Since you didn't reply to my reasoning for this patch, I have to ask more directly.
I didn't convince you of the usefulness of this conditional mode and you are not going to apply the patch, are you?
And if so, I could try to implement placement of the "limit_conn" into "if" context (not sure that I will be able to do that but at least I can try to). Will you agree to that feature?
_Any_ feedback is welcomed.
With best regards, Sergei Turchanov.
firstname.lastname@example.org писал 2020-09-23 00:39:
... Maxim Dounin писал 2020-09-22 23:19:
Quoting the documentation (http://nginx.org/r/limit_conn_zone):
: Requests with an empty key value are not accounted.
That is, limit_conn can be used conditionally without any code changes. The same applies to limit_req.
Yes, I know about that feature. The problem with that approach is a complication of the process to compute the value of a "key". We tried that before (with a cascade of "map" directives) and configuration became unreadable and very hard to understand when you add another indirection level to force a "key" to be an empty value depending on another condition. That was an incentive for this patch. This is especially true when there are several "limit_conn" directives in one location to limit user sessions by different identification attributes (and all of those attirbutes need to be "emptied" on a condition).
With best regards, Sergei Turchanov. _______________________________________________ nginx-devel mailing list email@example.com http://mailman.nginx.org/mailman/listinfo/nginx-devel