Community

Loading...

Azure Chaos Studio

25 Ideas

Stephane Lapointe

Support for App Services & Functions in Chaos Studio

Beside VM, NSG & AKS, it would be super useful if we could have App Services & Azure Functions as targets in Chaos Studio.

Company Response

Thank you for your suggestion! We now have a "Stop App Service" fault available in the fault library. We recognize there are more App Service related faults that might be useful, so please feel free to keep the conversation going and submit additional ideas for fault behavior.

Mike Wedderburn-Clarke (HE/HIM)

Availability Zone and Region failures

Ability to simulate entire AZ and Region failure. For AZ, ideally Chaos Studio would identify all resources in an AZ and stop them or block network access to them. Ideally it would also be able to simulate AZ outage on ZRS and other natively multi-AZ resources so we can ensure there are no unexpected dependencies.

Company Response

Thank you for the suggestion! As we continue expanding our fault library, AZ and Region failures are scenarios we want to enable. Today, you can perform this for Virtual Machine Scale Sets via dynamic targeting -- see the following tutorial on using dynamic targeting to shut down Virtual Machine Scale Sets in a certain availability zone: Create a chaos experiment that uses dynamic targeting to select hosts. Additionally, the Network Security Group Rules fault can simulate downstream or cross-region dependency outage: Chaos Studio fault and action library.

Giuseppe Lacagnina

Support Container Apps in Chaos Studio

As far as I know, Chaos Studio does not currently support Container Apps (they are not listed among possible Targets). I would like to be able to conduct Chaos Experiments on my ecosystem, which is implemented with Container Apps.

3rd party REST API Support

We would like to call 3rd party (non-Azure) REST API. For example, when simulating ExpressRoute and On-Prem network failure, it would be better if Azure Chaos Studio support 3rd party (network provider's) REST API.

Support for Private AKS Chaos Testing

We'd like to try Chaos Testing on our private AKS clusters. Currently, there is no support in Chaos Studio. Unless directly use Chaos Mesh.

Company Response

Hi Stanley! Chaos Studio now supports private networks via VNet Injection. You can find the documentation here: Integration of VNet Injection with Chaos Studio, and the Azure Update article here: Chaos studio - Public preview updates for January 2023.

Integrate with Azure Function for custom faults

In order to be able to implement complex or third party faults, it would be nice to integrate Chaos Studio with Azure function. Trigger a function to fire the fault, and another one for resetting the initial state for example.

Support AKS faults with local accounts disabled

Customers have requested using AKS Chaos Mesh faults when AKS local accounts are switched off for security reasons - currently this isn't supported. Upvote this item to help us prioritize!

Support built-in roles

Some customers have requested built-in roles in Chaos Studio to make managing permissions easier. Currently, permissions can be assigned via custom roles incorporating Chaos Studio's ARM operations. Please vote on this feedback item to help show us that you want or need built-in roles!

New

Pre/post conditions

Some customers have requested support for pre/post-conditions in Chaos Studio, such as checking and/or setup before running experiments. Please vote on this feedback item to help show us that this is important!

Sandhya Mohanan Anoop

Litmus Chaos usage for Simulating AKS Faults

I have been using Chaos mesh but I feel litmus offers more capabilities, The main differences or interesting facts I noticed are below: Chaos Workflows : Litmus Chaos provides predefined chaos workflows that encapsulate complex chaos experiments. These workflows make it easier for users to quickly execute common chaos scenarios without having to write custom chaos experiments from scratch. Web UI: More user friendly and extensive UI which enabled a faster turnaround. Wide Variety of experiments for Kubernetes: Litmus library for Kubernetes experiments are wide like many Pod Chaos, Node Chaos and even Application Chaos. Also it offers Azure Instance Stop and Disk loss. I have not tested all but documentation looks interesting and promising. Well maintained and I see active updates to the code repository. Where as Chaos mesh recently updated “As we currently haven't had enough maintainers to support the release cycle, We will release new versions from time to time, with a cycle of about 6 months.”. It would be great if we have a flexibity in terms of AKS agents being used. Kindly look into this idea.