Red Hat Bugzilla – Bug 135532
Bad exit codes for missing files in initscript
Last modified: 2007-11-30 17:07:04 EST
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; de-DE; rv:1.7.2) Gecko/20040806 Description of problem: The initscript /etc/rc.d/init.d/named contains the following checks: [ "${NETWORKING}" = "no" ] && exit 0 [ -x /usr/sbin/named ] || exit 0 [ -r ${ROOTDIR}/etc/named.conf ] || exit 0 Each of these should return '1' to indicate a failure. If, for example, named is started by the cluster manager and you forgot to provide a configuration, named is reported as started and running, even if it is not. Version-Release number of selected component (if applicable): bind-9.2.4-EL3_10 How reproducible: Always Steps to Reproduce: 1.delete /etc/named.conf 2.service named start/status 3.echo $? Actual Results: Gives '0' Expected Results: Should be '1' Additional info:
OK, in the next version, these checks will return 1 to indicate failure. These lines have remained unchanged since the very first version of named.init . I guess the rationale behind them is that the bind rpm install script always does a 'chkconfig --add named' on a clean install, and that if NETWORKING is disabled or named is not configured, the service is "disabled" and should not report errors on start up. This will be fixed in bind-9.2.4-2_EL3 .