User:Jukeboksi/BBA studies/User:Jukeboksi/Blog/2013: Difference between revisions
< User:Jukeboksi | BBA studies | User:Jukeboksi
(tweaking the HOWTO on linux backupping) |
(mentioning mysqldump and how to use it) |
||
| Line 8: | Line 8: | ||
::* ''Usually the full home-directory and a full dump of [[w:database]]s is all you need'' | ::* ''Usually the full home-directory and a full dump of [[w:database]]s is all you need'' | ||
::* ''Use a dedicated backup directory on every machine. '''The main point is that it needs to be somewhere outside the directories you are interested in backupping.''' I call mine <tt>/backupmedia</tt> . | ::* ''Use a dedicated backup directory on every machine. '''The main point is that it needs to be somewhere outside the directories you are interested in backupping.''' I call mine <tt>/backupmedia</tt> . | ||
:::<tt>'tar cvzf /backupmedia/machinename/homedir.tar.gz /home/yourusername' </tt> Will compress your home directory | :::<tt>'tar cvzf /backupmedia/machinename/homedir.tar.gz /home/yourusername' </tt> Will compress your home directory. | ||
:::Use the <tt>'mysqldump'</tt> command with switches <tt>--all-databases</tt>, the mysql root password and user | |||
# ) VERIFY THE BACKUPS TO WORK BY LOADING INTO A REFERENCE SYSTEM ON NON-PRODUCTION CLOSED SERVER INSTANCE. Not verified truly cannot be called backup because if they are found to be faulty shit will ensue. Check that the .tar.gz's unpack and that the database dumps can be read onto MySQL. | # ) VERIFY THE BACKUPS TO WORK BY LOADING INTO A REFERENCE SYSTEM ON NON-PRODUCTION CLOSED SERVER INSTANCE. Not verified truly cannot be called backup because if they are found to be faulty shit will ensue. Check that the .tar.gz's unpack and that the database dumps can be read onto MySQL. | ||
#) Criss-cross-copy from every machine to every machine / distribute the backups to as many servers, laptops and USB sticks as possible. | #) Criss-cross-copy from every machine to every machine / distribute the backups to as many servers, laptops and USB sticks as possible. | ||
Revision as of 11:59, 2 September 2013
September
2013-09-02
Full database w:backup performed
2013-09-02 Howto make cheap and easy backups on Linux?
To properly do a backup one must:
- ) Isolate the information objects that need to be successfully backupped and do it
- Usually the full home-directory and a full dump of w:databases is all you need
- Use a dedicated backup directory on every machine. The main point is that it needs to be somewhere outside the directories you are interested in backupping. I call mine /backupmedia .
- 'tar cvzf /backupmedia/machinename/homedir.tar.gz /home/yourusername' Will compress your home directory.
- Use the 'mysqldump' command with switches --all-databases, the mysql root password and user
- ) VERIFY THE BACKUPS TO WORK BY LOADING INTO A REFERENCE SYSTEM ON NON-PRODUCTION CLOSED SERVER INSTANCE. Not verified truly cannot be called backup because if they are found to be faulty shit will ensue. Check that the .tar.gz's unpack and that the database dumps can be read onto MySQL.
- ) Criss-cross-copy from every machine to every machine / distribute the backups to as many servers, laptops and USB sticks as possible.
- One can then just substitute /media/usb0 to make the off-line copies with command 'sudo cp -r /backupmedia/* /media/usb0/.' or 'cp -rf /backupmedia/* /media/usb0/.' if you don't need to force using sudo rights which I do for some reason on the w:USB stick. Be adviced that a normal USB key will accept any sudo password on any machine afaik.