1 |
Maybe you can just use two A records in your dns. |
2 |
|
3 |
Both point to one server, and both servers keep an eye on the other with |
4 |
hearthbeat, |
5 |
|
6 |
As soon as one of the servers gets unresponsive the other claims the ip |
7 |
as a secundary ip. |
8 |
|
9 |
ifconfig eth0 add otherip. |
10 |
// not forget to arp for it! |
11 |
|
12 |
And as soon as the hearthbeat finds the other server to be back online, |
13 |
it unsets the ip, and lets the other know that it can add the ip again. |
14 |
|
15 |
Not so hard, just a couple of shell scripts afaik. |
16 |
|
17 |
greets |
18 |
Jan |
19 |
|
20 |
|
21 |
Benjamin MALYNOVYTCH wrote: |
22 |
|
23 |
>********************************** |
24 |
>MB> You're faced with an impossible requirement here: |
25 |
> |
26 |
>MB> If, as you've stated "one machine is not enough due to high trafic" a |
27 |
>MB> single machine left over after a filure WON'T be able to keep you going. |
28 |
> |
29 |
>MB> For a high availability setup, you need a configuration where the |
30 |
>MB> remaining node(s) after failure are still powerfull enough to carry on. |
31 |
> |
32 |
>MB> Bye, Martin |
33 |
>********************************** |
34 |
> |
35 |
>Well, one machine is enough most of the time. It only slows down |
36 |
>sometimes, due to high trafic. That's why this cluster would be fine |
37 |
>enough, letting us enough time to restart the failling machine. |
38 |
> |
39 |
>Is this configuration still possible ? |
40 |
> |
41 |
>Best regards. |
42 |
> |
43 |
>Benjamin. |
44 |
> |
45 |
> |
46 |
> |
47 |
|
48 |
-- |
49 |
gentoo-cluster@g.o mailing list |