What is this?

This knowledgebase contains questions and answers about PRTG Network Monitor and network monitoring in general.

Intuitive to Use. Easy to manage.
More than 500,000 users rely on Paessler PRTG every day. Find out how you can reduce cost, increase QoS and ease planning, as well.

Free Download

How do I monitor system parameters like memory, CPU and disks on Linux systems via SNMP?

Votes:

0

I want to monitor system parameters like CPU load and memory on a Linux or UNIX system via SNMP. How does PRTG support this?

howto installation linux networking oidlib planning prtg snmp

Created on Feb 12, 2010 2:06:23 PM by Daniel Zobel [Product Manager]

Last change on Feb 12, 2010 3:27:26 PM by Daniel Zobel [Product Manager]

Permalink

1 Reply

Accepted Answer

Votes:

0

This article applies to PRTG Network Monitor 13 or later, as well as to previous (deprecated) versions

1. Install the SNMP daemon on the Linux Server

Please see the articles Checklist: Setting up SNMP on Linux and How do I install the SNMP daemon on Linux machines? for instructions.

2. Creating sensors in PRTG Network Monitor

Create a device for the Linux machine you want to monitor (enter this computer's IP address or DNS name).
On this device, create Linux sensors which are natively available in PRTG. You can find a list of the out-of-the-box sensors in the PRTG Manual.
Alternatively, you can add an SNMP Library sensor. Please follow the instructions below:
From the appearing menu window (drop-down menu in older PRTG versions), select the Basic linux library (ucd-snmp-mib).oidlib SNMP library file.
Click on OK ("Continue to step 2" in older PRTG versions)
On the next page you will see a list of available sensors. Select the sensors you want to monitor by marking the according checkboxes. PRTG will create one sensor for each marked Library OID.

3. Available Sensors of the Basic Linux Library (UCD-SNMP-MIB)

Here is a list of sensors that the Basic Linux Library supports. You may see less entries on your servers. This depends on the OS, your system and the (security) settings inside the NET-SNMP daemon.

Entries marked as "Table: yes" will provide the measured value for several items (e.g. for each CPU, disk or network card).

Group: "Processes"

Update: We now strongly recommend the following approach instead:

Monitoring processes in Linux Using the approach below is still possible, but the deployed sensors aren't as reliable as the ones from the link above.

Sensor name	Table?	Meaning
Processes Index	yes	Reference Index for each observed process.
Processes Count	yes	The number of current processes running with the name in question.
Processes Error Flag	yes	A Error flag to indicate trouble with a process. It goes to 1 if there is an error, 0 if no error.

Group: "Memory"

Sensor name	Table?	Meaning
Memory Total Swap	no	Total Swap Size configured for the host.
Available Swap Space	no	Available Swap Space on the host.
Memory Total Real	no	Total Real/Physical Memory Size on the host.
Available Real/Physical Memory	no	Available Real/Physical Memory Space on the host.
Memory Total Swap used by text	no	Total virtual Memory used by text.
Active virtual Memory used by text	no	Active virtual Memory used by text.
Memory Total Real used by text	no	Total Real/Physical Memory Size used by text.
Active Real/Physical Memory used by text	no	Active Real/Physical Memory Space used by text.
Memory Total Free	no	Total Available Memory on the host.
Memory Shared	no	Total Shared Memory
Memory Buffered	no	Total Buffered Memory
Memory Cached	no	Total Cached Memory
Memory Swap Error Flag	no	Error flag. 1 indicates very little swap space left

Group: "Disk"

Sensor name	Table?	Meaning
Disk index	yes	Integer reference number (row number) for the disk mib.
Disk minimum	yes	Minimum space required on the disk (in kBytes) before the errors are triggered. Either this or dskMinPercent is configured via the agent's snmpd.conf file.
Disk min percent	yes	Percentage of minimum space required on the disk before the errors are triggered. Either this or dskMinimum is configured via the agent's snmpd.conf file.
Disk total	yes	Total size of the disk/partion (kBytes)
Disk avail	yes	Available space on the disk
Disk used	yes	Used space on the disk
Disk percent	yes	Percentage of space used on disk
Disk percent node	yes	Percentage of inodes used on disk
Disk error flag	yes	Error flag signaling that the disk or partition is under the minimum required space configured for it.

Group: "System Stats"

Sensor name	Table?	Meaning
System Memory Swapped In	no	Amount of Memory swapped in from disk (kB/s).
System Memory Swapped Out	no	Amount of Memory swapped to disk (kB/s).
System IO sent (deprecated)	no	Blocks sent to a block device (blocks/s). Deprecated, replaced by the ssIORawSent object
System IO received (deprecated)	no	Blocks received from a block device (blocks/s). Deprecated, replaced by the ssIORawReceived object
System Interrupts (deprecated)	no	The number of interrupts per second, including the clock. Deprecated, replaced by ssRawInterrupts
System Context Switches (deprecated)	no	The number of context switches per second. Deprecated, replaced by ssRawContext
System CPU user (deprecated)	no	percentages of user CPU time. Deprecated, replaced by the ssCpuRawUser object
System CPU system (deprecated)	no	percentages of system CPU time. Deprecated, replaced by of the ssCpuRawSystem object
System CPU idle (deprecated)	no	percentages of idle CPU time. Deprecated, replaced by of the ssCpuRawIdle object
System CPU raw user	no	user CPU time in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second.
System CPU raw nice	no	nice CPU time in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second.
System CPU raw system	no	system CPU time in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second.
System CPU raw idle	no	idle CPU time in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second.
System CPU raw wait	no	iowait CPU time in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second. This is primarily a SysV thingie
System CPU raw kernel	no	kernel CPU time in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second.
System CPU raw interrupt	no	interruptlevel CPU time. in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second. This is primarily a BSD thingie
System IO raw sent	no	Number of blocks sent to a block device
System IO raw received	no	Number of blocks received from a block device
System Raw Interrupts	no	Number of interrupts processed
System Raw Context Switches	no	Number of context switches
System CPU raw soft irq	no	Soft IRQ CPU time. in ticks per second, as reported by the kernel. Total ticks is 100xNumber of Processors. For single processor machine numbers will appear to be percentages as the kernel will tally ticks at 100 per second. This is for Linux 2.6
System Raw Blocks Swapped In	no	Number of blocks swapped in
System Raw Blocks Swapped In	no	Number of blocks swapped out

Group "CPU Load"

Sensor name	Table?	Meaning
CPU Load (1 minute average)	no	The load average is the system load over a period of time. An idle computer has a load number of 0. Each proceSystem that is using CPU, waiting for CPU or is in uninterruptible sleep (usually waiting for disk activity) adds 1 to the load number.
CPU Load (5 minute average)	no	The load average is the system load over a period of time. An idle computer has a load number of 0. Each proceSystem that is using CPU, waiting for CPU or is in uninterruptible sleep (usually waiting for disk activity) adds 1 to the load number.
CPU Load (15 minute average)	no	The load average is the system load over a period of time. An idle computer has a load number of 0. Each proceSystem that is using CPU, waiting for CPU or is in uninterruptible sleep (usually waiting for disk activity) adds 1 to the load number.