2.44.0

New Features

  • All Kubernetes Component Docker container images installed by the Replicated Embedded Kubernetes installer are now scanned and patched prior to each Replicated release.
  • Files can now be omitted from the Replicated Support Bundle using the files property of the meta.redact spec.

Changes

  • The Docker version on Native and Swarm installs will now default to 19.03.8.
  • Replicated Docker dependencies have been updated to version 19.03.8.
  • The swarm-default-address-pool flag has been added to the Swarm quick install scripts. This flag is used to specify the default subnet pools for global scope networks. The value is passed to the swarm init command via the --default-addr-pool flag.
  • The replication level of the Rook and Ceph storage pools used by Replicated Embedded Kubernetes will be set hourly. Previously the replication level was only set after changes to the cluster size.
  • Secrets will now be encrypted before being stored in the database using AES-GCM encryption with a per-installation 192-bit key generated at installation time.
  • When the Replicated database is corrupted it will automatically attempt to repair itself on next start.
  • The Docker Version Custom Preflight Check no longer requires a maximum version.
  • The LdapCopyAuthFrom template function now accepts the ServerType parameter to return the LDAP server type.
  • Metrics reporting has been disabled for the audit logging service on both Kubernetes and Swarm installations.
  • Task logs are now stored in the Replicated temp directory instead of the database.

Bug Fixes

  • Memory and disk space Custom Preflight Checks now properly differentiate between SI and binary units.
  • Fixed scaling of RGW metadata pools for clustered Kubernetes installations.
  • Fixed an issue that could cause Replicated to prompt the user to dismiss Preflight Checks twice on initial installation.
  • The Replicated container process will now respond to signals from the service manager and attempt to shut down in a clean manner.
  • Certain error messages encountered during app updates will now be shown in modals because they were previously rendered too quickly to be seen.
  • Improvements have been made to properly reflect the Pod deployment status on the Cluster page of the Replicated Embedded Kubernetes Admin Console.
  • Fixed an issue that prevented the end-user from changing the Kubernetes API load balancer address on Replicated Embedded Kubernetes airgapped installations.
  • Fixed an issue that caused Replicated Embedded Kubernetes installations to fail due to failed registry pushes when in airgapped mode and an HTTP proxy is configured.
  • Fixed an issue that caused Replicated Embedded Kubernetes installations to fail due to failed creation of the registry S3 bucket when in airgapped mode and an HTTP proxy is configured.
  • The Replicated statsd-graphite container will now be restarted on Native installations when the application is upgraded allowing for changes in port configuration to take effect.
  • The Replicated statsd-graphite container will now log to stdout and stderr.
  • Fixed an issue that could cause Replicated Embedded Kubernetes to crash when a pod takes too long to start.
  • Silenced noisy logs when querying the /ping endpoint of the Replicated UI service with a console password set.
  • Fixed Docker installation scripts on CentOS 6, RHEL 6 and Fedora 30.
  • Fixed discrepancy in number of audit log events returned by search and CSV export.
  • The shutdown service on Replicated Embedded Kubernetes will ensure the Replicated pod is not bound to the node that is being shutdown.
  • All known CVEs with fixes have been patched in all images distributed by Replicated at the time of the release. For more information see this article.

2.43.1

Notes

  • This release is currently only available for the Swarm and Native schedulers.

Changes

  • The Native scheduler property ports[].public_port_initial has been added to prevent port collisions when exposing a public port when clustering is enabled and cluster_instance_count.initial is greater than one. This property functions the same as the public_port property but will be ignored by versions of Replicated prior to 2.43.1.

Bug Fixes

  • All known CVEs with fixes have been patched in all images distributed by Replicated at the time of the release. For more information see this article.

