Hi Olivier, Thanks for the detailed questions — happy to clarify how src_conn_cur behaves in these scenarios.
*1. HTTP/2 multiplexing* src_conn_cur counts *streams*, not TCP connections. So in your example, if a client opens a single TCP connection and issues 20 parallel HTTP/2 streams, src_conn_cur will indeed show *20*. This is expected, since the counter reflects concurrent request streams associated with the source, not the underlying transport connection. If you want a metric that reflects only the number of TCP connections, the counters to look at are the ones tied to connection-level tracking (e.g., stick tables using conn_rate, conn_cur, etc.), which operate strictly at the TCP layer and won’t multiply with HTTP/2 streams. *2. Interaction with http-request set-src* Yes — src_conn_cur is evaluated *after* any set-src rewrite. Once the source address is modified, the counter uses the rewritten value for lookups and increments. This allows you to group or normalize sources intentionally, but it also means that rewriting can affect rate/connection limits if you rely on those counters. Let me know if you’d like examples of stick-table configurations or ways to separate TCP‑level and stream‑level tracking. Best regards, Thanks & Best Regards Henry PAN Sr. Lead Cloud Architect (425) 802--3975 https://www.linkedin.com/in/henrypan1 On Wed, Feb 18, 2026 at 1:10 AM Olivier D <[email protected]> wrote: > Greetings, > > I am currently using HAProxy 2.8 and would like to clarify the exact > behavior of the src_conn_cur counter. > > From my understanding, src_conn_cur is intended to count the number of > concurrent connections from a single source IP address. However, I > have a couple of questions regarding its behavior in specific > scenarios: > > 1/ HTTP/2 multiplexing behavior > In HTTP/2, multiple requests can be multiplexed over a single TCP > connection. > If a client opens one TCP connection and issues, for example, 20 > parallel resource requests over that same connection, does > src_conn_cur increment to 20 in that case? > It seems to be that way from what I see but would be happy to have > confirmation on this point. > If so, is there another counter that operates strictly at the TCP > connection level (i.e., that would count only 1 in this situation)? > > 2/ Interaction with http-request set-src > Does src_conn_cur take into account modifications made by http-request > set-src? > In other words, if the source address is rewritten using set-src, is > the counter evaluated against the modified source value? > > Thank you in advance for your clarification. > > Best regards, > Olivier > > >
