↓Skip to main content

Release Radar - Q1 2023

Thanos 0.30
#

TLDR: new caching engine

  • --tsdb.out-of-order.time-window flag facilitates out-of-order sample ingestion
  • Redis client deprecated in favour of Rueidis (in Golang, faster, supports client-side caching)
  • Promtheus bumped to v0.40.7, but sans native histogram capability

Argo Rollouts 1.4
#

TLDR: fast rollbacks

  • minPodsPerReplicaSet param for canary deploys (new canaries must have x min pods)
  • ‘fast rollback’ capability (skip analysis steps), with configurable gate for count of previous revisions
  • dynamic shell completion for core CLI commands
  • reliability improvements to leader election logic
  • support for Apache APISIX (API gateway, ingress controller etc)

Coroot 0.9 - 0.12
#

TLDR: k8s workload autodiscovery

  • automatic Kubernetes workload observability via kube-state-metrics discovery
  • memory leak detection (defaults to +10MB/h, can be modified globally or per-project)
  • AMQP (RabbitMQ) protocol support
  • JVM awareness - out of box monitors for availability and stop-the-world pauses (enduring safepoint stops)

VictoriaMetrics 1.84 - 1.86
#

TLDR: pushgateway support

  • /api/v1/import/prometheus endpoint now (also) supports ingest of pushgateway format
  • support for migration of data originating from sources supporting Prometheus remote read API
  • drilldown into per-job or per-instance metric exports now possible vmui
  • support for discovery of Hashicorp Nomad services (nomad_sd_configs)
  • parsing of partial RFC3339 query args (eg 2022 becomes 2022-01-01T00:00:00Z)
  • support for aggregation of samples (based on time + label) in vmagent

Gitlab 15.8
#

TLDR: SCIM for self-managed instances

  • support for blocking MR’s unlesss external status checks pass (flagged behind only_allow_merge_if_all_status_checks_passed)
  • beta of ‘direct transfer’, a refactored group and project transfer approach (for both intra and inter instance transfers)
  • self-hosted GitLab deploys now SCIM compatible (programatically create/remove identities)
  • access requests create To-Do items for relevant (group/project) admins

Tailscale 1.36
#

TLDR: (Apple) Shortcuts

  • --json output for many more commands (eg version, lock status, lock logs)
  • support for UPnP port mapping of gateways when deployed in HA
  • Shortcuts actions for macOS and iOS clients
  • First-party container no longer fails when $TS_KUBE_SECRET is unset

Liqo 0.7
#

TLDR: Terraform provider

  • support for connectivity to parent cluster API server from offloaded pods
  • first-party Terraform provider (peering, namespace config etc)
  • better handling of failed worker nodes in remote cluster

vcluster 0.14
#

TLDR: in-tree CRD sync

  • new ‘multi-namespace’ mode, syncs child namespaces to parent cluster (as opposed to being nested inside parent cluster’s vcluster namespace)
  • generic-crd-sync plugin effectively moved in-tree as ‘generic resource sync’ capability, facilitating CRD sync and child cluster access to parent cluster controllers
  • Kubernetes 1.26 support for all three vcluster distros (k3s, k0, and upstream)

Prometheus 2.41 - 2.42
#

TLDR: X.509 hot reloads

  • keep_firing_for rule parameter
  • support for HTTP proxy headers
  • hot reload for file-bound certificates

VictoriaMetrics 1.87
#

TLDR: stream aggregation

  • PROXY protocol support for most components (needs allowlisting, eg -httpListenAddr.useProxyProtocol)
  • support for deduping input samples via new streamAggr CLI args
  • dark mode and local fonts for vmui web interface

Mimir 2.6
#

TLDR: reduced peak memory utilization

  • experimental (feature flagged) store-gateway memory enhancements (streaming instead of buffering)
  • grafanaExploreURL and tenantID template functions
  • experimental store-gateway reliability improvements (index header reads without using mapped memory)

Kyverno 1.9
#

TLDR: policy exceptions

  • PolicyException custom resource, facilitates exclusions decoupled from policies themselves
  • ClusterCleanupPolicy custom resource, for granular, automated removal of specified cluster resources
  • OpenTelemetry support (added in v1.8) improved, now includes distributed tracing capability

Tempo 2.0
#

TLDR: TraceQL

  • Parquet graduates to GA, and is now the default backend
  • initial introduction of TraceQL (new query language)
  • distributor support for (generic and OTel gRPC) forwarders

ArgoCD 2.6
#

TLDR: plugin param’s

  • syncOptions.createNamespace=true sub-params for adding labels and annotations
  • support for parametarization of CMP’s (plugins)
  • progressive rollout support for ApplicationSet resources [alpha]
  • new admin initial-password CLI subcommand prints bootstrapped admin credential

Nomad 1.5
#

TLDR: OIDC support

  • support for SSO, through addition of broad OIDC provider compatibility
  • API endpoint for updating node metadata (so also possible via CLI and web UI)
  • CLI and web UI quickstarts via new provide via ‘job templates’ (syntax boilerplate)

