to have some logging to make sure this is because there is no new API.

[shoenig]

Actually this was a bug, @Great-Stone - one that i just fixed in #14868. Nomad should have avoided placing the job on a known-to-be-outdated node (and now will in a future release). So, thanks for reporting!

[github-actions]

I'm going to lock this issue because it has been closed for 120 days ⏳. This helps our maintainers find and focus on the active issues.
If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.