v1.4.3
版本发布时间: 2022-11-22 21:37:14
hashicorp/nomad最新发布版本:v1.9.0-beta.1(2024-09-27 02:58:40)
1.4.3 (November 21, 2022)
IMPROVEMENTS:
- api: Added an API for counting evaluations that match a filter [GH-15147]
- cli: Improved performance of eval delete with large filter sets [GH-15117]
- consul: add trace logging around service registrations [GH-6115]
- deps: Updated github.com/aws/aws-sdk-go from 1.44.84 to 1.44.126 [GH-15081]
- deps: Updated github.com/docker/cli from 20.10.18+incompatible to 20.10.21+incompatible [GH-15078]
- exec: Allow running commands from mounted host volumes [GH-14851]
- scheduler: when multiple evaluations are pending for the same job, evaluate the latest and cancel the intermediaries on success [GH-14621]
- server: Add a git
revision
tag to the serf tags gossiped between servers. [GH-9159] - template: Expose per-template configuration for
error_on_missing_key
. This allows jobspec authors to specify that a template should fail if it references a struct or map key that does not exist. The default value is false and should be fully backward compatible. [GH-14002] - ui: Adds a "Pack" tag and logo on the jobs list index when appropriate [GH-14833]
- ui: add consul connect service upstream and on-update info to the service sidebar [GH-15324]
- ui: allow users to upload files by click or drag in the web ui [GH-14747]
BUG FIXES:
- api: Ensure all request body decode errors return a 400 status code [GH-15252]
- autopilot: Fixed a bug where autopilot would try to fetch raft stats from other regions [GH-15290]
- cleanup: fixed missing timer.Reset for plan queue stat emitter [GH-15134]
- client: Fixed a bug where tasks would restart without waiting for interval [GH-15215]
- client: fixed a bug where non-
docker
tasks with network isolation would leak network namespaces and iptables rules if the client was restarted while they were running [GH-15214] - client: prevent allocations from failing on client reconnect by retrying RPC requests when no servers are available yet [GH-15140]
- csi: Fixed race condition that can cause a panic when volume is garbage collected [GH-15101]
- device: Fixed a bug where device plugins would not fingerprint on startup [GH-15125]
- drivers: Fixed a bug where one goroutine was leaked per task [GH-15180]
- drivers: pass missing
propagation_mode
configuration for volume mounts to external plugins [GH-15096] - event_stream: fixed a bug where dynamic port values would fail to serialize in the event stream [GH-12916]
- fingerprint: Ensure Nomad can correctly fingerprint Consul gRPC where the Consul agent is running v1.14.0 or greater [GH-15309]
- keyring: Fixed a bug where a missing key would prevent any further replication. [GH-15092]
- keyring: Fixed a bug where replication would stop after snapshot restores [GH-15227]
- keyring: Re-enabled keyring garbage collection after fixing a bug where keys would be garbage collected even if they were used to sign a live allocation's workload identity. [GH-15092]
- scheduler: Fixed a bug that prevented disconnected allocations to be updated after they reconnect. [GH-15068]
- scheduler: Prevent unnecessary placements when disconnected allocations reconnect. [GH-15068]
- template: Fixed a bug where template could cause agent panic on startup [GH-15192]
- ui: Fixed a bug where the task log sidebar would close and re-open if the parent job state changed [GH-15146]
- variables: Fixed a bug where a long-running rekey could hit the nack timeout [GH-15102]
- wi: Fixed a bug where clients running pre-1.4.0 allocations would erase the token used to query service registrations after upgrade [GH-15121]