Opened 9 years ago

Last modified 9 years ago

#100 new

Resource mapper bug when clientId is the same as the name of a physical host

Reported by: nriga@bbn.com Owned by: somebody
Priority: major Milestone:
Component: AM Version: SPIRAL5
Keywords: Cc:
Dependencies:

Description (last modified by nriga@bbn.com)

Hi I created a slice (transgeo) with two raw nodes and two vms. The topology should look like: pc1-pc2-vm1-vm2

Notice that pc1 and pc2 are my clientIds and not the actual pc names (I know very unfortunate naming)

pc1 was mapped to pc3.instageni.gpolab.bbn.com and pc2 was mapped to pc4.instageni.gpolab.bbn.com

I asked for a custom Ubuntu image to be loaded in both machines. pc4.instageni.gpolab.bbn.com has indeed an Ubuntu image loaded while pc43.instageni.gpolab.bbn.com has a fedora one.

The rspec is attached.

The problem is probably caused by the unfortunate naming.

Attachments (2)

transgeo-iggpo-1vms-2raw.xml (2.5 KB) - added by nriga@bbn.com 9 years ago.
transgeo-iggpo-1vms-2raw.2.xml (2.6 KB) - added by nriga@bbn.com 9 years ago.

Download all attachments as: .zip

Change History (5)

Changed 9 years ago by nriga@bbn.com

Changed 9 years ago by nriga@bbn.com

comment:1 Changed 9 years ago by nriga@bbn.com

This time the OS loaded correctly on both raw pcs.

comment:2 Changed 9 years ago by nriga@bbn.com

Description: modified (diff)
Summary: Problem loading image in raw node on IG BBNResource mapper bug when clientId is the same as the name of a physical host

comment:3 Changed 9 years ago by nriga@bbn.com

While debugging Leigh said that :

Yep, you tripped over a pretty obscure bug in the resource mapper.
Basically, by naming your nodes the way you did, the resource mapper
mistakenly thought you wanted vm1 and vm2 mapped to pc1 which was mapped to
pc3. 

Basically do not use as clientIds the names of hosts

Note: See TracTickets for help on using tickets.