Summary: | vhosts priority order with wildcard | ||
---|---|---|---|
Product: | Apache httpd-2 | Reporter: | Ninos <me> |
Component: | Core | Assignee: | Apache HTTPD Bugs Mailing List <bugs> |
Status: | REOPENED --- | ||
Severity: | enhancement | CC: | me |
Priority: | P2 | ||
Version: | 2.2.22 | ||
Target Milestone: | --- | ||
Hardware: | PC | ||
OS: | Linux |
Description
Ninos
2014-05-07 23:03:34 UTC
This is working as desined -- If there's a wildcard and an exact match IP in the <virtualhost>, the exact match wins and the name-based vhost will be picked from the best IP-based match. I think the design is a little bit wrong or wrong implemented. The vhosts with the wildcard should have a higher priority on a not defined vhost. Here're some examples: EXAMPLE 1 Visit: Domain: example1.com IP: IP1 Defined vhost: Domain: example2.com IP: IP1 Priority: Current: IP1:example2.com -> wildcard Should be: wildcard -> IP1:example2.com EXAMPLE 2 Visit: Domain: example1.com IP: IP1 Defined vhost: Domain: example2.com IP: IP2 Priority: Current: wildcard -> IP2:example2.com Should be: wildcard -> IP2:example2.com EXAMPLE 3 Visit: Domain: example1.com IP: IP1 Defined vhost: Domain: * IP: IP1 Priority: Current: IP1:* -> wildcard Should be: IP1:* -> wildcard What I mean is, if all vhosts have a defined ServerName (expect whildcard), the vhosts shouldn't have a higher priority than the wildcard IF the domain can't be matched with a vhost. If someone want's to match all undefined ip-requests to a vhost, he can also use a wildcard for the ServerName. (Servername *,example1.com, *.example1.com). This design would be cleaner and more logical. Not a bug, this is by design. IP/interface based mapping first, name-based mapping second. If you just want name-based, you have to limit yourself to all *:port. Otherwise, ip-based discrimination happens first. I understand the current mapping and also know now, that's not a bug. In my second comment I asked for a little modified design, because for me my example seems more logical. :) (In reply to Ninos from comment #4) > I understand the current mapping and also know now, that's not a bug. Reclassifying as an enhancement. I don't think it's well defined here thouh. |