* Monitor the amount of dirty memory because we seem to have a problem with that. (`mount -o remount' or `sync' can take hours to complete on the Hydra server because there are gigabytes of dirty memory. During that time other processes become very slow and sit at 100% system time. Maybe related to http://bugzilla.kernel.org/show_bug.cgi?id=14830)
* Monitor the amount of dirty memory because we seem to have a problem with that. (`mount -o remount' or `sync' can take hours to complete on the Hydra server because there are gigabytes of dirty memory. During that time other processes become very slow and sit at 100% system time. Maybe related to http://bugzilla.kernel.org/show_bug.cgi?id=14830) svn path=/configurations/trunk/tud/; revision=20180
Eelco Dolstra committed
Feb 22, 2010 at 19:53 UTC
13cf5beaac8b3c6503557dbb71d24ef3b3542124
1 file changed
+1
common.nix
+1
@@ -24,5 +24,6 @@
24
services.zabbixAgent.extraConfig =
25
''
26
UserParameter=hardware.temp.cpu.average,cat /sys/devices/platform/coretemp.*/temp1_input | ${pkgs.perl}/bin/perl -e 'while (<>) { $n++; $sum += $_; }; print $sum / $n / 1000;'
27
+ UserParameter=vm.memory.dirty,echo $((1024 * $(cat /proc/meminfo | sed 's/Dirty: *\([0-9]\+\) kB/\1/; t; d')))
28
'';
29
}