NET-SNMP version: 5.5
OS: CentOS release 6.7 (Final)
Kernel: 3.10.0-229.24.2.el6.CUSTOM.x86_64
Issue description
The following set of OIDs defined in HOST-RESOURCES-MIB file provides information about cached memory on system :
.1.3.6.1.2.1.25.2.3.1.3.7 -> hrStorageDescr = STRING: Cached memory
.1.3.6.1.2.1.25.2.3.1.4.7 -> hrStorageAllocationUnits
.1.3.6.1.2.1.25.2.3.1.5.7 -> hrStorageSize
.1.3.6.1.2.1.25.2.3.1.6.7 -> hrStorageUsed
This schema corresponds to the one defined in RFC 2790, the Host Resources MIB.
HrStorageEntry ::= SEQUENCE {
hrStorageIndex Integer32,
hrStorageType AutonomousType,
hrStorageDescr DisplayString,
hrStorageAllocationUnits Integer32,
hrStorageSize Integer32,
hrStorageUsed Integer32,
hrStorageAllocationFailures Counter32
}
When performing checks on the above set of OIDs, external management tools (eg.SolarWind) reports errors about low disk space.
Alert description : Disk space low on host "172.19.110.32 "NTNX-15SM65450073-B-CVM . Free space is 0. Volume Cached memory
Total disk size: 3.2 G
Disk Free space : 0
Disk used : 3.2G(100 %)
Alert Generated time : 6/18/2016 11:29 PM
Investigation shows that both hrStorageSize and hrStorageUsed are collected from /proc/meminfo:cached entry.
The management tools perform a diff between the two values and that implies that every time the check is executed, independently from the used value, the result will always be the same:
Disk Free space : 0
So, it seems to me that, at least at Linux level, this check is broken. Am I missing something?
Seems same as https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=596339, affecting debian.
There is not really a "size" for cached. It would say that you should ignore the value from your management application. The same goes for buffers and shared.