1 |
On Thu, Mar 28, 2013 at 11:38 AM, Nick Khamis <symack@×××××.com> wrote: |
2 |
|
3 |
> Hello Everyone, |
4 |
> |
5 |
> Just got a ticket assigned to me where we need to update our production |
6 |
> servers. |
7 |
> |
8 |
> uname -a |
9 |
> Linux noun 3.4.9-gentoo #2 SMP Sat Oct 13 09:35:07 EDT 2012 x86_64 |
10 |
> Intel(R) Xeon(TM) CPU 3.60GHz GenuineIntel GNU/Linux |
11 |
> |
12 |
> eselect |
13 |
> [18] hardened/linux/amd64 * |
14 |
> |
15 |
> I don't think they have been updated since the initial install and |
16 |
> wanted to get a little feedback on some safe practices and methods |
17 |
> that should be performed before and while doing so. |
18 |
> |
19 |
> Thanks in Advance, |
20 |
> |
21 |
> Nick. |
22 |
> |
23 |
> |
24 |
Personally, I would recommend pulling an rsync (databases and such might |
25 |
cause a hiccup with that) of one of them to a nonessential system and |
26 |
testing updating there, building packages (assuming matching use flags, |
27 |
etc, across your systems), documenting the pitfalls you run into as you go. |
28 |
After you're up to date there, run through and test it again from a base |
29 |
copy, then test the actual services to ensure changes to them don't hose |
30 |
your environment's configuration, and once that's good, it then depends |
31 |
entirely on what failover, or downtime allowances you have available. If |
32 |
you have no failover to rely on, and can't afford enough downtime to update |
33 |
the system in place from the packages you've built, clone each off, update, |
34 |
then migrate the changes that've occured in the time between... time |
35 |
consuming, and requires a lot of care, but doable. |
36 |
|
37 |
-- |
38 |
Poison [BLX] |
39 |
Joshua M. Murphy |