Summary: | mod_proxy_fcgi and AH01068 Got bogus version X error message | ||
---|---|---|---|
Product: | Apache httpd-2 | Reporter: | Luca Toscano <toscano.luca> |
Component: | mod_proxy_fcgi | Assignee: | Apache HTTPD Bugs Mailing List <bugs> |
Status: | RESOLVED FIXED | ||
Severity: | normal | ||
Priority: | P2 | ||
Version: | 2.5-HEAD | ||
Target Milestone: | --- | ||
Hardware: | PC | ||
OS: | Linux | ||
Attachments: |
Force httpd to read the whole response from a FCGI backend even in presence of errors
Force httpd to read the whole response from a FCGI backend even in presence of errors |
Description
Luca Toscano
2016-07-10 10:26:09 UTC
Another thing to consider for a good solution is the "enablereuse=on" use case. The solution that I proposed should ensure that no data will be left to read generating spurious error for subsequent requests, but not super sure about it. After a chat with Jacob on IRC we figured out a simpler way to make this change, namely avoid breaking in the HTTP_NOT_MODIFIED branch case, since it is not an error condition. Added a new patch! Created attachment 34032 [details]
Force httpd to read the whole response from a FCGI backend even in presence of errors
First attempt to patch trunk in http://svn.apache.org/viewvc?view=revision&revision=1752347 As discussed on IRC it would be great to fix the HTTP_PRECONDITION_FAILED use case (HTTP 412) since it causes the following confusing error messages: AH01070: Error parsing script headers AH01075: Error dispatching request to : HTTP_PRECONDITION_FAILED is a value that can be returned by ap_scan_script_header_err_brigade_ex Tried to add another improvement to my change, namely http://svn.apache.org/r1754732 The new behavior that I am proposing should be used only when connection reuse is set to avoid inconsistencies. The issue has been backported to 2.4.x with http://svn.apache.org/r1757670 |