How to create and move (clone) backup of the system drive from one linux VPS to another?

Greetings to all.
On the server system disk partition is online store of Bitrix with a large attendance. There was a question about backup cloning your system drive from one linux VPS to another and then updating the system on the server the donor (php, mysql, web environment the Bitrix). Server use from the company TimeWeb, which has no functions of the VDS cloning, boot from the live CD boot backup drive.
What is the best and with minimal downtime for the site to create and move (clone) backup of the system drive from one linux VPS to another?
June 3rd 19 at 20:59
6 answers
June 3rd 19 at 21:01
Thank God this is Linux and you can just rsync without system files and everything will be exactly
June 3rd 19 at 21:03
dump/restore
clumsy method - tar everything from the root. Stupidly, voraciously, a lot of excess, but certainly did not miss...
Tried so
tar-cvfp /mnt/backup/archive.tar.gz /dev/sda3, swears removing leading '/' from names /dev/sda3
tried so tar-cvfp /mnt/backup/archive.tar.gz dev/sda3, it complains cannot stat: no such file or directory
Tell me how - patience.Fisher commented on June 3rd 19 at 21:06
,
dev/sda3 slash at the beginning of the missed /dev/sda3 - Jaqueline.Parisian commented on June 3rd 19 at 21:09
tar always removes the first slash, if it is in the way. For what? And while unpacking came out so that the unpacked files will overwrite the root filesystem. So the message removing the leading / is a perfectly normal message - Esta_Upton commented on June 3rd 19 at 21:12
,
Samplessshaft

and I had samplesofVAT. - Brenden_Terry commented on June 3rd 19 at 21:15
come on, grammar Nazis... Maybe cilavegna Russian Radney.... - Esta_Upton commented on June 3rd 19 at 21:18
June 3rd 19 at 21:05
Easier and cheaper to contact support TimeWeb.
However you can run into the typical big companies clumsiness. Bureaucracy, regulations, etc.
Turned, can't do anything - patience.Fisher commented on June 3rd 19 at 21:08
, backup site and do the update. Recently updated environment, everything went fine.
p.s. We have an environment without dirty hacks. Certbot and ProFTPD from a third party. - Jaqueline.Parisian commented on June 3rd 19 at 21:11
In the past, developers have gone and left me this farm. Also installed on the server ssl(https) certificate which can fly after the upgrade. What version of entourage You were? - Esta_Upton commented on June 3rd 19 at 21:14
, https://timeweb.com/ru/support/faq/vds/common/kak-... - Brenden_Terry commented on June 3rd 19 at 21:17
, Probably 7.2. Half a year ago that was relevant.
Will fly better, nothing to worry about. We didn't fly. LE certificates issued means certbot --nginx - Esta_Upton commented on June 3rd 19 at 21:20
June 3rd 19 at 21:07
https://toster.ru/answer?answer_id=426823
Duplicates,
To choose the right strategy the migration of the site to another server, you first need to decide whether otlozhenie of writes/updates to the database during the transfer of the resource, i.e. the actual simple? Suppose that we started the transfer of the loaded online store, where every three minutes the order. If you stupidly remove the database dump, make an archive of files, overtake on another server, deploy the whole thing, the lag time will be a minimum of 30 minutes, even if that new piece of hardware will be in the same rack. While there will be a migration, the old database will already recorded new orders. As a variant it is possible during migration to prevent the creation of orders and users, but a simple 30 minutes (actually 2 hours) is not satisfied, so on large projects practiced a seamless move - first, install the balancer, it is overwritten DNS, and he, in turn, proxies the traffic to the old server, then raised a second server, it unfolds the database, which poluchaetsya to the database on the old server as slave, the sync scripts rsync/csync, the files are submitted to the cloud. Finally, we obtain the classic master-slave model. In conclusion, changing the roles in the database, make a new master and old slave. Simmer the old server.
This is some wild overhead. You can migrate with minimal downtime. You do not serve the NPP. - patience.Fisher commented on June 3rd 19 at 21:10
depends on what will outweigh: the value of orders in two hours of downtime vs. 16 hours of devops - Jaqueline.Parisian commented on June 3rd 19 at 21:13
if everything is so expensive, then the failover should have thought originally. - Esta_Upton commented on June 3rd 19 at 21:16
June 3rd 19 at 21:09
I understand there is a need to boot from the live cd, but this is not possible. Thanks for the advice - patience.Fisher commented on June 3rd 19 at 21:12
June 3rd 19 at 21:11
In General 2 options:
1. Rent reserve VDS. Becaem tarом disk, ssh move to a new server, restored. Custom domain on the new server. When the move will take place to synchronize the database using SQLyog. A simple site about 30 min. On the old server all updated, tested. Repeat the move. Costs: renting a new server + additional disks (for a month). The risk is minimal
2. Becaem means of the hosting disk. Rolling updates, testing. With the appearance of critical issues are rolled back. Simple problems with more than 1 variant and the risk of more

Soon I will write what happened...

Find more questions by tags 1C-BitrixLinux