Event Rate Limit Configuration (v1alpha1)

limits are the limits to place on event queries received. Limits can be placed on events received server-wide, per namespace, per user, and per source+object. At least one limit is required.

type is the type of limit to which this configuration applies

qps is the number of event queries per second that are allowed for this type of limit. The qps and burst fields are used together to determine if a particular event query is accepted. The qps determines how many queries are accepted once the burst amount of queries has been exhausted.

burst is the burst number of event queries that are allowed for this type of limit. The qps and burst fields are used together to determine if a particular event query is accepted. The burst determines the maximum size of the allowance granted for a particular bucket. For example, if the burst is 10 and the qps is 3, then the admission control will accept 10 queries before blocking any queries. Every second, 3 more queries will be allowed. If some of that allowance is not used, then it will roll over to the next second, until the maximum allowance of 10 is reached.

cacheSize is the size of the LRU cache for this type of limit. If a bucket is evicted from the cache, then the allowance for that bucket is reset. If more queries are later received for an evicted bucket, then that bucket will re-enter the cache with a clean slate, giving that bucket a full allowance of burst queries.

The default cache size is 4096.

If limitType is 'server', then cacheSize is ignored.