2.43.0

New Features

  • Airgapped licenses can now be updated by uploading a file on the Console Settings (/console/settings) page of the Replicated Admin Console.
  • The optional flag taint-control-plane has been added when using the Replicated Embedded Kubernetes installation scripts to prevent Pods from running on master nodes alongside the control plane. This will add the node-role.kubernetes.io/master: NoSchedule taint to all master nodes.
  • The Replicated Native Scheduler will now map a range of ports when clustering is enabled and cluster_instance_count.initial is greater than one and a port.public_port is exposed explicitly on the host.
  • The command replicatedctl snapshot rm has been added to allow for the removal of a snapshot from the Replicated CLI.

Changes

Bug Fixes

  • Fixed an issue that caused failures from the kubeadm init command to be ignored by the Replicated Embedded Kubernetes installation script.
  • Increased the minimum Ceph pool size on multi-node Replicated Embedded Kubernetes clusters to 2 to prevent data loss from the loss of a single node.
  • Fixed an issue that caused the Rook Ceph Object Gateway replication factor not to be applied, leading to data loss from the loss of a single node.
  • Fixed an issue that could prevent a Replicated Embedded Kubernetes application from starting when the application defines an APIService.
  • Fixed an issue when using the Native scheduler that could cause the application to restart when Replicated is updated.
  • Failure to meet the minimum replicated_api_version requirement will no longer prevent Replicated from downloading new versions of an application. An application update will still fail if the requirement is not met.
  • When running the command replicatedctl app-config set, Test Procs with the run_on_save property set to true will only run for items that are changing, no longer preventing the command from succeeding.
  • Snapshot restore now supports AWS Instance Profiles.
  • Fixed an issue that could cause the SFTP Snapshots temporary directory not to be properly cleaned up when a snapshot failure occurs.
  • Fixed an issue that could cause a Support Bundle to be empty when a task timeout occurs.
  • Weave container logs will now default to info level rather than debug level.
  • The Replicated Premkit container logs will now default to info level rather than debug level.
  • Silenced noisy TLS handshake errors from the Replicated UI server logs.
  • Silenced long errors from i18n in the Replicated Daemon logs when a template fails to evaluate.
  • All known CVEs with fixes have been patched in all images distributed by Replicated at the time of the release. For more information see this article.

2.42.5

Bug Fixes

  • This release contains bug fixes.

2.42.4

Bug Fixes

  • This release contains bug fixes.

2.42.3

Notes

  • This release is currently only available for the Kubernetes scheduler.

Changes

  • All Pods that use the Ceph Shared Filesystem will have an initContainer injected to verify the mount succeeded before proceeding.

Bug Fixes

  • Fixed an issue when using the Ceph Shared Filesystem that could cause the application to lose data by inadvertently writing to the ephemeral container filesystem instead of the shared filesystem.

2.42.2

New Features

  • Added the route GET /console/v1/auth to the Integration Console API to retreive auth type and login parameters.

Bug Fixes

  • Fixed an issue that prevented successful Replicated Embedded Kubernetes airgapped installations when an HTTP proxy was configured.
  • Fixed an issue that would cause Swarm snapshots to fail with an error pulling the Postgres database export image.
  • All known CVEs with fixes have been patched in all images distributed by Replicated at the time of the release. For more information see this article.

2.42.1

Notes

  • This release is currently only available for the Native and Swarm schedulers.

Bug Fixes

  • Fixes a Native scheduler issue where container shared memory settings were incorrectly set to 10 bytes after a Replicated upgrade.

2.42.0

New Features

Changes

  • The command replicatedctl license-load will no longer wait for the installation to complete without the --attach flag.
  • The timeout enforced by Replicated when restoring node volumes from a snapshot has been increased from 1 hour to 12 hours.

Bug Fixes

  • Fixed an issue that caused airgapped installations to stall due to large airgap bundle sizes, requiring a browser refresh.
  • Fixed an issue that caused Replicated to crash when the audit log database is corrupt.
  • Fixed an issue that could cause Replicated to pull images from the Replicated Registry with the wrong credentials after an auto upgrade of Replicated, resulting in image pull failures.
  • All known CVEs with fixes have been patched in all images distributed by Replicated at the time of the release. For more information see this article.

2.41.1

Bug Fixes

  • This release contains bug fixes.