What causes "Error querying job evaluations: %s" in hashicorp/nomad?

Check the wrapped cause: if it's permission denied, add evaluation read rights to the ACL policy for the namespace.

Error querying job evaluations: %s

Error message

Error querying job evaluations: %s

What it means

Raised by outputJobInfo when Jobs().Evaluations(jobID, q) fails while collecting evaluation history for the status output. The raw API error is wrapped into this message. Allocations were already fetched successfully at this point, so the failure is specific to the evaluations read (usually permissions or a transient server error).

Source

Thrown at command/job_status.go:441

// outputJobInfo prints information about the passed non-periodic job. If a
// request fails, an error is returned.
func (c *JobStatusCommand) outputJobInfo(client *api.Client, job *api.Job) error {
	var q *api.QueryOptions
	if job.Namespace != nil {
		q = &api.QueryOptions{Namespace: *job.Namespace}
	}

	// Query the allocations
	jobAllocs, _, err := client.Jobs().Allocations(*job.ID, c.allAllocs, q)
	if err != nil {
		return fmt.Errorf("Error querying job allocations: %s", err)
	}

	// Query the evaluations
	jobEvals, _, err := client.Jobs().Evaluations(*job.ID, q)
	if err != nil {
		return fmt.Errorf("Error querying job evaluations: %s", err)
	}

	latestDeployment, _, err := client.Jobs().LatestDeployment(*job.ID, q)
	if err != nil {
		return fmt.Errorf("Error querying latest job deployment: %s", err)
	}

	jobActions := make([]map[string]string, 0)
	for _, tg := range job.TaskGroups {
		for _, task := range tg.Tasks {
			for _, action := range task.Actions {
				jobActions = append(jobActions, map[string]string{
					"group":  *tg.Name,
					"task":   task.Name,
					"action": action.Name,
				})
			}
		}

View on GitHub (pinned to 482b49bf1a)

Solutions

Check the wrapped cause: if it's permission denied, add evaluation read rights to the ACL policy for the namespace.
Confirm namespace/region flags match the job's scope.
Retry the command if the server just failed over (no cluster leader / 500).
Verify agent connectivity and NOMAD_ADDR if the cause is a connection error.

Example fix

// before (HCL policy missing eval read)
namespace "default" {
  job = "read"
}
// after
namespace "default" {
  job       = "read"
  eval      = "read"
  allocation = "read"
}

Defensive patterns

Strategy: retry

Validate before calling

curl -sf -H "X-Nomad-Token: $NOMAD_TOKEN" \
  "$NOMAD_ADDR/v1/job/$JOB_ID/evaluations?namespace=$NOMAD_NAMESPACE" > /dev/null

Try / catch

jobEvals, _, err := client.Jobs().Evaluations(id, q)
if err != nil {
    if strings.Contains(err.Error(), "No path to a job") || isTransient(err) {
        // retry or adjust namespace
    }
    return err
}

Prevention

Include eval = read in least-privilege ACL policies used for status tooling.
Use consistent -namespace/-region across all queries in one command invocation.
Retry reads after server failover events.
Version-pin CLI against the cluster to avoid endpoint drift.

When it happens

Trigger: client.Jobs().Evaluations(*job.ID, q) returns an error: agent unreachable mid-command, ACL token cannot read evaluations in the namespace, namespace/region mismatch in q, or server-side error during the read.

Common situations: Fine-grained ACL policy allowing job/alloc reads but not evaluation reads; leadership change on the server between the alloc and eval queries; temporary network blip during a long status command.

Related errors

AI-assisted analysis of hashicorp/nomad@482b49bf1a (2026-09-04). Data as JSON: /api/errors/3864e9dc92a4ab7f. Report an issue: GitHub.