Boundary 0.12
#

TLDR: more authentication options

  • support for mapping Vault credentials directly to a user/target (previously required target-per-credential)
  • credential injection using Vault SSH certificates (HCP only)
  • support for sessions requiring multi-hop traffic pathing (HCP only)

Cilium 1.13
#

TLDR: Gateway API

  • support for L7 load balancing of Services via service.cilium.io/lb-l7 annotation, including in ClusterMesh scenarios
  • upstream conformant implementation of Gateway API (for N-S traffic flows, effectively their v2 take on ingress orchestration)
  • LoadBalancer IPAM + BGP service announcement facilitate dynamical allocation of external-facing IP’s on bare metal, as is done on your behalf ‘in the cloud’ (effectively obviating the need for MetalLB)
  • Hubble datasource plugin for Grafana ties together Prometheus metrics / Tempo trace emissions with Timescape [Enterprise]
  • CiliumNodeConfig CRD facilitates node-by-node migration from other CNI’s (supersedes global config)
  • first-party container images now ship signed (via cosign), and with a SBOM (in SPDX format)
  • Tetragon file integrity monitoring (akin to Sysdig) [Enterprise]
  • “BIG TCP” support, enabling higher throughput whilst maintaining low latency on 100Gbps+ capable networks
  • IPv6 NAT46/64 translation capability
  • SCTP (often a VoIP transport) support
  • K8s 1.26 internalTrafficPolicy support
  • datapath mTLS support (precursor but not yet enabler of real world mTLS capability for Cilium’s mesh)

Dapr 1.10
#

TLDR: multi-app workflows

  • pluggable component SDK’s for Golang, .NET and Java; self-hosted binaries for bespoke extensions to the framework
  • API endpoints (and CRD’s) for bulk publish and subscribe operations
  • (alpha) ‘Workflows’ API for building long running or persistent workflows, including those spanning multiple app
  • support for Cloudflare Workers KV as SQLite3 as state stores

GitLab 15.9
#

TLDR: guest access to private repos

  • read access to private repositories for Guest role [Ultimate]
  • support for requiring multiple CODEOWNERS approvers
  • closed beta of Code Suggestions (GH Copilot competitor) [SaaS Ultimate]
  • gitlab-sshd (in Golang) support for self-managed instances
  • displayable Discord ID’s for user profile pages
  • support for filtering by primary (programming) language on Explore Projects page

VictoriaMetrics 1.88
#

TLDR: VM remoteWrite

  • vmagent support for VictoriaMetrics improved remote write protocol (lower network bandwidth + disk I/O)
  • vmagent discovery of Kuma Control Plane targets (via new kuma_sd_configs)
  • vmalert support for reading alerting and recording rules from an S3-compatible backend [Enterprise]
  • vmauth now re-attempts failing GET requests on all configured backends before returning a result to the client

Coroot 0.13 - 0.14
#

TLDR: Pyroscope integration

  • more notification sinks: Teams, PagerDuty and Opsgenie (Slack already supported prior)
  • integration with Pyroscope (a continuous profiling platform), facilitating deep dives into process execution, correlated against collected metrics (like CPU, I/O etc)
  • support for adding an additional Prometheus selector to all queries (enabling aggregate metrics from federated backends eg Thanos/Mimir)

Grafana 9.4
#

TLDR: better keyboard navigation

  • Canvas panel (introduced in v9.3) enhanced via support for data links and arrows (shapes)
  • topnav and topNavCommandPalette flags unlock new [CMD/CTRL]+K keyboard-driven navigation UX
  • better search and export of Grafana Alerting rules
  • newPanelChromeUI flag unearths redesigns to existing panel types

Teleport 12.0
#

TLDR: many more auth methods

  • transparent Azure and GCP CLI authentication (adding to existing support for the same in AWS)
  • Device Trust - authentication via macOS Secure Enclave, branded as [Alpha, Enterprise]
  • X.509 based (‘passwordless’) auth for Windows local users [Enterprise]
  • Kubernetes pod-level RBAC, facilitating more granularily than when role mapping K8s users/groups

Vault 1.13
#

TLDR: multi-namespace access ergonomics

  • Azure auth method support for Functions and App Service workloads (Virtual Machines already supported)
  • cross-namespace secret sharing
  • event-based notifications (via websockets API) [Alpha]
  • Vault Operator (first-party K8s secrets sync) [Beta]

Terraform 1.4
#

TLDR: improved CLI output

  • terraform_data resource, a replacement for the lion’s share of null provider use cases
  • OPA results emitted in plan and apply output, similar to that of TFC Sentinel policies
  • Terraform Cloud’s structured run output now surfaced in terraform CLI output (if using TFC backend)

Prometheus 2.43
#

TLDR: split scrape_configs into separate files

  • new scrape_config_files block
  • no_proxy (exclude a subset of URLs) and proxy_from_environment (read from ENV) args
  • promtool support for HTTP query commands (aligns with existing amtool functionality)

Thanos 0.31
#

TLDR: per series sample limits

  • Redis Sentinel support (addition of master_name param)
  • --disable-caching-index-header-file facilitates Store component running without persistent disk
  • --store.limits.request-{series,samples} options
  • experimental support for Query component load balancing across multiple Store components