Bug 41109 - connection broken when keepalive=on & maxrequestperchild are set
Summary: connection broken when keepalive=on & maxrequestperchild are set
Alias: None
Product: Apache httpd-2
Classification: Unclassified
Component: mpm_worker (show other bugs)
Version: 2.0.59
Hardware: PC Linux
: P2 major with 2 votes (vote)
Target Milestone: ---
Assignee: Apache HTTPD Bugs Mailing List
Keywords: PatchAvailable
Depends on:
Reported: 2006-12-05 03:26 UTC by Cyryl Płotnicki-Chudyk
Modified: 2015-01-16 20:38 UTC (History)
1 user (show)

2.3.x patch (ap_graceful_stop_signalled) (687 bytes, patch)
2007-06-18 00:56 UTC, Dmytro Fedonin
Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Cyryl Płotnicki-Chudyk 2006-12-05 03:26:45 UTC
when max request per child is reached , then the thread is immediately killed ,
there's no wait for closure / no reset of the tcp connections saved along
keep-alive feature

such apache's behaviour makes third-party modules , such as mod_webobjects, feel
as the user killed connection while receiving response after his/her request
Comment 1 Dmytro Fedonin 2007-06-15 01:20:42 UTC
I see the same behavior on 2.2.x and 2.3.x. And not only current connection will
be broken but also at least one (up to ThreadsPerChild) following connection
will ignore KeepAlive and terminate immediately after first request is served.
Noticeable amount of broken connections makes the effect annoying.
Patch will follow.
Comment 2 Dmytro Fedonin 2007-06-18 00:56:18 UTC
Created attachment 20362 [details]
2.3.x patch (ap_graceful_stop_signalled)

It fixes return value of ap_graceful_stop_signalled() to be 0 all the time. Fix
for 2.2.x is the same.
Comment 3 Eric Covener 2015-01-16 20:38:37 UTC
old ticket, behavior seems to work fine even on 2.2.x (long running CGI holds httpd process open)