Questo contenuto non è disponibile nella lingua selezionata.
Chapter 1. OpenShift Service Mesh release notes
Review new features, compatibility updates, fixed issues, and known issues for Red Hat OpenShift Service Mesh to stay informed about changes across different product versions.
1.1. Red Hat OpenShift Service Mesh version 3.5 new features and enhancements Copia collegamentoCollegamento copiato negli appunti!
This release makes Red Hat OpenShift Service Mesh 3.5 generally available, adds new features, addresses Common Vulnerabilities and Exposures (CVEs), and is supported on OpenShift Container Platform 4.20 and later.
For lists of supported component versions (such as the Istio control plane) and feature support levels, see:
- "Service Mesh component versions"
"Service Mesh feature support tables"
- Ambient mode multi-primary multi-network topology is generally available
The ambient mode multi-primary multi-network topology (a multi-cluster service mesh topology), introduced in OpenShift Service Mesh 3.3 as Technology Preview, is fully supported in this release. Each cluster runs its own control plane and communicates through ambient east-west gateways. With HBONE (HTTP-Based Overlay Network Environment) mutual TLS, workloads communicate securely across clusters. Before deploying this topology, review the documented configuration requirements and known limitations.
- IPv6 single-stack and dual-stack networking support for ambient mode is generally available
OpenShift Service Mesh can operate on clusters with IPv6 single-stack or dual-stack networking, enabling deployment in cloud environments that require IPv6.
To enable dual-stack operation:
- Configure your Kubernetes cluster for dual-stack networking (Kubernetes 1.23 or later).
-
Set the
ISTIO_DUAL_STACKenvironment variable totruein your Istio configuration. Configure
pilot.ipFamilyPolicytoRequireDualStackin your Istio installation.After you configure these settings, ambient proxies automatically route traffic to both IPv4 and IPv6 endpoints without requiring changes to applications. Organizations can operate a single mesh across different network types rather than maintaining separate infrastructure for each protocol.
- Security-hardened configuration for internet-facing gateway proxies
Gateway proxies can use a new
EDGEprofile that applies security-focused settings for internet-facing deployments. The profile configures a subset of Envoy edge proxy hardening recommendations through standard Istio APIs, reducing the need for customEnvoyFiltersfor these settings.The
EDGEprofile provides:- Defaults for Envoy listener and cluster buffers, HTTP timeouts, and downstream HTTP/2 limits.
- Enhanced header and path handling for edge-proxy deployments.
Support for configuring a global downstream connection limit through
ProxyConfig.connectionSettings.profile.Enable the profile by setting
ProxyConfig.connectionSettings.profiletoEDGE. Existing gateway configurations are not affected unless the profile or individualConnectionSettingsfields are explicitly configured.
- Integration with the Zero Trust Workload Identity Manager is generally available
With this release, OpenShift Service Mesh integration with the Red Hat Zero Trust Workload Identity Manager (ZTWIM) Operator is generally available. ZTWIM deploys and manages SPIRE (the SPIFFE Runtime Environment), which provides deep workload attestation for Service Mesh. This feature was introduced as a Technology Preview in Service Mesh 3.4.
SPIRE replaces the built-in Istio certificate authority (CA). Envoy sidecar proxies and gateways receive X.509 SPIFFE Verifiable Identity Documents (SVIDs) from the SPIRE Agent through the Secret Discovery Service (SDS) API. This gives you stronger workload attestation and central control of workload identities.
Service Mesh 3.5 also adds support for SPIRE federation. With federation, workloads in different trust domains can authenticate each other by exchanging SPIFFE trust bundles. Use cases include multiple OpenShift clusters that each run their own SPIRE server, and external workloads that run SPIRE agents outside the mesh.
Known limitations:
- The integration supports sidecar mode only.
-
You must create SPIRE registrations before you start workload pods. For example, use
ClusterSPIFFEIDresources. A pod without a matching registration waits indefinitely for an identity. -
SPIFFE IDs must use the format
spiffe://<trust_domain>/ns/<namespace>/sa/<service_account>. -
The SPIRE server trust domain must match the Istio
meshConfig.trustDomainvalue. For federation, you must list each federated trust domain and its
spiffeBundleUrlinmeshConfig.caCertificateson every participating cluster.
- Upgrade OpenShift Service Mesh two versions at a time in sidecar mode
You can upgrade sidecar mode deployments two versions at a time with the
RevisionBasedstrategy, reducing upgrade time and workload disruption. For example, upgrading directly from OpenShift Service Mesh 3.3 to 3.5 requires only one workload restart instead of multiple restarts at each intermediate version.The
InPlacestrategy still upgrades OpenShift Service Mesh one version at a time.- Upstream enhancements and behavior changes in OpenShift Service Mesh 3.5
As an enterprise distribution based on upstream Istio, OpenShift Service Mesh inherits enhancements from the core project, Envoy, and the broader community ecosystem. Review the following upstream changes in this release to decide whether your deployment needs configuration updates.
