1 |
Am Mon, 20 Mar 2017 17:15:04 -0600 |
2 |
schrieb thelma@×××××××××××.com: |
3 |
|
4 |
> Besides standard "data" backup, if I was to plan for a disaster |
5 |
> recovery; what to include in a backup system if I was to rebuild a |
6 |
> new box? |
7 |
> |
8 |
> - /etc |
9 |
> - /var/lib/portage/world |
10 |
> - /usr/src/linux/.config |
11 |
> - /var/spool/fax/ (if needed) |
12 |
> - /var/www/localhost/htdocs/ (if needed) |
13 |
> - crontab (users and root) |
14 |
> |
15 |
> What else did I miss? |
16 |
|
17 |
I simply backup everything. Because whatever you try not to miss, |
18 |
there's always something you're gonna miss in the disaster case. |
19 |
|
20 |
A very good application to do this space efficient is borgbackup, plus |
21 |
it's very fast. |
22 |
|
23 |
Otherwise, I'd also add /boot to the list, to have a working kernel to |
24 |
start with. BTW: Having kernel modules, this is not enough: Also |
25 |
backup /lib/modules. |
26 |
|
27 |
If you store crontab, you should also store home. So your list is |
28 |
incomplete by definition. Why store only partly user data? |
29 |
|
30 |
Looking at "fax", this suggest there may be lots of other location you |
31 |
may want to backup. Why not the whole spool folder? It contains the |
32 |
mail spool. |
33 |
|
34 |
Backups of part of a system are not backups at all. By definition. Part |
35 |
of a disaster recovery plan is also to have a working rescue system |
36 |
with all the tools to restore your backup (backup software, network |
37 |
access, access to backup storage, encryption keys). |
38 |
|
39 |
For me, storing .config, /etc, and var/lib/world is enough and just a |
40 |
starting point for building new systems from scratch. But rebuilding a |
41 |
system takes much more. So I simply do not bother with rebuilding the |
42 |
packages and then trying to implant data back into such a system, I |
43 |
simply also store the system installation itself. The system is not |
44 |
that big that it outweights the hassle of rebuilding the system and |
45 |
fiddling around with putting data back into different places of the |
46 |
system. |
47 |
|
48 |
Restoring/rebuilding system has taught me that the hard way. There's |
49 |
always configuration outside of /etc. There's always something your |
50 |
"backup" missed, which is hard to rebuild. Simple solution: backup |
51 |
everything. |
52 |
|
53 |
And then, one of the most important factors of "disaster recovery" is |
54 |
time: In a disaster, it is important to get the system back online as |
55 |
fast as possible. You're going to be under stress anyways in that case, |
56 |
so take every chance for mistakes out of the process right from the |
57 |
beginning: Making restore more complex only introduces mistakes. Follow |
58 |
the KISS principle. |
59 |
|
60 |
-- |
61 |
Regards, |
62 |
Kai |
63 |
|
64 |
Replies to list-only preferred. |