[checkmk-commits] 5941 FIX Fixed nagios using outdated config when not using "cmk -R/-O"

Module: check_mk Branch: master Commit: 2df7ba9f2c2700849ec15af8e74c563f54b93f03 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=2df7ba9f2c2700… Author: Lars Michelsen <lm(a)mathias-kettner.de> Date: Thu Mar 29 09:37:11 2018 +0200 5941 FIX Fixed nagios using outdated config when not using "cmk -R/-O" When using "omd start/restart" or the init script etc/init.d/nagios to start or restart nagios, it was possible that the core started with an outdated configuration. During start/restart the nagios init script is working in two steps: 1. config validation and 2. start/restart of the nagios process. During both steps the whole object configuration (etc/nagios/conf.d) needs to be processed. To optimize this the first step is perfoming some precompiling of the read config files. The result is saved in the file var/nagios/objects.precache and read by the core during startup instead of the single object config files. The init script had an option set (CORE_NOVERIFY=yes) in all cases that prevented the config verification even when executed manually. Without the config verification the object precache was not updated before the restart, making the core load an outdated file. The variable CORE_NOVERIFY=yes should only be set when the config verification has already been done. For example cmk -R/-O is doing exactly this. A workaround is to trigger the restarts using "cmk -R". Change-Id: I92cb76637094e5a136561423fe3a6bbfed2e1267 --- .werks/5941 | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/.werks/5941 b/.werks/5941 new file mode 100644 index 0000000..e320653 --- /dev/null +++ b/.werks/5941 @@ -0,0 +1,28 @@ +Title: Fixed nagios using outdated config when not using "cmk -R/-O" +Level: 1 +Component: packages +Compatible: compat +Edition: cre +Version: 1.5.0i4 +Date: 1522308528 +Class: fix + +When using "omd start/restart" or the init script etc/init.d/nagios to +start or restart nagios, it was possible that the core started with an +outdated configuration. + +During start/restart the nagios init script is working in two steps: 1. config +validation and 2. start/restart of the nagios process. During both steps the +whole object configuration (etc/nagios/conf.d) needs to be processed. To +optimize this the first step is perfoming some precompiling of the read config +files. The result is saved in the file var/nagios/objects.precache and read +by the core during startup instead of the single object config files. + +The init script had an option set (CORE_NOVERIFY=yes) in all cases that +prevented the config verification even when executed manually. Without the +config verification the object precache was not updated before the restart, +making the core load an outdated file. The variable CORE_NOVERIFY=yes should +only be set when the config verification has already been done. For example cmk +-R/-O is doing exactly this. + +A workaround is to trigger the restarts using "cmk -R".