Destructive Chaos Testing

[Bolodya1997]

• [x] Decomposition - https://github.com/networkservicemesh/sdk/issues/823
• [x] Sandbox healing simple cases - https://github.com/networkservicemesh/sdk/issues/898
• [ ] Sandbox healing complex cases - https://github.com/networkservicemesh/sdk/issues/899
• [x] Integration tests - https://github.com/networkservicemesh/deployments-k8s/issues/1174
• [ ] Integration tests (multi death cases) - https://github.com/networkservicemesh/deployments-k8s/issues/1928
• [ ] Refactor healing - https://github.com/networkservicemesh/sdk/issues/950

Blockers

1. clock in sandbox - #785
2. ~registry chain healing - #713~
3. ~issue with endpoint restore healing - #918~ - no more blocker since we removed restore cases
4. ~issue in connect client / heal client - #925~
5. ~client healing - #938~

[Bolodya1997]

[Bolodya1997]

@edwarnicke Please share your thoughts about what do we need from this scheme in 1.0?

[denis-tingaikin]

@Bolodya1997 Could you also provide estimations for each of the todo/in progress tasks?

[Bolodya1997]

Approximate estimations:

1. Registry healing [in progress] - 1d
2. Client healing - 2d
3. NSMgr k8s testing - 5d
   • 1 death case - 2d
   • rest cases - 3d
4. Forwarder k8s testing - 2d
5. Endpoint k8s testing - 2d
6. Registry k8s testing - 2d
7. Sandbox simple testing [in progress] - would be finished by sandbox rework, clock in sandbox, registry healing, client healing
8. Sandbox complex testing - 5d
9. Restore name issue [in progress] - 2d

And so additionally there should be some time for review and merge.

[denis-tingaikin]

@edwarnicke

Currently, we plan to deliver these things till release if we'll not face critical bugs:

Registry healing [in progress] - 1d
Client healing - 2d
NSMgr k8s testing
* 1 death case - 2d
Forwarder k8s testing - 2d
Endpoint k8s testing - 2d

Let us know if you want to edit this list.