0% found this document useful (0 votes)
8 views

014 Reviewing Metrics and Logging Data Using Performance Co-Pilot Part 2 Study Guide

014 Reviewing Metrics and Logging Data Using Performance Co-Pilot Part 2 Study Guide

Uploaded by

fqchina
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
8 views

014 Reviewing Metrics and Logging Data Using Performance Co-Pilot Part 2 Study Guide

014 Reviewing Metrics and Logging Data Using Performance Co-Pilot Part 2 Study Guide

Uploaded by

fqchina
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Reviewing Metrics and Logging

Data Using Performance Co-Pilot

Mike Bunch
LINUX TRAINING ARCHITECT
REVIEWING METRICS AND LOGGING DATA USING PERFORMANCE COPILOT

Lesson Objectives and Overview

Lesson Objective

Review and demonstrate various Performance Co-Pilot tools (e.g., pminfo, pcp atop, and pcp
mpstat) and use them to view various system metrics.

Lesson Notes
Performance Co-Pilot (PCP)

PCP was developed by Silicon Graphics (SGI and first included in RHEL 6.6. It has become the
preferred performance metrics analysis tool due to its ability to manage metrics for local and
remote systems as well as being highly configurable.

Performance Co-Pilot (PCP) versus Sysstat

The sysstat tools (e.g., iostat and vmstat) are well know among Linux System
Administrators. PCP offers similar tools for a familiar experience.

Sysstat does not have a management service, relying on cron for repeating data collection,
limiting the collection intervals unless complex scripts are used while PCP operates as a systems
service, allowing for more flexibility for data collection and intervals.
REVIEWING METRICS AND LOGGING DATA USING PERFORMANCE COPILOT

Lesson Objectives (continued)

Lesson Notes (continued)


Performance Co-Pilot (PCP) Utility Comparisons

Common Cmd PCP Comments


iostat pcp iostat PCP output is similar to iostat -x
vmstat pcp vmstat
mpstat pcp mpstat
top pcp atop Not part of sys stat.

NOTE This is not an all inclusive list, just a few examples for comparison.
REVIEWING METRICS AND LOGGING DATA USING PERFORMANCE COPILOT

Command Examples from the Lesson

# View Performance Co-Pilot Configuration on the localhost:


pcp

# Compare iostat to pcp iostat and pmchart:


iostat -x
pcp iostat
pmchart -c Iostat

# Compare vmstat to pcp vmstat:


vmstat
pcp vmstat

# Compare mpstat to pcp mpstat:


mpstat
pcp mpstat
REVIEWING METRICS AND LOGGING DATA USING PERFORMANCE COPILOT

Command Examples from the Lesson

# Compare top to pcp atop:


top
pcp atop
-- Pressing 'h' will open the help menu, just like top and similar apps

# Monitor CPU load, memory usage, and disk write operations per partition:
pmdumptext -Xlimu -t 2sec 'kernel.all.load[1]' mem.util.used
disk.partitions.write -h localhost

# Review CPU load, memory usage, and disk write operations from an archive:
pmdumptext -Xlimu -t 2sec 'kernel.all.load[1]' mem.util.used
disk.partitions.write -a archive_name

# Check container running state:


pminfo --fetch containers.name containers.state.running containers.pid
REVIEWING METRICS AND LOGGING DATA USING PERFORMANCE COPILOT

Command Examples from the Lesson

# Check network interfaces:


pmprobe -I network.interface.up

# Check container interfaces:


pmprobe -I --container CONTAINERNAME network.interface.up

# View pmlogger directory:


ls /var/log/pcp/pmlogger/HOSTNAME

# View pmlogger file details by name:


pmdumplog -L /var/log/pcp/pmlogger/HOSTNAME/LOGNAME

# Extract all pmlogger archives to "hostarchive":


pmlogextract *.0.xz ../hostarchive
REVIEWING METRICS AND LOGGING DATA USING PERFORMANCE COPILOT

Command Examples from the Lesson

# Use pmchart to view extracted archive in graphical format:


pmchart -z -a hostarchive -t 10m -O-0 -s 400 -v 400 -geometry 800x450 -c CPU -c
Loadavg
-z = set reporting timezone to log source timezone
-t = time interval
-s = number of samples
-v = visible samples
-geometry = size of the pmchart window
-c = pmchart views to load

# Use prep to view extracted archive in text format:


pmrep -z -a hostarchive -t 10m -p -S'@15:45' -s 20 kernel.all.load
kernel.all.cpu.{user,sys,wait.total} disk.all.total_bytes
-p = show timestamps
-S = start time

You might also like