Nomad - v1.5.1

1.5.1 (March 10, 2023)

BREAKING CHANGES:

• api: job register and register requests from API clients older than version 0.12.1 will not longer emit an evaluation [GH-16305]

SECURITY:

• variables: Fixed a bug where a workload identity without any workload-associated policies was treated as a management token CVE-2023-1299 [GH-16419]
• variables: Fixed a bug where a workload-associated policy with a deny capability was ignored for the workload's own variables CVE-2023-1296 [GH-16349]

IMPROVEMENTS:

• cli: Add job prefix match to the nomad job dispatch, nomad job eval, nomad job scale, and nomad job scaling-events commands [GH-16306]
• cli: Add support for the wildcard namespace * to the nomad job dispatch, nomad job eval, nomad job scale, and nomad job scaling-events commands [GH-16306]
• cli: Added -json and -t flag to alloc checks command [GH-16405]
• env/ec2: update cpu metadata [GH-16417]

DEPRECATIONS:

• api: The Restart(), Stop(), and Signal() methods in the Allocations struct will have their signatures modified in Nomad 1.6.0 [GH-16319]
• api: The RestartAllTasks() method in the Allocations struct will be removed in Nomad 1.6.0 [GH-16319]

BUG FIXES:

• api: Fix Allocations().Stop() method to properly set the request LastIndex and RequestTime in the response [GH-16319]
• cli: Fixed a bug where the -json and -t flags were not respected on the acl binding-rule info command [GH-16357]
• client: Don't emit shutdown delay task event when the shutdown operation is configured to skip the delay [GH-16281]
• client: Fixed a bug that prevented allocations with interpolated values in Consul services from being marked as healthy [GH-16402]
• client: Fixed a bug where clients used the serf advertise address to connect to servers when using Consul auto-discovery [GH-16217]
• docker: Fixed a bug where pause containers would be erroneously removed [GH-16352]
• scheduler: Fixed a bug where allocs of system jobs with wildcard datacenters would be destructively updated [GH-16362]
• scheduler: Fixed a bug where collisions in dynamic port offerings would result in spurious plan-for-node-rejected errors [GH-16401]
• server: Fixed a bug where deregistering a job that was already garbage collected would create a new evaluation [GH-16287]
• server: Fixed a bug where node updates that produced errors from service discovery or CSI plugin updates were not logged [GH-16287]
• server: Fixed a bug where the system reconcile summaries command and API would not return any scheduler-related errors [GH-16287]
• service: Fixed a bug where attaching a policy to a job would prevent workload identities for the job from reading the service registration API [GH-16316]
• ui: fixed an issue where system/sysbatch jobs with wildcard datacenters (like ["dc*"]) were not showing client status charts [GH-16274]
• ui: fixed outbound link to outage recovery on error page [GH-16365]