[BlueOnyx:00917] Re: Tar.gz installer on Amazon Cloud EC2 + eth0 settings

[Michael Stauber]

Hi Julien,

> Infact it should be nice to install everything and do not launch new
> services (unless confirmed with a prompt) so that SysAd can check network
> settings..
>
> curl: (7) Failed to connect to 169.254.169.254: Network is unreachable


169.254.169.254? Is that an IP fetched from DHCP? Also make sure your 
/etc/sysconfig/network has this line in it:

NOZEROCONF=yes

> Do you know if there is a way to have everything installed without new
> services to be started ? should I comment the lines:
>
> ## post scripts
> /usr/sausalito/constructor/base/wizard/\:sysreset\:linkToWizard.pl
>
> ?

*That* script has nothing to do with restarting the network. You can leave 
that in. It just creates the static HTML pages that bring up the post install 
Wizard.

In fact nothing in the install.sh script modifies or restarts the Network. 

However, during the install of the base-network RPMs the existing network 
information is read from the system and stored into CODB. And exactly during 
that first storage of the network config into CODB it will rewrite the network 
configuration, writing back the previously gathered settings. 

It happens automatically during the CCEd restart, which is trigged by the RPM 
installation.  Not much you or we can or should do about that.

Typically: If your network config was OK before, it should be fine afterwards, 
too. It doesn't come up with it's own network settings out of thin air or 
makes something up.

Best procedure (in your case) I'd say is this:

Start over, double check your network settings first. Make sure that 
/etc/sysconfig/network has the line ....

NOZEROCONF=yes

... in it. Use static IPs, too. Maybe put in a DHCP failback if you like that 
kicks in after 10-15 minutes of not aborting it manually. Or a script that 
copies a backup of your network settings back after 10-15 minutes of 
inactivity and then restarts the network.

-- 
With best regards

Michael Stauber