-
Unhealthy endpoint routing: With this release, Istio routes traffic to unhealthy endpoints when most healthy endpoints are unavailable, preventing complete service failure. If you already configured
OutlierDetection.minHealthPercentin aDestinationRule, this new default behavior does not apply to that service. To disable this behavior globally, set thePILOT_AUTO_SEND_UNHEALTHY_ENDPOINTSenvironment variable tofalsein youristioddeployment. Gateway resource merging: The new
PILOT_ENABLE_STRICT_GATEWAY_MERGINGenvironment variable prevents IstioGatewayresources in different namespaces from merging with managed Gateway APIGatewayresources. (Unmanaged Gateway APIGatewayresources are not affected.) This new variable is enabled by default. To disable it, setPILOT_ENABLE_STRICT_GATEWAY_MERGINGtofalse.ImportantIf your existing mesh configuration merges Istio
Gatewayresources with managed Gateway APIGatewayproxies across namespaces, either refactor your gateway setup or disable strict gateway merging by settingPILOT_ENABLE_STRICT_GATEWAY_MERGINGtofalse.Gateway distributed tracing: The
PILOT_SPAWN_UPSTREAM_SPAN_FOR_GATEWAYfeature flag was removed. As a result, gateway distributed tracing of requests to backend services is always enabled. If you had set this flag to false, you should remove that configuration, because it no longer has effect.For complete details, see the upstream Istio release documentation:
- Istio 1.31 change notes
- What’s new in Istio 1.31
-
Unhealthy endpoint routing: With this release, Istio routes traffic to unhealthy endpoints when most healthy endpoints are unavailable, preventing complete service failure. If you already configured
1.2. Red Hat OpenShift Service Mesh version 3.5 Technology Preview features Copia collegamentoCollegamento copiato negli appunti!
This release includes some features that are currently in Technology Preview. These experimental features are not intended for production use.
For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.
- External VM workload integration (Technology Preview)
You can include external RHEL virtual machine workloads in OpenShift Service Mesh, whether they are managed by Red Hat OpenShift Virtualization or running as standalone VMs.
A Red Hat-built RPM package installs the Istio proxy on the external VM, so that the VM joins the mesh. The Istio control plane manages the VM workload. As a result, the VM workload benefits from standard mesh capabilities, such as mutual TLS, observability, traffic management, and authorization policies.
Mesh management of external VM workloads is a Technology Preview feature.
- New Kiali perspective for monitoring service meshes on RHACM-managed clusters (Technology Preview)
If you run meshes on clusters managed by Red Hat Advanced Cluster Management for Kubernetes (RHACM), the OpenShift Service Mesh Console offers a new perspective: the Fleet Service Mesh perspective. It provides a hub-wide inventory of Istio meshes and control planes across the RHACM fleet.
The Fleet Service Mesh perspective is a Technology Preview feature.
For more information, see OSSMC Fleet Service Mesh Technical Preview.
- Automated metric scraping configuration for Istio telemetry (Technology Preview)
Metric scraping for the Istio control plane and sidecar proxies can be automatically configured, enabling you to view mesh metrics in Prometheus, Kiali, and the OpenShift web console without manually creating monitoring resources.
To enable automated metric scraping, add the
sailoperator.io/monitoring: enabledannotation to anIstiocustom resource. The Sail Operator reconcilesServiceMonitorandPodMonitorresources with default metrics scraping configuration for each mesh. ExistingServiceMonitorandPodMonitorresources that you created manually are not overwritten.Automated metric scraping configuration is a Technology Preview feature.
- Weighted waypoint canaries (Technology Preview)
In Istio ambient mode, you can safely test configuration and version updates to waypoint proxies before rolling them out across your mesh. Use this feature to run a canary waypoint alongside your primary waypoint and shift a configurable percentage of traffic to verify changes under real conditions. Services and namespaces reference the canary proxy by using the
istio.io/use-waypoint-canaryandistio.io/use-waypoint-canary-namespacelabels, with the traffic split controlled by theistio.io/use-waypoint-canary-weightannotation.Weighted waypoint canary proxies are a Technology Preview feature.
For more information, see the upstream Istio release documentation: Istio 1.31 change notes.
1.3. Red Hat OpenShift Service Mesh version 3.5 fixed issues Copia collegamentoCollegamento copiato negli appunti!
OpenShift Service Mesh inherits fixes for Common Vulnerabilities and Exposures (CVEs) and bugs from upstream Istio and related projects.
- Information about CVEs
- See OpenShift Service Mesh Security Advisories.
- Upstream fixed issues in OpenShift Service Mesh 3.5
-
DestinationRuleand backend policy precedence: When both aDestinationRuleand a Gateway API backend policy (BackendTLSPolicyorXBackendTrafficPolicy) target the same host,DestinationRulefields take precedence. This fix ensures the backend policy only fills in fields that theDestinationRuleleaves unset, regardless of which resource was created first. -
Consistent hash load balancing:
DestinationRuleconfigurations that useconsistentHashproperly send traffic to new endpoints after scaling. -
Subset TLS configuration: When a
DestinationRulesubset does not specify a TLS mode for a port, it correctly falls back to the top-level traffic policy TLS setting. This fix ensures endpoint mTLS mode is properly inherited. -
Zone-aware routing configuration: Setting
zoneAwareLbSetting.enabled: falseproperly disables Envoy zone-aware routing. This fix prevents uneven traffic distribution during gradual rollouts when some pods have self-discovery enabled and others do not.
-
For complete details, see the upstream Istio release documentation: Istio 1.31 change notes
1.4. Red Hat OpenShift Service Mesh version 3.5 removed features Copia collegamentoCollegamento copiato negli appunti!
Removed functionality no longer exists in the product.
- Service Mesh
must-gatherimage removed This release removes the
must-gatherimage specifically for debugging OpenShift Service Mesh. Instead, use the Service Mesh debugging capabilities in the default OpenShift Container Platformmust-gatherimage.For more information, see the OpenShift Container Platform documentation: About the must-gather tool