Fixed
- (app) Close a channel instead of racing on a bool when the aggregator fails at startup in #1114 by @teemow
Full Changelog: https://github.com/giantswarm/muster/compare/v5.6.8...v5.6.9
Full Changelog: https://github.com/giantswarm/muster/compare/v5.6.8...v5.6.9
Full Changelog: https://github.com/giantswarm/muster/compare/v5.6.7...v5.6.8
Full Changelog: https://github.com/giantswarm/muster/compare/v5.6.6...v5.6.7
Full Changelog: https://github.com/giantswarm/klaus/compare/v0.0.309...v0.0.310
Full Changelog: https://github.com/giantswarm/klaus-operator/compare/v0.0.144...v0.0.145
Full Changelog: https://github.com/giantswarm/klaus-operator/compare/v0.0.143...v0.0.144
Full Changelog: https://github.com/giantswarm/muster/compare/v5.6.5...v5.6.6
muster test --readiness-timeout: the per-instance deadline for expected resources (tools, workflows, MCP servers) to become available after startup is now configurable (default 15s, previously hardcoded). (#1101)muster test no longer starts every scenario’s instance at t=0: at high --parallel the cold-start herd (50 Go processes at once) is what starved small CI containers into intermittent 15s readiness timeouts and scenario-timeout kills. Startups are now bounded (--startup-parallel, default 8) without limiting steady-state scenario parallelism, and spawned test instances run with GOMAXPROCS=2 so a container that reports the host’s core count (CircleCI docker executors allocate CPU by cgroup shares, which Go cannot right-size against) no longer produces fifty host-sized runtimes on a 4-vCPU slice. The suite banner prints GOMAXPROCS/NumCPU so CI logs show the runtime sizing. (#1101)Full Changelog: https://github.com/giantswarm/muster/compare/v5.6.3...v5.6.4
Full Changelog: https://github.com/giantswarm/muster/compare/v5.6.2...v5.6.3