Add retry_back_off into envoy

[DmitroDemchenko]

Good morning, currently there are a lot of useful params that we can pass via annotations like num-retries, per-try-timeout, retry-on an so on.
Is it possible to configure also retry_back_off / base_retry_backoff_ms ? Because default is 25ms and maximum interval is 10 times in Envoy and I'd like to increase it little beat.
Please help with it.

[cheshir]

Hi there. Also looking for a solution to tweak base retry delay a bit.

[vadim-sychov]

Recently I faced the same problem, anyone knows how can to solve this?

[youngnick]

Thanks for logging this issue @DmitroDemchenko. We can certainly look at adding support for configuring the retry back off. Can I ask how you (and everyone else watching this issue) are configuring Contour?

In general, we try not to add more annotations to our Ingress support, but encourage people to use either HTTPProxy or the forthcoming Gateway API support. That's not to say we won't, but we want to confirm we'll actually be adding a thing you need.

[DmitroDemchenko]

Hi, @youngnick , I'm adding Contour in my k8s cluster. Deploying it via helm chart. And I need different ingress resources for my applications. When I configure Ingress resource I use this annotations:
annotations:
kubernetes.io/ingress.class: contour
projectcontour.io/num-retries: "10"
projectcontour.io/response-timeout: "60s"
projectcontour.io/retry-on: "5xx,gateway-error,reset,connect-failure"

But I'd like increase time between retries, and found opportunely in envoy do it via retry_back_off. If it possible now do it another way, not in annotations please let me know.

[cheshir]

Hi, @youngnick

I'm using contour in front of hundreds of sites (SEO park). Sites are deployed on GKE preemptible nodes.
I want to configure backoff to cover scenario when user trying to open site and it's nodes rotates in the same time.

[xaleeks]

+1 that this is useful

Thanks for sharing @DmitroDemchenko & @cheshir. If you don’t mind, please also share your usage on this ticket #1269

[github-actions]

The Contour project currently lacks enough contributors to adequately respond to all Issues.

This bot triages Issues according to the following rules:

• After 60d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, the Issue is closed

You can:

• Mark this Issue as fresh by commenting
• Close this Issue
• Offer to help out with triage

Please send feedback to the #contour channel in the Kubernetes Slack

[github-actions]

The Contour project currently lacks enough contributors to adequately respond to all Issues.

This bot triages Issues according to the following rules:

• After 60d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, the Issue is closed

You can:

• Mark this Issue as fresh by commenting
• Close this Issue
• Offer to help out with triage

Please send feedback to the #contour channel in the Kubernetes Slack