Hi, Currently, you can only do this based on source IP address. I know Willy is working to improve this and in1.5-dev9 you'll be able to do it, since you'll be able to track any users using strings (ie your token) then apply conn_rate or sess_rate, http_req_rate, etc... using gpc0 as a trigger.
cheers On Wed, May 2, 2012 at 5:47 AM, Ben Hood <0x6e6...@gmail.com> wrote: > Hi, > > I was wondering if HAProxy has the capability to rate limit HTTP POSTs > based on the response from the backend. > > The clients identify themselves with an token passed as query > parameter in the POST. I would like to implement the business logic > for calculating rate limits in my backend app. If the backend decides > that a limit has been breached for a particular client token, it would > response with a certain non-2xx code and would supply a TTL value to > express the period of time that the client will be throttled. > > Hence I was wondering whether it is possible to configure HAProxy to > cache the fact that a given client token is to be throttled for the > period expressed by the TTL. > > Any help is appreciated, > > Cheers, > > Ben >