How do we arrange a site shift to a backup server?
We have a main site, a large portal with thousands of pages, personal offices that use thousands of people, with a lot of information for clients, an internal CRM system for our staff, and dozens of postal accounts. It's all working on the same house and gathered together - i.e. a monolithic system, redefining something as it says.
Most recently, there was an accident at our client's house, and our VPS'ca died for a long time.
Naturally, we've developed the main server system. They went into the dominator control panel and changed the NS server.
In a few hours, the site earned on the standby suite and and everything was fine!
Question is: How can the system be implemented in such a way that we can quickly and expeditiously switch our dominance from one hosting to another (from the main server to standby)? The option of a backup server at the same hosting service provider is not a very good option for us - once we've already contacted one company, and in the end, the two servers weren't working.
The records of the housekeepers from the base are used not immediately but usually on schedule. Plus, dns serves the NS addresses (in fact, like all the other addresses). This is a major delay in changing NS servers.
Take out the NS server with the other services. There's a lot of DNS support where they offer and even free. The same house registrars offer services (usually paid). If you have your own backup server on the Internet, you can also raise a full DNS that's going to be responding to the ip of the work server. In the event of an emergency situation, it is necessary to change the records not from the registrar (NS) but in its DNS, to change the main A records. The maximum switching time will be the TTL.
It is only necessary to repair the warning of master DNS of slays and the operational fence of the zone. Either you lift all the DNS in the master's mode and change the recordings in each individual accident. The truth is, a few masters are making it difficult to support because all changes in the zone must always be done with their hands on all servers.
Option number 2 (for web):
On main and stand-by servers, contact content-servers to organize through proxy (nginx), with proxy on the server reserve redirecting queries to the main content server. In an emergency, it's switching to its own content server. Proxy can be raised again.