Differences between revisions 32 and 79 (spanning 47 versions)
Revision 32 as of 2017-11-28 16:52:01
Size: 1122
Editor: 192
Comment:
Revision 79 as of 2021-03-16 21:00:58
Size: 2639
Comment:
Deletions are marked like this. Additions are marked like this.
Line 1: Line 1:
= Its all about uptime! = = It is all about uptime! =
Line 3: Line 3:
The main goal is to go as close to no downtime as possible. System uptime and availability is being increasingly important. Various services are expected to be accessibly at all times. The challenge is to configure and maintain such a system, with minimal downtime, and preferable no downtime at all. At first this sounds like a very complicated task, but breaking it down to its components makes it possible. Configuring and deploying one component at a time, in a redundant way, allows for added complexity in a simple way, which increasingly builds on a stable redundant platform, keeping the redundancy all the way through the component stack. As the system is comprised of redundant components, it is ensured that any component at any time can be taken down for maintenance, without compromising the uptime and availabilty of the system.
Line 5: Line 5:
== These are our worst foes == == The Challenge ==

Many factors affects such a system, and usually we are not controlling all of them.
Line 12: Line 14:
To get around this a redundant system has to be built. Realizing that a redundant system consists of many redundant parts, virtualization naturally comes to mind. Redundancy for all factors will be very expensive, so here we will focus on the software redundancy, and the physical hardware the software runs on.

Realizing that a redundant system consists of many redundant parts, virtualization naturally comes to mind.
Line 28: Line 32:
== Network Planning ==

 * [[Network Planning|Network Planning]]
Line 30: Line 38:
 * [[DNS|DNS]]
Line 31: Line 40:
 * [[DNS|DNS]]
 
''This project is based on the GNU/Linux software packaged by Debian, currently the stretch release. The information provided here will most likely work on other GNU/Linux systems as well. Basic knowledge of GNU/Linux commandline tools, file editing etc. is required to use this guide.''
 * [[NTP|NTP]]
 * [[GlusterFS|GlusterFS]]
 * [[Corosync and HA-proxy|Corosync and HA-proxy]]
 * [[Mariadb|Mariadb]]
 * [[Redis|Redis]]
 * [[Apache|Apache]]
 * [[LDAP|LDAP]]
 * [[Postfix|Postfix]]
 * [[Dovecot|Dovecot]]
 * [[Nextcloud|Nextcloud]]
 * [[OPNsense|OPNsense]]

== Configure a Service ==

 * [[Syncthing|Syncthing]]
 * [[Minidlna|Minidlna]]
 * [[Checkmk|Checkmk]]

== Tools ==

=== Networking ===

 * [[nethogs|nethogs]]
 * [[bwm-ng|bwm-ng]]

''This project is based on the GNU/Linux software packaged by Debian. KVM hosts are running on Buster, guests are for the most part running on Buster. The information provided here will most likely work on other GNU/Linux systems as well. Basic knowledge of GNU/Linux commandline tools, file editing, network etc. is required to use and understand this guide.''

It is all about uptime!

System uptime and availability is being increasingly important. Various services are expected to be accessibly at all times. The challenge is to configure and maintain such a system, with minimal downtime, and preferable no downtime at all. At first this sounds like a very complicated task, but breaking it down to its components makes it possible. Configuring and deploying one component at a time, in a redundant way, allows for added complexity in a simple way, which increasingly builds on a stable redundant platform, keeping the redundancy all the way through the component stack. As the system is comprised of redundant components, it is ensured that any component at any time can be taken down for maintenance, without compromising the uptime and availabilty of the system.

The Challenge

Many factors affects such a system, and usually we are not controlling all of them.

  • Power outage
  • Internet access
  • Hardware malfunction
  • Software updates

Redundancy for all factors will be very expensive, so here we will focus on the software redundancy, and the physical hardware the software runs on.

Realizing that a redundant system consists of many redundant parts, virtualization naturally comes to mind.

Configure your system

Manage Virtual Domains

Network Planning

Configure a Redundant Service

Configure a Service

Tools

Networking

This project is based on the GNU/Linux software packaged by Debian. KVM hosts are running on Buster, guests are for the most part running on Buster. The information provided here will most likely work on other GNU/Linux systems as well. Basic knowledge of GNU/Linux commandline tools, file editing, network etc. is required to use and understand this guide.

TODO

None: Uptime (last edited 2021-12-31 11:46:57 by Kristian Kallenberg)