[Bug 595796] New: "killproc" violates LSB
http://bugzilla.novell.com/show_bug.cgi?id=595796 http://bugzilla.novell.com/show_bug.cgi?id=595796#c0 Summary: "killproc" violates LSB Classification: openSUSE Product: openSUSE 11.3 Version: Factory Platform: All OS/Version: SLES 11 Status: NEW Severity: Major Priority: P5 - None Component: Basesystem AssignedTo: werner@novell.com ReportedBy: werner@novell.com QAContact: qa@suse.de CC: lmb@novell.com, ro@novell.com, kukuk@novell.com Depends on: 578246 Found By: Development Blocker: --- +++ This bug was initially created as a clone of Bug #578246 +++ I've just spent some time hunting down why the cluster daemon was dieing during shutdown, and couldn't find anything in the logs. (Because I somehow didn't expect that a SIGKILL would be send magically out of nowhere.) Then, I stumbled across a suse-specific killproc oddity: killproc sends signals to all processes that use the specified exe‐ cutable. If no signal name is specified, the signal SIGTERM is sent. If this program is not called with the name killproc then SIGHUP is used. Note that if SIGTERM is used and does not terminate a process the signal SIGKILL is send after a few seconds (default is 5 seconds, see option -t). If a program has been terminated successfully and a veri‐ fied pid file was found, this pid file will be removed if the termi‐ nated process didn't already do so. .. -t<sec> The number <sec> specifies the seconds to wait between the sent signal SIGTERM and the subsequentially signal SIGKILL if the first SIGTERM does not show any result within the first few milli seconds. This defaults to 5 seconds. So, if -TERM is specified, this automatically escalates to KILL (this escalation cannot be disabled). However, LSB states (http://refspecs.freestandards.org/LSB_3.1.1/LSB-Core-generic/LSB-Core-generi...): killproc [-p pidfile] pathname [signal] The killproc function shall stop the specified program. The program is found using the algorithm given above. If a signal is specified, using the -signal_name or -signal_number syntaxes as specified by the kill command, the program is sent that signal. Otherwise, a SIGTERM followed by a SIGKILL after an unspecified number of seconds shall be sent. [...] Note that this says that this automatic escalation only happens if no signal is specified, not if the specified signal happens to be the default SIGTERM. ("killproc" on RHEL/Fedora behaves like this.) I think killproc must differentiate between "No signal specified" versus "SIGTERM explicitly specified". The impact here is real, since it could (and did!) break 3rd party init scripts in unexpected ways. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=595796 http://bugzilla.novell.com/show_bug.cgi?id=595796#c1 Dr. Werner Fink <werner@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED --- Comment #1 from Dr. Werner Fink <werner@novell.com> 2010-04-12 15:52:51 UTC --- submitted -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=595796 http://bugzilla.novell.com/show_bug.cgi?id=595796#c2 Johannes Meixner <jsmeix@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jsmeix@novell.com --- Comment #2 from Johannes Meixner <jsmeix@novell.com> 2010-07-15 10:34:19 UTC --- FYI: For a possible unexpected side effect of the fix here see bug 622058 - or perhaps even a more general issue: "<service> restart" doesn't wait for stop before issuing start -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=595796 http://bugzilla.novell.com/show_bug.cgi?id=595796#c3 --- Comment #3 from Bernhard Wiedemann <bwiedemann@suse.com> --- This is an autogenerated message for OBS integration: This bug (595796) was mentioned in https://build.opensuse.org/request/show/42970 Factory / cups -- You are receiving this mail because: You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply@novell.com