Anypoint Runtime Fabric 1.x.x Release Notes Archive

Added: Support for Kubernetes version 1.19 for Self-Managed Kubernetes clusters.

Added: External log forwarding support for Datadog logging service.

Added: Patches for CVE-2020-15999, CVE-2020-1971, and CVE-2021-3449.

Updated: External log forwarding improvements for performance under load.

Updated: Logging improvements for Runtime Fabric agent logs.

Updated: Deprecated support for Kubernetes version 1.16 for Self-Managed Runtime Fabric clusters.

Fixed: Toggling the TLS option did not clean up the CA cert configuration for external log forwarding.

Fixed: External log forwarding issue with character escaping in JSON-formatted log messages.

Fixed: Intermittent SSL connection errors that occurred under stress load when using external log forwarding.

Fixed: Intermittent Resource not found error occurred when toggling log forwarding.

Fixed: Persistence object store returned intermittent authorization errors.

Added the Persistence Gateway that enables Mule applications deployed to a Mule runtime engine (Mule) to store and share data across Mule application replicas and application restarts. This feature is supported on both Runtime Fabric on Self-Managed Kubernetes and Runtime Fabric on VMs / Bare Metal.

Added: Support for Runtime Fabric on Self-Managed Kubernetes with Google Kubernetes Engine (GKE).

Added: Security improvements for the resource-fetcher service.

Added: Support for Enhanced API Metrics with Anypoint Monitoring.

Added: Support for new Anypoint Monitoring endpoint (dias-ingestor) with port 443; the existing endpoint will be deprecated.

Fixed: Cluster health reporting may encounter issue when Anypoint Monitoring Log Forwarding is disabled.

Added: External log forwarding support for Syslog logging service.

Added: Runtime Fabric and Anypoint Security Edge support for WebSockets.

Fixed: Runtime Manager reporting incorrect state in some cases, when multiple applications were deployed concurrently.

Fixed: OpsCenter CPU usage dashboard displaying only last 24 hours of CPU data.

Fixed: Application logs missing in Anypoint Monitoring in some occasions.

Added: Support for Runtime Fabric on Self-Managed Kubernetes with Amazon Elastic Kubernetes Service (EKS) and Azure Kubernetes Service (AKS).

Added: Updates that improve the security of components running on Java.

Fixed: The internal load balancer would remove the 2nd occurrence of the "=" character in application URLs containing a query string.

Fixed: The internal load balancer would incorrectly add multiple "Via" HTTP headers to requests sent to deployments.

Added: Further improvements to cluster status reporting.

Fixed: Increased Anypoint Monitoring sidecar memory limit to better handle data bursts.

Fixed: Increased Runtime Fabric Component upgrade timeout to accommodate upgrades on slower networks.

Fixed: Issue with missing priority classes on restoring from the backup.

Added: External log forwarding support for AWS CloudWatch logging service.

Added: Ability to report node healthiness and Runtime Fabric component version inconsistency as part of cluster health monitoring.

Added: Support for multiple certificates and domains with the internal load balancer.

Added: Support for routing non-secure, HTTP requests with the internal load balancer.

Added: Support for GET request with body at the internal load balancer.

Fixed: Issues affecting the reporting of deployment statuses in some cases.

Fixed: Error with chunked encoding at the internal load balancer.

Note: JVM heap metric is now only available under Anypoint Monitoring.

Added: Ability to exclude logs from Runtime Fabric components and the appliance when forwarding logs to an external provider.

Added: Improvements in reporting cluster health and updating cached URLs when enabling mutual authentication for connectivity to Anypoint Platform.

Added: Show download progress while fetching application assets.

Fixed: Issues affecting the stability of cluster health reporting.

Fixed: Failure to deploy applications with names starting with "http".

Fixed: Ingress not returning the appropriate response codes when http protocol limits were violated.

Fixed: Security improvements and docker image hardening.

Added: Ability to send a test message to validate external log forwarding configuration.

Added: Ability to include 'CA Path Certificate File' when configuring external log forwarding.

Added: Ability to disable external log forwarding at application level.

Fixed: Security improvements which reduce scope of access.

Added: Ability to configure external log forwarding from Runtime Manager.

Added: Memory optimization improvements with the internal load balancer.

Fixed: Ingress dashboard in Ops Center not displaying correctly.

Fixed: Issue with reporting cluster health status when pods are in a pending state.

Fixed: Security fixes.

Added: Ability to enforce application replicas are deployed on different worker nodes.

Added: Ability to disable log forwarding to Anypoint Monitoring at application level.

Fixed: Replicas with Mule clustering enabled would sometimes fail to form a cluster.

Fixed: Updates to the appliance caused credentials for the internal metrics database to be out of sync.

Fixed: Cluster health reporting in Runtime Manager would not update in some cases.

Fixed: Improvements to increase stability in the reporting of Runtime Fabric health.

Fixed: Runtime Fabrics managed by the EU control plane were running validations against US control plane URLs.

Fixed: Internal load balancer upgrade failed in some cases.

Added: Detailed health visibility for Runtime Fabrics shown in Runtime Manager.

Added: Ability to enable and disable forwarding logs to Anypoint Monitoring (Requires Titanium).

Added: Appliance version is displayed on Runtime Fabric settings page with upgrade instructions.

Added: Upgrades to internal load balancer are treated the same as other Runtime Fabric components.

Fixed: Optimized Runtime Fabric agent for better resource utilization.

Fixed: Optimized Runtime Fabric agent limiter behavior when multiple deployments are in Pending state.

Fixed: Stability improvements and critical bug fixes.

Fixed: Upgrades may not complete when performed on Installer package version 1.0.7 or below.

Fixed: Installations performed using an outdated version of the install scripts failed without a detailed error message.

Fixed: Installations performed with a configuration consisting of zero available worker nodes would fail by time out.

Fixed: "%Inf" appeared incorrectly in some alert emails.

Added: CPU Core Usage dashboard to view Mule core usage across Runtime Fabric.

Fixed: Applications which failed due to lack of memory were reported with a generic "JVM exited unexpectedly" message.

Fixed: Heap memory metrics URL was incorrect after upgrading from installer package version 1.0.6 or earlier.

Added: Support for enabling CPU bursting for Mule application and API gateway deployments.

Fixed: Failed to reconnect to control plane after rebooting a controller node due to a race condition between pod starting and Kubernetes API being available.

Fixed: Failed to publish events if Kubernetes resource version became out of date.

Updated: Rolling out new replicas happens concurrently instead of one by one.

Updated: Reduced DNS query timeout for internal load balancer to reduce recovery time in the case of DNS pod restart.

Fixed: Proxy authentication header was not being sent in some cases.

Updated: Timeout limit increased while downloading resources from control plane to 2 minutes

Added: Deployments will better follow an eventually consistent model by continuously retrying. Read more about eventual consistency.

Added: Runtime Manager will report "Pending" status for replicas which are being scheduled on Runtime Fabric.

Added: Enhanced error reporting during deployment initialization.

Added: Enhanced status reporting while deployments are pulling images.

Fixed: TLS certificate chain wasn’t included during certificate renewal in some cases.

Fixed: Incorrect Runtime Fabric health was reported after installation in some cases.

Fixed: Internal load balancer metrics were not available in Ops Center Ingress dashboard.

Fixed: Incorrect available capacity reported for application deployments in some cases.

Added: Runtime Fabric agent components will re-initialize if disconnected from control plane for more than six hours.

Fixed: Applications failed to deploy when using a certificate with an uppercase common name.

Updated: Deployments are now labeled with rtf.mulesoft.com/id.

Fixed: DNS queries to Anypoint control plane were performed even when running through an HTTP proxy

Added: Java version upgraded to 8u212.

Fixed: HTTPS_PROXY environment variable was not used when refreshing credentials in some cases.

Added: Support for HTTP proxies with authentication enabled.

Fixed: Errors during application deployment did not show the correct container name in some cases.

Added: Support for viewing application metrics and logs with Anypoint Monitoring.

Added: Support for consuming secure properties in applications.

Added: Cluster Service upgraded to Java 8u191.

Added: Automatic HTTP proxy configuration for API Gateways.

Fixed: Agent upgrade did not successfully complete in some cases.

Fixed: Improved error message when application deployments failed due to a missing Mule license.

Fixed: JVM heap memory dashboard was not visible in Ops Center UI in some cases.

Fixed: Agent readiness probe did not report error when control plane connectivity was lost in some cases.

Added: Support for RHEL/CentOS v7.6.

Added: HTTPS-based connectivity to Anypoint control plane. (Replaces TCP-based connectivity).

Added: JVM memory heap monitoring in Ops Center 'Applications' dashboard.

Added: Java version upgraded to 8u191.

Fixed: User-specified HTTP proxy configuration was lost across updates.

Fixed: Application shows as Recovering instead of Updating after unexpected restart in Application Manager.

Fixed: Changing settings for Inbound Traffic failed in some cases.

Added: Proactive renewal of certificate used to connect to control plane.

Added: Changes to share Runtime Fabric with business groups and specify resource limits with environments.

Fixed: Removed a hardcoded dependency with a component used to collect metrics for the Ops Center. Note: newer builds of the Runtime Fabric installation package will require this update.

Fixed: Runtime Fabric unable to register to control plane if HTTP proxy is configured.

Fixed: Runtime Fabric reports worker node count incorrectly in some cases.

Added: More detailed information on the root cause if an installation fails.

Fixed: Tokenization health probes occasionally reported unhealthy under high CPU load.

Fixed: Runtime Fabric reports disconnected status when leader is re-elected in some cases.

Fixed: Cluster status is not reported after connection to control plane is re-established.

Fixed: Unable to restart services if Runtime Fabric is offline for more than 12 hours.

Fixed: Removed sensitive resource URLs from log output.

Fixed: Side-channel test of connection to control plane when running via proxy.

Fixed: Increased time allowed for connecting to control plane during installation.

Added: Monitoring dashboard in Ops Center to provide more visibility on the health of the inbound traffic controller and Mule applications.

Added: Support for directing outbound traffic of Runtime Fabric services through an HTTP and TCP proxy.

Added: Support for running Runtime Fabric services with an existing user and group ID.

Fixed: Ingress resources were not removed when inbound traffic is disabled.

Fixed: Some environments were not visible as a filter in Ops Center monitoring dashboard.

Fixed: Inbound traffic controller health probes occasionally reported unhealthy under high CPU load.

Fixed: The last mile security setting was not correctly applied in some cases.

Fixed: Application updates can cause disruption of inbound traffic in certain cases.

Support for the following Mule runtime and API gateway versions: