v1.1.1

This patch release includes some bug fixes for Chaos Dashboard and IoChaos.

Enhancements

• Supports setting node selector when deploying chaos-daemon #1382

Fixes

• Chaos Dashboard
  • Chore: delete archives also delete related events #1372
  • Fixed: missing stressors checking #1391
  • Fixed: scope pods' isolation problem #1399
  • Fixed: sets config for sqlite3 to avoid error database is locked #1413
  • Fixed: resets failed message when applying chaos action successfully #1429
  • Fixed: corrects the scheduler and annotation filling after loading an experiment from LoadFrom or external #1438
• toda
  • Fixes ENOENT after renaming files ba1783
  • Fixes failing to detach if the process has died f96954
  • Ignores ESRCH when attaching tasks 2a4192
  • Fixes the behavior of utimensat 6dd650
  • Fixes unexpected lock before injecting latency
  • Fixes ENOENT caused by inode cache

v1.1.0

Chaos Mesh v1.1.0 refactors Chaos Dashboard component and makes it easier to use, and introduces authentication to restrict users’ permissions and actions. We also implemented DNSChaos to help users simulate DNS failures. In addition, to help users debug Chaos Mesh, we introduced a new tool, chaosctl, to provide debug information.

New Feature & Enhancements

• Refactors Chaos Dashboard and supports authentication
  • Refactors dashboard homepage and some visible changes #1285
  • Adds authority for archiving experiments and events #1261
  • Sets ns automatically #1343
  • Supports set config to disable token #1249
  • Supports using token generating from Kubernetes to login #1188
  • Adds networkchaos external targets in dashboard #1168
  • Uses YAML format as experiment description #1029
  • Adds global search for events, experiments, and archives #1034
• chaosdaemon: supports more tc filters #1309 #1325
• chaosdaemon: makes network device configuration #1242
• Makes selector support expression selectors #1277
• Implements chaosctl tool for print debug info #1074
• Supports set protocol for iptables and tc, make it compatible with chaos-daemon #1179
• Improves controllers #1014 #1256 #1221
• IoChaos: supports specifying container name #1159
• Implements DNSChaos to simulate DNS failures #1039 #935
• Adds host-network flag in install.sh #954
• Updates toda to v0.1.11 #1367

Major Bug Fixes

• Fixed: fail to reuse archived experiment by chaos-dashboard #1323

v1.0.3

This patch release includes bug fixes for Scheduler, NetworkChaos, IoChaos, and Chaos Dashboard.

Enhancements

• controller-manager now connects chaos-daemon by using endpoints to find chaos-daemon IP, instead of using hostPort directly #1253
• toda updated to v0.1.9 and nsexec is used to replace nsenter #1280

Fixes

• Some fixes for Chaos Dashboard
  • Fixed: Chaos Dashboard panic due to invalid TargetScope filed #1210
  • Fixed: repeated parsing of attr field when creating IoChaos #1223
• Fixed: NetworkChaos/IoChaos slow recovery #1241
• Fixed: scheduler error nextRecover shouldn’t be later than nextStart #1284

v1.0.2

This patch release has some fixes for IoChaos and Chaos Dashboard.

Enhancements

• Disable privileged mode when deploying Chaos Daemon component #1126
• Update toda to v0.1.8 #1146

Fixes

• Import more DB drivers to fix the issue that MySQL cannot be used as the DB store #1076
• Some fixes for Chaos Dashboard
  • Fix error handling in Chaos Dashboard to avoid panic #1105
  • Update API response and add the response to POST/PUT request #1119
  • Make Chaos Dashboard be compilable with namespace scoped mode #1135
  • Fix unexpected namespace via "Load From Archives" during creating a new experiment #1128
  • Fix repeated pod records in chaos event #1141
  • Fix the failure of resubmission experiment after failed to submit a chaos experiment #1158

v1.0.1

This patch release has some fixes for IoChaos, NetworkChaos, and Chaos Dashboard.

Fixes

• Add a check to avoid injecting NetworkChaos on a pod with hostNetwork: true #1017
• Fix the issue that NetworkChaos does not take effect on the GKE platform #1038
• Some fixes for Chaos Dashboard
  • Fix the failure of setting the direction of the network partition #1063
  • Fix the failure of setting namespace selectors and label selectors #1041

Enhancements

• Add host-network flag in install.sh #1044
• Update toda to v0.1.4 #1056

v1.0.0

On September 25, 2020, Chaos Mesh v1.0 is now released! We have made some significant improvements in this release, including the refactor of IoChaos, providing more security guarantees for chaos experiments, supporting the setting of protected Namespaces, and supporting the use of Namespace’s permissions. In addition, this release is also focused on Chaos Dashboard and Chaos Operator components, which helps users create more stable, safe, and controllable chaos experiments.

