mirror of https://github.com/OpenVidu/openvidu.git
e2d05aa8ef
Redeploying an Elastic or High Availability template on Azure always
failed, so no parameter could be changed after the first deployment:
- The Master Nodes' subnets were child resources outside the virtual
network. A redeploy PUTs the virtual network with the subnets it lists,
so ARM tried to delete the Master Nodes' subnets while they were in use
(InUseSubnetCannotBeDeleted). Every subnet is now declared inside the
virtual network; the symbolic names stay as `existing` references.
- Azure Automation returns a webhook's URI only when it creates it, so the
scale-in action group got an empty serviceUri on every redeploy ('Service
Uri' must not be empty). Every deployment now creates the scale-in webhook
under a new name, from a utcNow() parameter (new on every CLI deployment)
and the deployment's name (new on every portal Redeploy, which refills the
parameters with their previous values).
- The Media Nodes' scale set no longer sets its capacity: a redeploy with a
capacity below the scale-in-protected Media Nodes running was refused. The
autoscale profile's default seeds the size at creation.
Verified live on Elastic: two redeploys in a row succeeded. A redeploy still
cannot set the autoscale minimum or maximum below the Media Nodes running
(UnsupportedCapacity); the CHANGELOG and the documentation say how to remove
them first.
|
||
|---|---|---|
| .. | ||
| cf-openvidu-ha.bicep | ||
| cf-openvidu-ha.json | ||
| createUiDefinition.json | ||