Varnish Cluster using sharding director ?

[arthurzinck]

Hi Craig, we are pretty satisfied about how VarnishCluster works but, as we have 3 VarnishCluster replicas for resiliency we need 3 requests to build up cache on the 3 instances.

I was wondering if there is a way to use sharding as they would do it here in this example.
And avoid going back to the app, if one of the Varnish instances may have it in cache. But automatically set up by the controller inside the backends.vcl.tmpl

Thanks,
Arthur

[cin]

There sure is. :) You can change the behavior by altering the VCL to suit your needs, but here are the docs we have around the feature. https://github.com/IBM/varnish-operator/blob/54c5aff649c8c13652d232a19d8451d8ac37ee1e/docs/vcl-configuration.md#advanced-vcl-with-varnish-sharding. Let me know if you have any problems getting it setup.

[arthurzinck]

This is awesome, thank you for your answer.
I seemed to have issue with the acl_cluster in the example you provided me. This configuration was sending a 503 Backend fetch failed until I modified the example a bit and add the /32 inside the brackets.
I have done a PR to fix this, maybe you will agree to merge it : #105

[cin]

@arthurzinck I merged this change into my fork. For now (and maybe going forward), my fork is only going to be maintained.

[arthurzinck]

Hello Craig, yes for sure. I will be using your fork from now one 👍
If you need help maintaining I can give a hand from time to time