New Feature & Enhancements

• New implementation of IoChaos
  • Refactor the API and controller of IoChaos #932
  • Implement the toda project, a hook filesystem for injecting IO chaos https://github.com/chaos-mesh/toda
  • Add pod status in IoChaos #958
• Support namespace scoped chaos #872
• Record the error information on the Chaos object #931
• Update Reconcile logic and make updated cron take effect immediately #920
• Add command to check install script #916
• Add hostNetwork option for controller-manager #874

Major Bug Fixes

• Fix the error in retrying failed NetworkChaos #949
• Fix the issue that fields selectors do not work #920
• Fix the abnormal error when deleting chaos object #943
• Fix incomplete experiments caused by the restart of Chaos Dashboard #933
• Fix the bug caused by the empty target mode #930
• Fix invalid usage of cgo/unsafe pointer in TimeChaos #904
• Fix the selectors issue in new experiments scope page #881

v0.9.1

This patch release has some important fixes to StressChaos and NetworkChaos.

Fixes

• Fix the issue that controller-manager panics due to concurrent access to map when to run stress chaos #855
• Fix the error in building TC commands #867
• Fix the issue of failure to install SideCar template by running install.sh #861
• Always recreate controller-manager pod while upgrading helm charts #875

v0.9.0

Chaos Mesh v0.9.0 mainly introduces the Chaos Dashboard component, which is the web UI for users to manage and monitor chaos experiments. In this version, NetworkChaos has been refactored to support setting multiple network attack rules on the same Pod at the same time, a one-click installation script is added to help users quickly get started with Chaos Mesh.

New Features & Enhancements

• Introduce Chaos Dashboard component
  • Support creating/updating/deleting/pausing PodChaos, NetworkChaos, StressChaos, TimeChaos, IoChaos, KernelChaos through the web interface #481
  • Support directly uploading YAML files through the interface to create chaos experiments #665
  • Support obtaining specific fault injection event details through the interface #628
  • Support direct reuse of archived chaos configurations #783
• Supports forcibly cleaning chaos experiment by setting chaos experiment object annotations to #415 #478
• Support the use of installsh script to quickly install Chaos Mesh #466 #506 #511
• Add a new sidecar configuration template to simplify the IoChaos configuration file #502
• Support setting protected namespaces #661
• Support injecting StressChaos into a specified container in the Pod #759 #794
• Refactor NetworkChaos to support setting multiple network attack rules on the same Pod #788

Major Bug Fixes

• Fix burn-memory not taking effect #776
• Fix the failure to restore NetworkChaos #788

v0.8.0

Chaos Mesh v0.8.0 provides the ability to orchestrate chaos experiments in the Kubernetes environment, with support of comprehensive types of failure simulation, including Pod failures, container failures, network failures, file system failures, system time failures, and kernel failures. Helm installation method is also supported so that users can quickly deploy Chaos Mesh for chaos experiments. Chaos Mesh uses YAML to define chaos experiments, and provides a rich range of preset chaos test samples for users to quickly try Chaos Mesh.

New Features and Enhancements

• Add PodChaos to simulate the failure on Pods and Containers, including Pods and Containers being killed, Pods being continuously unavailable
• Add NetworkChaos to simulate network failures, including delay, packet duplication, packet loss, partition, etc
• Add TimeChaos to simulate failures on the system clock, such as clock skew
• Add IOChaos to simulate failures on the file system, including file system I/O delay, and file system I/O errors
• Add KernelChaos to simulate kernel failures
• Add StressChaos to simulate CPU burn and Memory burn
• Support rich selectors to specify the scope of the chaos experiment
• Support rich schedulers, including using cron to schedule chaos experiments
• Support pausing a chaos experiment provisionally
• Support defining chaos experiments using YAML file
• Support ValidatingAdmissionWebhook for verifying the chaos object
• Support cert-manager for certificate management
• Support deploying Chaos Mesh using Helm to
• Support saving metrics using Prometheus
• Support recording information of chaos experiment in Kubernetes events
• Support the complete e2e testing framework