Admin password on ebs root-vol-from-snap

[Brian Hendrickson]

Hi for-poets fans,

I made a snapshot of my server, but when I tried to roll back (see
method below) - my Windows Administrator password is now invalid. I re-
fetched the password with my .pem key from the aws console but that
didn't help.

This is how I rolled back:
1) stopped the instance 2) detached orig vol 3) attached vol-from-snap
at "/dev/sda1" 4) started instance - the snapshot was made while the
Windows instance was running - could that have caused the problem?

Thanks for any ideas :-)

-Brian

[Brian Hendrickson]

I found the cause of my Windows Admin password problem - the Public DNS address of the instance changed! I didn't know that could happen

[Dave Winer]

The ip address is not preserved across stops and starts.

[Brian Hendrickson]

Thanks Dave - yeah I guess I need an elastic IP

On Apr 21, 4:17 pm, Dave Winer <dave.wi...@gmail.com> wrote:
> The ip address is not preserved across stops and starts.
>
>
>
> On Saturday, April 21, 2012, Brian Hendrickson wrote:
> > I found the cause of my Windows Admin password problem - the Public DNS
> > address of the instance changed! I didn't know that could happen
>
> > On Thu, Apr 19, 2012 at 8:47 PM, Brian Hendrickson <

> > openmicroblog...@gmail.com <javascript:_e({}, 'cvml',

[Dave Winer]

I just use DNS.

[Andrew Grumet]

 

Guessing that Route 53 knows when it is pointing to an EC2 instance and updates automatically.  Brian, if you're using something besides Route 53, an elastic IP should do the trick.  The elastic IP is free as long as it's in use.

[Dave Winer]

Pretty sure Route 53 doesn't do that, but it would be easy to check.

I'm a pretty heavy user of that service. :-)

Dave

[Frank McPherson]

What I have learned is that when you stop an EC2 instance, the next time you start it up the instance will likely have a new IP address and thus a new public DNS name provided by Amazon. If you have a CNAME pointing to Amazon's public DNS, you will need to edit the CNAME so that it points to the new public DNS. 

I did not want to edit my CNAME every time I restarted the EC2 instance, so I use an Elastic IP. With an Elastic IP, every time you associate that IP to an instance Amazon recreates the same public DNS, which is what I have the CNAME pointing to. Consequently, whenever I want to restart my EC2 instance I do the following:

1. Launch the instance in EC2

2. Associate my elastic IP to that instance, then wait for a few minutes for the public DNS to propagate

3. Connect to my instance via RDP using the CNAME and verify that the OPML editor is running

From that point forward I can then log in to Radio and my River refreshes.  Of course, all of this is moot if you never stop or terminate the instance.

[Andrew Grumet]

I haven't used the service before, but this piqued my interest.  I tried assigning a CNAME to the public DNS record for a running instance, then stopping the instance, then starting again and seeing a new DNS name as expected.  Switching over to the Route 53 tab, I could see that the CNAME still pointed to the old DNS name.  This was not a big surprise, per Dave's comments, and also because I couldn't find anything in the docs to indicate support for this.

 

Interestingly, they do have an "alias" record type that can be used in a similar way with elastic load balancers, but there's no way to set the target to an individual instance.