On May 21, Thomas Föcking
Once a week this simply script is ran: (only this script has cap_sys_rawio and write access on /dev/sdb) dd if=/dev/sda of=/dev/sdb bs=16M You will have two problems with this: Minor: Only data lost during the last week can be restored. If you delete a file, go on holiday for two weeks and want to restore it later, there is no chance. Major: Changes in the filesystem during this "backup" will most likely ruin your copy ... !
You also didn't tell us, which services you are offering. Web only? User authentication? SMTP/POP/IMAP? ... Every service that is used to authenticate users should use encryption (SSL/TLS). World: ftp, ssh, http, https Anonymous ftp only, hopefully?
Markus -- _____________________________ /"\ Markus Gaugusch ICQ 11374583 \ / ASCII Ribbon Campaign markus@gaugusch.at X Against HTML Mail / \
* Markus Gaugusch wrote on Tue, May 21, 2002 at 17:08 +0200:
dd if=/dev/sda of=/dev/sdb bs=16M You will have two problems with this: Minor: Only data lost during the last week can be restored. If you delete a file, go on holiday for two weeks and want to restore it later, there is no chance. Major: Changes in the filesystem during this "backup" will most likely ruin your copy ... !
Additionally, it's difficult to restore single files. And even if you do a file based backup, i.e. with tar, from a rw mounted filesystem, chances are high that at least a few files or directories are inkonsistent (i.e. a file is TRUNC, with contents in a temp file, tar may backup the zero file and such races). But with LVM, there is the very nice snapshooting feature. I think for servers it is the onliest way to have consistent backups without outages. oki, Steffen -- Dieses Schreiben wurde maschinell erstellt, es trägt daher weder Unterschrift noch Siegel.
dd if=/dev/sda of=/dev/sdb bs=16M
You will have two problems with this: Minor: Only data lost during the last week can be restored. If you delete a file, go on holiday for two weeks and want to restore it later, there is no chance.
This is not what I want. If sda crashes I put sdb into the sda case and a new sd into the second hdb case. hdb has the same (max. 1 week old data) like hda. The html files and mysql databases are automaticly copied (scp) from our "intranet" server with content managemant system to the internet server. So there is no need so make a backup of them... (The intranet-server has raid1 and a seperated backup system)
Major: Changes in the filesystem during this "backup" will most likely ruin your copy ... !
I didn't mentioned that swap and other partionions are first unmounted or mounted readonly. The backup of 17gb take 15 minutes at a time where no one will work on the system. And if so, s/he won't be able to ;-) regards, Thomas
participants (3)
-
Markus Gaugusch -
Steffen Dettmer -
Thomas Föcking