- The number of images pulled in parallel is now determined by the number of CPUs. This will help remove contention when pulling several images on underpowered servers.
- Fixed an issue that could cause Replicated to start containers out of order when they are specified in the same component and at least one is optional.
- When using the Replicated scheduler, all Docker images will be pulled in parallel to speed up installations when running Docker version 1.10.0 or later.
- Replicated now fully supports Docker 17.06. New installations will start with 17.06 unless otherwise specified in preflight checks.
- The Admin Console will now use a single connection to Graphite when querying metric data.
- When automating an installation for testing, boolean values no longer require quotes.
- The Replicated data directory can be overridden using the
DefaultDataDir environment variable or a
- Added a top level YAML key named
support where all custom support files and commands can be declared.
- Replicated console settings changes are now included in the audit log.
- Fixed an issue that would cause the Replicated easy install script to exit with no output when updating Replicated on Debian based distributions.
- Password fields on the Admin Console settings page can now be erased.
- Replicated will now respect the removal of the http proxy in runtime configuration
- Disconnecting a node will no longer prevent app from starting when preflight checks fail.
- The hosted
docker-compose.yml for installing Replicated on Docker Swarm was updated to work correctly with Docker 17.03 and later.
- LDAP anonymous binds will no longer be supported and will now require a password.
- Fixes an issue where users were able to log into the admin console without a password when LDAP authentication was enabled using AD or when anonymous binds were enabled on the LDAP server.
- Fixes an issue where resource contention causes operators to disconnect.
- Fixed an issue where
ConfigOption templates will fail to evaluate when the target is a sub-item of an item with type
- Fixes an issue introduced in 2.9.0 where the
ConfigOption template was failing to get resolved in the case where the target item’s when clause evaluates to false.
- Fixes an issue introduced in 2.9.0 that causes the connection between the operator and the master to drop when in contention for CPU resources for an extended period of time.
- Fixes an issue on the preflight checks page where the continue button is not visible when an operator is disconnected.
- Fixes an issue in some cases where the application YAML sent from Replicated Studio will cause Replicated to crash.
- Removes a request from the UI prior to app installation causing an error log in replicated related to a missing app ID query param.
- Customer licenses can optionally be assigned to multiple channels, giving customers the ability to choose one of the selected release channels for each installation.
- A new Replicated CLI is installed that exposes the ability to manually perform most actions which would previously require the Admin Console. The original CLI is still available with no changes. To manually enable
replicatedctl on servers upgraded from previous versions, see the manual instructions
- Previous versions of an application can be installed using a parameter in the easy install script or by creating a Replicated configuration file.
- The Premkit and Statsd external networks are now automatically added as environment variables into application containers on the Swarm scheduler.
- Preflight checks are now enabled on the Kubernetes scheduler with several Kubernetes-specific items added.
- Added all detected parameters from a running Replicated server into the support bundle, regardless of where they were set.
- Admin Console Boolean configuration options are now allowable as values to templatable when properties in the application YAML. This eliminates the need to build off multiple hidden config options.
- The maximum clock skew between nodes in a Replicated cluster is configurable on the Console Settings page of the Admin Console.
- The Admin Console enforces CSRF tokens on browsers with low security settings.
- Automated installs will now work with airgap licenses.
- Kubernetes deployments with replica values that evaluate to 0 will no longer prevent the application from starting.
- Increased the reliability of application start/stop state in the Admin Console.
- Fixed an issue where updating a configuration option would not be automatically applied until the next application restart.
- Fixed an issue that would cause invalid YAML to prevent an update from being applied or skipped.
- Fixed a bug that would cause an application update to fail to apply after Replicated was automatically updated.
- Fixed a bug that would cause the proxy server settings to be written incorrectly on systemd systems.
- Added additional Replicated features when deploying on Docker Swarm:
- Added the ability to utilize Docker secrets in an application.
- Airgapped installations are now supported.
- Admin commands are now supported.
- Better error messages are displayed in the admin console and logs when a Docker Swarm stack fails to deploy or start.
- Application containers will now be automatically added to the premkit and statsd networks. This makes it easier to use the Replicated Integration API in your application.
- The application specific installation scripts will now apply custom branding to all screens, including the screens before the license is uploaded.
- Component names are now a required field. They were strongly recommended before, but it’s not possible to save YAML without component names now.
- When installing using the easy install script, supported operating systems will receive Docker version 17.03.1-ce. Older operating systems will continue to receive older Docker versions as needed.
- Unsupported console settings are hidden from the console settings page in the admin console.
- The public port property in container configurations are now optional. When not specified, Replicated will let Docker choose the best port.
- The initial cluster instance count property can now use Replicated template functions.
- The support bundle will now report if the installation is airgapped.
