Skip to content

Thin down available Infoblox connections #463

Closed
@kuritka

Description

Related to #462

When zone delegation creation via EdgeDNS provider API fails (error), the reconciliation loop resets immediately.

k8gb is trying to connect but the provider's API can be under a load of waiting connections, so the loop for a single GSLB looks like this:
10s (connecting..., timeout error) + 2s (reconciliation reset) + 10s (connecting..., timeout error) + 2s (reconciliation reset) + 10s + 2s,….

Connection is recreated because the reconciliation loop returns an error. If we return nil and wait until the reconciliation period passes, the loop for a single GSLB will look like: 10s (connecting..., timeout error) + RECONCILE_REQUEUE_SECONDS (recover) (30s by default) + 10s (connecting..., timeout error) + RECONCILE_REQUEUE_SECONDS (recover)…CONNECTED

With waiting for the next reconciliation loop instead of the reset loop, the Infoblox will not be under connection load.
There will be more time to share the available connections between all gslbs. In addition, it doesn't produce such amount of timeout errors.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions