Locating Server Metric Counters

In some scenarios, it may be necessary to monitor server metrics outside of the Advanced Server Monitoring Agent. Many of these metrics can be replicated by following the reference table below.

 

For a description of the various metrics collected, please see the Server Metrics section.

Windows

Most performance counters for Windows machines are also available to developers using Perfmon.

 

Counter Name
Perfmon Counter Path
Processor Counters

CPU %

\Processor(_Total)\% Processor Time

Context switches/sec

\System\Context Switches/sec

Process Queue Length

\System\Processor Queue Length

Memory Counters

Memory %

\Memory\% Commited Bytes In Use

Page reads/sec

\Memory\Page Reads/sec

Page Writes/sec

\Memory\Page Writes/sec

Cache Memory Allocation

\Memory\Cache Bytes

Disk Counters

% I/O Time Utilized

\PhysicalDisk(_Total)\% Idle Time

(Calculation taken as 100 - counter value)

Service time

\PhysicalDisk(_Total)\Avg. Disk sec/Transfer

Queue Length

\PhysicalDisk(_Total)\Current Disk Queue Length

Reads/sec

\PhysicalDisk(_Total)\Disk Reads/sec

Writes/sec

\PhysicalDisk(_Total)\Disk Writes/sec

Network Counters

Note: for values coming from a Network Interface, the Server Monitoring Agent will record the sum of all instances, excluding the local loop-back interface.

Packets Received/sec

\Network Interface(interface name)\Packets Received/sec

Packets Sent/sec

\Network Interface(interface name)\Packets Sent/sec

Bytes received/sec

\Network Interface(interface name)\Bytes Received/sec

Bytes sent/sec

\Network Interface(interface name)\Bytes Sent/sec

Packets Received Errors

\Network Interface(interface name)\Packets Received Errors

Packets Sent Errors

\Network Interface(interface name)\Packets Outbound Errors

Connections Established

\TCP\Connections Established

Connection Failures

\TCP\Connection Failures

TCP Segments Retransmitted

\TCP\Segments Retransmitted/sec

 

Linux

Performance counters for Linux are calculated by examining the following values, if available. Some counters may be measured from multiple sources, in which case a source will be selected which appears applicable to the current kernel.

 

Counter Name
File
Relevant lines, columns
Processor Counters

CPU %

/proc/stat

Line: cpu

4th numeric column is idle time

Context switches/sec

/proc/stat

 

Line: ctxt

Process Queue Length

/proc/stat

 

Line: procs_running

Memory Counters

Memory %

/proc/meminfo

Lines: MemTotal, MemFree, Buffers, Cached

Cache Memory Allocation Ratio

/proc/meminfo

Lines: MemTotal, Cached, SwapCached

Major Page Faults

/proc/vmstat

Line: pgmajfault

Page ins/sec

/proc/stat

Line: page, Column: 1

/proc/vmstat

Line: pgpgin

Page outs/sec

/proc/stat

Line: page, Column: 2

/proc/vmstat

Line: pgpgout

Disk Counters

Note: Counters are measured by summing values from all physical disks, excluding logical disks

% I/O Time Utilized

/proc/partitions

Column: 14

/proc/diskstats

Column: 13

Average Service time

/proc/partitions

Columns: 5, 9, 13, 15

/proc/diskstats

Columns: 4, 8, 12, 14

Queue Length

/proc/partitions

Column: 13

/proc/diskstats

Column: 12

Reads/sec

/proc/partitions

Column: 5

/proc/diskstats

Column: 4

Writes/sec

/proc/partitions

Column: 9

/proc/diskstats

Column: 8

Network Counters

Note: for values coming from a Network Interface, the Server Monitoring Agent will record the sum of all instances, excluding the local loop-back interface.

Packets Received/sec

/proc/net/dev

Column: 3

Packets Sent/sec

/proc/net/dev

Column: 11

Bytes received/sec

/proc/net/dev

Column: 2

Bytes sent/sec

/proc/net/dev

Column: 10

Packets Received Errors

/proc/net/dev

Column: 4

Packets Sent Errors

/proc/net/dev

Column: 12

Collisions/sec

/proc/net/dev

Column: 15

Connections Established

/proc/net/snmp

Line: Tcp, Column 9

Connection Failures

/proc/net/snmp

Line: Tcp, Column 8

TCP Segments Retransmitted

/proc/net/snmp

Line: Tcp, Column 12