| 1 |
On Thu, Feb 25, 2010 at 17:11, Ward Poelmans <wpoely86@×××××.com> wrote: |
| 2 |
> On Thu, Feb 25, 2010 at 16:41, Alan McKinnon <alan.mckinnon@×××××.com> wrote: |
| 3 |
>> A much better way is to run a dedicated agent on the client. If the server |
| 4 |
>> needs to schedule backups, it can ask the agent to do so using regular tcp |
| 5 |
>> traffic. The client can then do it's backup and rsync it over to the server |
| 6 |
>> when it's done, and that push can be done as a regular user on both ends. The |
| 7 |
>> actual backing up on the client must be done by root of course, no other user |
| 8 |
>> has the necessary access. |
| 9 |
|
| 10 |
If anyone is still interested, i had some time and this is what i did: |
| 11 |
|
| 12 |
On the client: |
| 13 |
rsync -a -X -b --backup-dir=../backup.0/ --link-dest=../backup.0/ |
| 14 |
/home/ward backupserver:Backup-Laptop/backup.cur/ |
| 15 |
ssh backupserver /home/ward/shiftbackups.sh |
| 16 |
|
| 17 |
This makes a directory backup.cur on the backupserver with a full |
| 18 |
backup in it, but it's exactly only a incremental backup as it |
| 19 |
hardlinks from backup.0 (the previous backup). The script |
| 20 |
shiftbackups.sh moves backup.0 to backup.1 and backup.cur to backup.0 |
| 21 |
and so on... |
| 22 |
|
| 23 |
This does more or less exactly what i wanted. |
| 24 |
|
| 25 |
Regards, |
| 26 |
|
| 27 |
Ward |
Archives