Methods for automatic installation of servers and virtulaolk?

Question

linux

What methods, excluding puppet, ansible etc, can be used to automatically install a preconfigured machine

, so I only know this method
1. PXE + DHCP + tftp + kickstart file (this method seems very difficult to use when you have 10,000 thousand machines per hour rising , well, or how many AWS have there)


And how do eminent providers, such as DigitalOcean, Amazon, and any others that deploy much faster than the above method.

Answer 1

Just deploy from the finished image. You will have a virtual one in a few seconds.

Answer 2

Using an API (for example for Amazon EC2 docs.aws.amazon.com/AWSEC2/latest/APIReference/Wel... you create machines and insist them via ansible. The
question is too general and too simple, please clarify what you mean.

Answer 3

Well, I'll sign off.
If we need to install an iron server. Then we load the prepared special image (about 30 megabytes) via PXE + dhcp.
After loading, an agent is launched on the system, which collects characteristics from this server and sends them to the central server, and in the opposite direction takes the file from the installation script for this server.
It downloads a real image according to the script, and in 2-3 minutes it formats disks and deploys the system, configures it and restarts. Everything, the server with the necessary characteristics is ready.
This is how a lot of povijin projects work, for example FAI - fai-project.org
The second option, when you have centralized disk storage, for example via FibreChannel, you do not need to download anything over the network at all, just say at the storage level which image should be available for download to which FC adapter. Same if you have iSER or iSCSI storage.
The third option, when you have virtual machines, clone the desired image in the file system (local or distributed) and voila. For example, you have a CEPH distributed storage, an RBD image is cloned in 2-3 seconds, plus a template for a new virtual machine. Well, automation for the deployment of virtual machines, chef, ansible, puppet, then everywhere.

Answer 4

Docker, Vagrant на гугле

Answer 5

Хостинг-провайдеры разворачивают виртуалки из преднастроенных готовых образов. Берётся виртуалка, ставится внутрь система, обновляется, настраивается, и т.п. Разделы внутри виртуалки автоматически ресайзятся при первом запуске развёрнутой из образа виртуалки до размера всего диска (неважно, сколько там выделено, разделы займут всё дисковое пространство). Управление IP-адресами осуществляется с помощью различных система IPAM (IP address management). У VMware и Hyper-V есть механизмы IP pool, когда внутрь виртуалок прописывается статический адрес из определённого преконфигурированного IP-пула. На KVM/Xen, думаю, тоже такие решения есть. Как бы, это никакая не проблема -- после клонирования базового образа подмонтировать раздел нового виртуального диска и там в /etc/sysconfig/network-scripts/ifcfg-eth0 прописать IP-адрес для адаптера, это можно даже скриптами на bash накостылить самому.
Самый простой вариант выдачи адреса, понятно -- это DHCP.