- Added ability to start snapshots via Replicated CLI.
- Added an admin console dashboard warning if Replicated detects a clock skew higher than 5 minutes on any operator.
- Easy installation script now supports Amazon Linux 2017.03.
- Fixed issue where running admin commands with a
--help flag would not correctly render line breaks.
- Global services in Swarm will no longer prevent Replicated applications from stopping completely.
- Fixed the instructions on the Cluster page of the admin console when running in Airgap mode to show proper instructions.
- The Docker version in the host requirements is now treated as a minimum.
- Added support for SUSE Enterprise Linux 12 in the install scripts.
- Fixed an issue that was preventing preflight checks from being able to determine the number of CPU cores when executed on single core Google Cloud Platform and Amazon Web Services EC2 instances.
- Easy installation script will now correctly tag Replicated component images as current on older versions of Docker.
- Fixed an issue when no files were selected for a multiple file input, an empty directory was created on the server.
- Fixed an issue where a failed preflight check on a server running Docker 17.04 could prevent user from continuing with the installation.
- Fixed an issue with file owner from being applied if the permissions were not also set in the YAML.
- Fixed an issue preventing Replicated from properly parsing the Docker Enterprise Edition version strings.
- Fixed an issue preventing Replicated from properly parsing the Oracle Enterprise Linux version strings.
- Replicated now supports deploying a Swarm (docker-compose v3/3.1 YAML). The following features are supported:
- Added a new easy install script to provision a new Swarm cluster while installing Replicated. Documentation is available.
- Added support for Replicated Automatic Updates when running in Swarm.
- Added support for private images pushed to the Replicated registry. Swarm supports both public images and private images in the Replicated registry with no additional configuration required.
- Added statsd/graphite support and custom monitors on the dashboard. The standard CPU and memory graphs are not supported currently.
- The support bundle now contains Swarm specific data when running Replicated in Swarm.
- Added the Replicated Integration API endpoint as a new network, joined to both the Replicated and the application network.
- An new section of our docs was written to help get started with deploying Replicated on Swarm at https://www.replicated.com/docs/packaging-an-application/docker-swarm.
- An example application (Voting app) showing how to deploy a Swarm application on Replicated is at https://www.replicated.com/docs/examples/swarm-votingapp/.
- Added support for pointing the AppURL on the Admin Console to a service endpoint.
- Added several new template functions to support Swarm installations.
- Added SwarmIngressAddress template function to determine the generated ingress address of the service.
- Added SwarmNodeIDs template function to get a list of all Docker node ids in the cluster.
- Added Namespace template function to read the generated namespace of the application in Swarm.
- When running Replicated on Kubernetes, it’s now possible to run Replicated in a non-default namespace.
- The ready state default time has changed from 2 seconds to 5 seconds.
- The ready state timeout is configurable in the application YAML.
- Added an autoscale property to the component level in the Replicated scheduler.
- Added a feature to the license page that allows a license to be locked to a specific version of the application. This will prevent updates past this version and will force new installs to start at this version.
- Custom preflight checks can use config files from the application YAML.
- Fixed a missing error message when multiple Replicated Operator containers with the same ID attempt to join the Replicated service. This is relevant for Replicated scheduler only.
- Fixed the “download” button on the snapshot restore page to properly download the recoverable volume.
- Fixed issue where using a AWS generated DNS name in the SSL Hostname input would generate errors.
- The following features were added to Replicated when running in Kubernetes:
- When an application update and an update to Replicated are both available at the same time, it no longer requires the user to click accept, then go back and accept the other. All pending updates will be applied at the same time.
- The install script will warn when Docker is installed with devicemapper in loopback mode. This is not a recommended configuration from Docker. When this warning is displayed, a link to the Docker recommendations is also provided. We also added an argument to skip this check by passing in
bypass-storagedriver-warnings to the install script.
- Preflight checks support parsing the new Docker versioning semantics (YY.MM instead of semver).
- Install scripts parsing the new Docker versioning semantics.
- Removed the requirement that LDAP users belong to at least one group to use in Replicated. This works with services such as Jumpcloud, using the default configuration now.
- Added a heartbeat on the Replicated and Replicated Operator communication channel. This will help reconnect when the network connection is broken and not detected as broken by either endpoint.
- Updated default preflight checks to allow for Docker up to version 1.13.1.
- Updated the base Replicated Docker images from Alpine 3.3 to Alpine 3.4.
- Changed the port-available custom preflight checks to run sequentially instead of in parallel when there are multiple checks.
- Fixed a bug that was causing Replicated to leak one file handle every time the Replicated UI page was loaded.
- Fixed a bug that was causing the cluster page in the Admin Console to show the wrong IP address after a snapshot restore process completed.