Equivalent of hash_always_miss

[fairbairn]

We want to run a cluster of nuster instances for rolling upgrades and disaster recovery.

To insure the caches are in sync across the cache cluster, we intend to integrate AWS EFS and use the newer configuration to share and check the disk first, insuring that the caches are properly coordinated.

We want to programmatically make a request through nuster to bypass the cache request and reach our back-end, to essentially refresh a single page with one request.

We do not want to PURGE the specific path, as it will create a temporary race condition through the nuster cluster for a period of time and make multiple requests downstream to our web server for the rendered page, which is a long-running request.

Instead, we would like to define a rule where where we might passing in a specific header value (known only to our internal services),so it performs the equivalent of a "hash_always_miss" varnish rule.

In this way, it takes only a single request to update the entire cache.

I have not yet found an example on how to define the rule to achieve this.

Is this possible in nuster?

[jiangwenyuan]

Hi

acl always_miss req.hdr(x-always-miss) -m found
nuster rule ...  if key always_miss

How about this?