Ubuntu 14.04 LTS VMs not accessible on 2nd day

Hello,

I've ran into this problem twice now and I don't know what I'm doing wrong so I figured I'd ask the community for help.  I'll try to detail all of my steps/setup below in an effort to get to the cause of my issue ASAP.  Thank you everyone in advance.

1. I created 3 Ubuntu 14.04 LTS VMs (Basic A3).  VMs do not have static public IP addresses and that is expected.
2. I open 2 additional ports for all incoming traffic (50070 & 8088) on 1 of the VMs.
3. I disable Ubuntu firewall and IPTABLES (just in case, I don't think Ubuntu has it configured by default).
4. Modify /etc/sysctl.conf to disable IPv6 and disk swapping on each VM.
5. Modify my /etc/hosts file on each VM so they can talk to each other using hostnames via internal IP addresses.
6. Create a new user group and a new user account that does not have sudo access.
7. Configure intra-VM passwordless SSH on each node.
8. Install some software from the internet (Oracle JDK & Apache Hadoop).
9. Make necessary environment variable, Hadoop config changes and verify that Apache Hadoop (HDFS & YARN) are functioning as expected across this 3 node cluster.
10. Shut down the Apache Hadoop services and then log off the SSH sessions I have to each VM.
11. Stop each VM using the Azure Portal stop button and see that they shut off.

The next day I take the following steps and this is where my issue is coming up:

1. Login to the Azure Portal.
2. Select the VM from the Startboard.
3. Start the VM.
4. Wait for the VM to become operational (OS startup).
5. Connect via SSH client.
6. Type in known admin username & password (same one that worked yesterday, recorded in a log and I verified this multiple time to make sure I wasn't mistyping something).
7. Receive "Access is denied" message as if I had mistyped the password.
   

This has happened twice (I tried deleting and starting over from scratch yesterday) now and I'm not sure why.  My expectation is that I can shutdown my services and my VMs when I'm not using them and then start them back up when I need them and be able to access via SSH using the known security credentials.  My SSH client did give me the expected "host key is not cached in the registry" message when connecting to the new dynamic public IP address that the VMs are assigned upon startup on day 2.  These are not issues - it is an issue that it seems the password is changed or account is disabled on second day after starting the VMs back up again.  I tested shutting down the VMs and then starting them back up on the same day and did not have this issue.  I also tested accessing via SSH with the user account I created and was successful however I then wasn't able to su to my admin account for admin purposes.

All configurations (firewall settings, IPv6, etc.) used are required by the software I intend to use on these VMs (Apache Hadoop and other ecosystem tools).

Thank you again for any help.

Orrin