The test case for PR 45434 seems to have regressed across 2.2->2.4. https://issues.apache.org/bugzilla/show_bug.cgi?id=45434
I have not tried to understand the mechanics here, but a dumb side-by-side analysis found a missing piece, below. 2.2 hardcodes this as "real + 11" but 2.4 uses the constant elsewhere. Any reason why this would be wrong? It fixes the test case I added to t/modules/proxy.t. 2.2.x code for reference: http://svn.apache.org/viewvc/httpd/httpd/branches/2.2.x/modules/proxy/proxy_util.c?annotate=1058624#l1090 Index: modules/proxy/proxy_util.c =================================================================== --- modules/proxy/proxy_util.c (revision 1365029) +++ modules/proxy/proxy_util.c (working copy) @@ -860,7 +860,7 @@ (balancer = ap_proxy_get_balancer(r->pool, sconf, real, 1))) { int n, l3 = 0; proxy_worker **worker = (proxy_worker **)balancer->workers->elts; - const char *urlpart = ap_strchr_c(real, '/'); + const char *urlpart = ap_strchr_c(real + sizeof(BALANCER_PREFIX) - 1, '/'); if (urlpart) { if (!urlpart[1]) urlpart = NULL;
