[checkmk-commits] Check_MK Git: check_mk: #1564 FIX check_mk_agent.linux: fix situation where async plugin is not executed after crash

Module: check_mk Branch: master Commit: 7104b7963f82cf1403abfd9bdbf2923a0b74ea2b URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=7104b7963f82cf… Author: Mathias Kettner &lt;mk(a)mathias-kettner.de&gt; Date: Wed Dec 3 15:06:17 2014 +0100 #1564 FIX check_mk_agent.linux: fix situation where async plugin is not executed after crash When executing asynchronous sections or plugins the Check_MK Agent uses <tt>run_cached()</tt> for putting them in the background. This function creates a cachefile with the extension <tt>.new</tt> and removes that suffix when the actual plugin has finished. If due to a server crash the file keeps lying around the the plugin would not be executed anymore. We now remove that file if it is older then twice the cache age <i>and we kill</i> any process that still has open that file. That way we avoid overlapping plugins in case of a real hanger. --- .werks/1564 | 17 +++++++++++++++++ ChangeLog | 1 + agents/check_mk_agent.linux | 4 ++++ 3 files changed, 22 insertions(+) diff --git a/.werks/1564 b/.werks/1564 new file mode 100644 index 0000000..27feb51 --- /dev/null +++ b/.werks/1564 @@ -0,0 +1,17 @@ +Title: check_mk_agent.linux: fix situation where async plugin is not executed after crash +Level: 1 +Component: checks +Compatible: compat +Version: 1.2.5i7 +Date: 1417615386 +Class: fix + +When executing asynchronous sections or plugins the Check_MK Agent uses +<tt>run_cached()</tt> for putting them in the background. This function +creates a cachefile with the extension <tt>.new</tt> and removes that suffix +when the actual plugin has finished. If due to a server crash the file keeps +lying around the the plugin would not be executed anymore. + +We now remove that file if it is older then twice the cache age <i>and +we kill</i> any process that still has open that file. That way we avoid +overlapping plugins in case of a real hanger. diff --git a/ChangeLog b/ChangeLog index f3d3dce..2680676 100644 --- a/ChangeLog +++ b/ChangeLog @@ -60,6 +60,7 @@ * 1602 FIX: dell_om_esmlog: Fixed typo in plugin output * 1563 FIX: Reworked configuration of process monitoring... NOTE: Please refer to the migration notes! + * 1564 FIX: check_mk_agent.linux: fix situation where async plugin is not executed after crash... Multisite: * 1508 Allow input of plugin output and perfdata when faking check results... diff --git a/agents/check_mk_agent.linux b/agents/check_mk_agent.linux index abb95f2..c734bb1 100755 --- a/agents/check_mk_agent.linux +++ b/agents/check_mk_agent.linux @@ -81,6 +81,10 @@ function run_cached () { if [ -e "$CACHEFILE.new" ] ; then local CF_ATIME=$(stat -c %X "$CACHEFILE.new") if [ $((NOW - CF_ATIME)) -ge $((MAXAGE * 2)) ] ; then + # Kill the process still accessing that file in case + # it is still running. This avoids overlapping processes! + fuser -k -9 "$CACHEFILE.new" >/dev/null 2>&1 + rm -f "$CACHEFILE.new" return fi fi