general/381: spare servers stay alive after a failed restart via SIGUSR1

[Ei...@hyperreal.com, Lars <sf...@unix-ag.org>]

>Number:         381
>Category:       general
>Synopsis:       spare servers stay alive after a failed restart via SIGUSR1
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    apache (Apache HTTP Project)
>State:          open
>Class:          sw-bug
>Submitter-Id:   apache
>Arrival-Date:   Sun Apr 13 13:10:04 1997
>Originator:     sfx@unix-ag.org
>Organization:
apache
>Release:        1.2b7
>Environment:
Linux 2
>Description:
 When doing a 'kill -1 httpd.pid' the httpd
 is restarted unless there is an error
 in one of the *.conf files. When an error
 occurs no new server and child servers are
 started... that's ok. But there's IMHO a
 little problem when using -SIGUSR1 to
 gracefully restart the server.
 If there is an error somewhere in the
 config files the 'kill -SIGUSR1 fails',
 but only the httpd 'root' process is
 killed and the child servers stay alive.
 At least the stay alive until a client
 connects which gets a 'document contains
 no data error' (this request makes the
 child servers disappear, at least this
 was the case when I tested it).

>How-To-Repeat:

>Fix:
 So if the Apache cannot be restarted
 via SIGUSR1 it should take care that
 all servers process are shut down.
 But maybe it's possible to parse the
 *.conf files _before_ Apache tries
 to restart with the new config files?
 Is it?

 On a busy site -SIGUSR1 is a very good
 idea, but a faulty change in a config
 files shuts the servers down when
 trying to restart it... To refuse the
 restart request if a config file contains
 an error and output a message about it
 is maybe helpful to webmasters of very
 busy sites who are afraid to see there
 servers down for more than just a second.
%0
>Audit-Trail:
>Unformatted: