lost ssh access while running the two user two experiment scenario IG-EXP-2

[lnevers@bbn.com]

While executing IG-EXP-2 an experiment that has two concurrent slivers by two different users, lost ssh access to the nodes for one of the experiments.

Sequence of events that lead to the problem:

As user ​lnevers@bbn.com executed the following:

1. Create slice named "2vm"
2. Created sliver with 2 OpenVZ nodes using default image.
3. Logged into each of the assigned hosts, both on pc5, port 30011 and port 30010.
4. Ran some commands and left a ping running while executing 5,6, and 7 below.

As user ​lnever1@bbn.com executed the following:

5. Created a slice "lnubuntu12b"
6. Created a sliver with 2 ubuntu 12 custom images
7. Logged into each of the assigned hosts (pc1 and pc4) and ran some command

As user ​lnevers@bbn.com executed the following:

8. Exited the two ssh sessions from 3 above, to log back in and verify that key were still in place.
9. Tried to ssh back in, but connection is refused by the assigned hosts pc5, port 30011 and port 30010.

   lnevers@arendia:~$ ssh pc5.utah.geniracks.net -p 30010
   ssh: connect to host pc5.utah.geniracks.net port 30010: Connection refused
   lnevers@arendia:~$ ssh pc5.utah.geniracks.net -p 30011
   ssh: connect to host pc5.utah.geniracks.net port 30011: Connection refused
   

The user ​lnevers@bbn.com can no longer login into the active sliver nodes. Have the ​lnevers@bbn.com keys been overwritten?

Both experiment are left running.

[lnevers@bbn.com]

Attaching sliverstatus for the two experiments.

[lnevers@bbn.com]

I think the slice name existed from previous testing and expired less than two hours after I started the sliver, which I had not expected. I will re-ran the test just to make sure that I can re-execute the entire IG-EXP-2 procedure and close this ticket.

On 5/25/12 1:11 PM, Leigh Stoller wrote:

�1. Create slice named "2vm" �5. Created a slice "lnubuntu12b"

�Both experiment are left running.

I see the sliver lnubuntu12b, but 2vm is gone so I cannot debug it. Perhaps the slice expired and that is why you could get back to it?

Lbs

On 5/25/12 1:18 PM, Leigh Stoller wrote:

I see the sliver lnubuntu12b, but 2vm is gone so I cannot debug it. Perhaps the slice expired and that is why you could get back to it?

To followup, the log shows that 2vm expired and was terminated by the system at 10:08 (MDT).

Be sure to use renew to keep your slivers around longer, or whatever Omni provides for this purpose.

Lbs

[lnevers@bbn.com]

This scenario has been executed without problems, able to run concurrent experiments as two different users and successfully exchange traffic between endpoint. Closing ticket.