1 |
On Thu, December 8, 2011 5:01 pm, Pandu Poluan wrote: |
2 |
> On Dec 8, 2011 9:46 PM, "Jarry" <mr.jarry@×××××.com> wrote: |
3 |
>> |
4 |
>> This server is ~50 miles away, and if I screw something |
5 |
>> and it does not boot up, I will have to go there and fix it |
6 |
>> on place. One small typo in ~50 config-files which must be |
7 |
>> updated is just enough to cause it... |
8 |
>> |
9 |
> |
10 |
> That's why I no longer deploy baremetal servers these days. Always as a |
11 |
> VM |
12 |
> on top of a hypervisor, with a small VM dedicated as an SSH tunnel. If I |
13 |
> mess up, I can use the hypervisor management tool to reboot the VM, or |
14 |
> open |
15 |
> a console session, or even revert to a "known good" state. |
16 |
|
17 |
A seperate machine with serial-console or Ethernet-KVM on the mainboard |
18 |
also helps. Especially when updating the host :) |
19 |
|
20 |
My new servers will all have Ethernet-KVM for this reason. |
21 |
|
22 |
-- |
23 |
Joost |