Setting up PCP – Performance Co-Pilot
by Uday R. Sawant, William Leemans, Jonathan Hobson, Oliver Pelz
Linux: Powerful Server Administration
- Linux: Powerful Server Administration
- Table of Contents
- Linux: Powerful Server Administration
- Linux: Powerful Server Administration
- Credits
- Preface
- 1. Module 1
- 1. Managing Users and Groups
- 2. Networking
- Introduction
- Connecting to a network with a static IP
- Installing the DHCP server
- Installing the DNS server
- Hiding behind the proxy with squid
- Being on time with NTP
- Discussing load balancing with HAProxy
- Tuning the TCP stack
- Troubleshooting network connectivity
- Securing remote access with OpenVPN
- Securing a network with uncomplicated firewall
- Securing against brute force attacks
- Discussing Ubuntu security best practices
- 3. Working with Web Servers
- Introduction
- Installing and configuring the Apache web server
- Serving dynamic contents with PHP
- Hosting multiple websites with a virtual domain
- Securing web traffic with HTTPS
- Installing Nginx with PHP_FPM
- Setting Nginx as a reverse proxy
- Load balancing with Nginx
- Setting HTTPs on Nginx
- Benchmarking and performance tuning of Apache
- Securing the web server
- Troubleshooting the web server
- 4. Working with Mail Servers
- 5. Handling Databases
- Introduction
- Installing relational databases with MySQL
- Storing and retrieving data with MySQL
- Importing and exporting bulk data
- Adding users and assigning access rights
- Installing web access for MySQL
- Setting backups
- Optimizing MySQL performance – queries
- Optimizing MySQL performance – configuration
- Creating MySQL replicas for scaling and high availability
- Troubleshooting MySQL
- Installing MongoDB
- Storing and retrieving data with MongoDB
- 6. Network Storage
- 7. Cloud Computing
- 8. Working with Containers
- Introduction
- Installing LXD, the Linux container daemon
- Deploying your first container with LXD
- Managing LXD containers
- Managing LXD containers – advanced options
- Setting resource limits on LXD containers
- Networking with LXD
- Installing Docker
- Starting and managing Docker containers
- Creating images with a Dockerfile
- Understanding Docker volumes
- Deploying WordPress using a Docker network
- Monitoring Docker containers
- Securing Docker containers
- 9. Streaming with Ampache
- 10. Communication Server with XMPP
- 11. Git Hosting
- Introduction
- Installing Git
- Creating a local repository with Git CLI
- Storing file revisions with Git commit
- Synchronizing the repository with a remote server
- Receiving updates with Git pull
- Creating repository clones
- Installing GitLab, your own Git hosting
- Adding users to the GitLab server
- Creating a repository with GitLab
- Automating common tasks with Git hooks
- 12. Collaboration Tools
- 13. Performance Monitoring
- 14. Centralized Authentication Service
- 2. Module 2
- 1. Installing CentOS
- Introduction
- Downloading CentOS and confirming the checksum on Windows or OS X
- Creating USB installation media on Windows or OS X
- Performing an installation of CentOS using the graphical installer
- Running a netinstall over HTTP
- Installing CentOS 7 using a kickstart file
- Getting started and customising the boot loader
- Troubleshooting the system in rescue mode
- Updating the installation and enhancing the minimal install with additional administration and development tools
- 2. Configuring the System
- Introduction
- Navigating text files with less
- Introduction to Vim
- Speaking the right language
- Synchronizing the system clock with NTP and the chrony suite
- Setting your hostname and resolving the network
- Building a static network connection
- Becoming a superuser
- Customizing your system banners and messages
- Priming the kernel
- 3. Managing the System
- Introduction
- Knowing and managing your background services
- Troubleshooting background services
- Tracking system resources with journald
- Configuring journald to make it persistent
- Managing users and their groups
- Scheduling tasks with cron
- Synchronizing files and doing more with rsync
- Maintaining backups and taking snapshots
- Monitoring important server infrastructure
- Taking control with GIT and Subversion
- 4. Managing Packages with YUM
- 5. Administering the Filesystem
- 6. Providing Security
- 7. Building a Network
- 8. Working with FTP
- 9. Working with Domains
- 10. Working with Databases
- 11. Providing Mail Services
- 12. Providing Web Services
- 13. Operating System-Level Virtualization
- 14. Working with SELinux
- 15. Monitoring IT Infrastructure
- 1. Installing CentOS
- 3. Module 3
- 1. Working with KVM Guests
- 2. Deploying RHEL "En Masse"
- 3. Configuring Your Network
- 4. Configuring Your New System
- 5. Using SELinux
- 6. Orchestrating with Ansible
- 7. Puppet Configuration Management
- 8. Yum and Repositories
- 9. Securing RHEL 7
- 10. Monitoring and Performance Tuning
- Bibliography
- Index
Over the years, a lot of tools have been created to troubleshoot performance issues on your systems, such as top, sar, iotop, iostat, iftop, vmstat, dstat, and others. However, none of these integrate with each other, some are extensions to others, and so on.
PCP seems to have a couple of things right: it monitors just about every aspect of your system, it allows the centralized storage of (important) performance data, and it allows you to use not only live data, but also saved data among others.
In this recipe, we'll look at both the "default" setup and "collector" configuration, which allows you to pull in all the performance data you want.
This is the basic setup of PCP:
- Let's install the necessary packages; run the following command:
~]# yum install -y pcp - Now, enable and start the necessary daemons, as follows:
~]# systemctl enable pmcd ~]# systemctl enable pmlogger ~]# systemctl start pmcd ~]# systemctl start pmlogger
- If you want to have the system monitored by a central collector, execute the following:
~]# firewall-cmd --add-service pmcd --permanent
Each host that is to act as a collector requires additional configuration. Here's how you can do this:
- Add a line per system to collect data from
/etc/pcp/pmlogger/control, as follows:<hostname> n n PCP_LOG_DIR/pmlogger/<hostname> -r -T24h10m -c config.<hostname>
Here,
<hostname>is the FDQN to this host. Take a look at the following example:guest.example.com n n PCP_LOG_DIR/pmlogger/guest.example.com -r -T24h10m -c config.guest.example.com
- After adding a host in this way, you need to restart the
pmloggerdaemon. Execute the following command line:~]# systemctl restart pmlogger
By default, PCP logs information every 60 seconds. If you want to increase this and want to gather performance statistics every 30 seconds, you need to change the line starting with LOCALHOSTNAME and add -t 30s at the end.
Modifying the statistics you gather is a bit more difficult. You can find the configuration for pmlogger in /var/lib/pcp/config/pmlogconf/. Every file in this directory contains information about which pointers to gather. The syntax is not very hard to understand, but it is complex to explain. The pmlogconf (1) man page contains everything you need to know.
If you want to visualize the data on a host, you need to install pcp-gui, as follows:
~]# yum install -y pcp-gui dejavu-sans-fonts
This package comes with a tool called pmchart, which allows you to create graphics with the data collected by PCP. The fonts are needed to properly display the characters.
For more information about PCP and its components, refer to their online manuals, which you can find at http://www.pcp.io/documentation.html.
-
No Comment