Several sliver in GPO InstaGENI have not configured the requested client_id value

[lnevers@bbn.com]

Several sliver requests in the GPO InstaGENI Rack this morning have resulted in nodes which did not have the client_id requested in the RSpec. As an example, the sliver IG-EXP-6-exp2 in the GPO rack requested the following client_ids:

<node client_id="GPO-VM"
component_manager_id="urn:publicid:IDN+instageni.gpolab.bbn.com+authority+cm"
exclusive="false">

<node client_id="GPO-PC"
component_manager_id="urn:publicid:IDN+instageni.gpolab.bbn.com+authority+cm"
exclusive="true">

But the nodes that are available are named "vhost1" and "pc5" and not as requested "GPO-VM" and "GPO-PC".

This has occurred several times on the GPO rack, but has not happened with requests to the Utah Rack. Is this an issue?

[lnevers@bbn.com]

Agreed this is very minor, as the ticket was originally marked. Is there a known issues page were we can capture this?

---

Capturing email exchanges for this ticket:

On 1/8/13 2:33 PM, Leigh Stoller wrote:
>> Several sliver requests in the GPO InstaGENI Rack this morning have 
>> resulted in nodes which did not have the ''client_id'' requested in the 
>> RSpec. As an example, the sliver IG-EXP-6-exp2 in the GPO rack requested 
>> the following client_ids: 
>
> Sadly, the hostname is too long:
>
> [root@pc5 logs]# /usr/local/etc/emulab/sethostname.dhclient 
> Resetting hostname to gpo-pc.ig-exp-6-exp2.pgeni-gpolab-bbn-com.instageni.gpolab.bbn.com ... hostname: name too long
>
> It would be nice to catch this in the front end, and I suppose we could 
> pick 64 as the typical value, and not let you pick slice,node names that
> are too long.
>
> That said, this is probably best considered a minor issue since it does
> not effect functionality, just what your shell prints out. 
>
> Lbs

[lnevers@bbn.com]

Hostnames that are too long are still not handled, and request for documentation in known issues still not found.