Opened 9 years ago

Closed 9 years ago

#44 closed (fixed)

RSpecs: Clean up link component_id for readability

Reported by: ahelsing@bbn.com Owned by: ibaldin@renci.org
Priority: minor Milestone:
Component: AM Version: SPIRAL4
Keywords: Cc:
Dependencies:

Description

Clean up link component_id for readability

Current is: urn:publicid:IDN+geni-orca.renci.org+interface+osgrenciNet.rdf#OsgrenciNet/Juniper/1600/GigabitEthernet/gB/1/ethernet

Can you do something like: "<prefix>+<site>+interface+<node or switch name>:<port#>:<remote end name>, EG: "<prefix>+emulab.net+interface+procurve-pgeni-wash:10:ion"

Maybe it is just using ':' or '+' instead of '#' and '/'.

Name the port# where possible, and the remote end where possible.

Change History (11)

comment:1 Changed 9 years ago by ibaldin@renci.org

Owner: changed from somebody to ibaldin@renci.org
Status: newassigned

comment:2 Changed 9 years ago by ibaldin@renci.org

This should be fixed.

comment:3 Changed 9 years ago by ibaldin@renci.org

Remote end information is not readily available.

comment:4 in reply to:  3 Changed 9 years ago by ahelsing@bbn.com

This requires manual configuration.

For this element, naming the remote end could be as simple as "ion", and is mostly just helpful.

In general though, GENI stitching requires that each AM know the remote side's name for their shared links. Something somewhere needs to know that when you call it 'Foo', you are referring to the thing the other end calls 'Bar' - otherwise it's impossible to reason about the topology.

Within the stitching extension, for connections from 1 AM to another AM, the <link> has a <remote_link> sub-element -- and this needs to name the remote end, in the way that the remote AM names the link. This should be the same basic information.

I realize this takes work to gather this information. But for GENI stitching to work, it is required. So prioritize it with other GENI stitching efforts, but don't write it off.

Replying to ibaldin@renci.org:

Remote end information is not readily available.

comment:5 Changed 9 years ago by ahelsing@bbn.com

The funny long component_ids look like they remain. Your comment above says this is fixed. What changed?

comment:6 Changed 9 years ago by ibaldin@renci.org

I replaced '#' and '/' with ':'

At least I thought I did.

comment:7 Changed 9 years ago by ahelsing@bbn.com

I don't see it.

EG:

urn:publicid:IDN+exogeni.net:acisrenciNet+link+acisrenciNet.rdf#AcisRenciNet/Juniper/3200/GigabitEthernet/gB/1/ethernet

comment:8 Changed 9 years ago by ibaldin@renci.org

Is this ad or manifest? Main or extension?

comment:9 Changed 9 years ago by ahelsing@bbn.com

This is a quote from the Ad Rspec from the ExoSM, from the main section. From the BBN rack, the links are similar.

The IDs match in the stitching extension (as desired).

comment:10 Changed 9 years ago by ahelsing@bbn.com

Looking much better. Thank you.

We're down to (were we already there?) quibbling. You have: urn:publicid:IDN+exogeni.net:bbnNet+link+NLR:BBN:Cisco:6509:TenGigabitEthernet:1:2:BbnNet:IBM:G8052:TenGigabitEthernet:1:0

With all colons separating fields, it's a little hard to parse. What about using a hyphen to separate the 2 endpoints, as in: urn:publicid:IDN+exogeni.net:bbnNet+link+NLR:BBN:Cisco:6509:TenGigabitEthernet:1:2-BbnNet:IBM:G8052:TenGigabitEthernet:1:0

I don't know what the 1:2 or 1:0 numbers are at the end, but the convention that others seem to use for similar pairs of numbers at the end is to use a single forward slash (matching something in the DCN world maybe?), so it would be: urn:publicid:IDN+exogeni.net:bbnNet+link+NLR:BBN:Cisco:6509:TenGigabitEthernet:1/2-BbnNet:IBM:G8052:TenGigabitEthernet:1/0

But I don't know if that makes sense in your case.

comment:11 Changed 9 years ago by ahelsing@bbn.com

Resolution: fixed
Status: assignedclosed

It doesn't look like anything changed here since the last roll-out. The 2 endpoints of a link are still separated by a colon.

When we get to stitching, there may be other things to worry about - like naming end points.

For now, I'm going to close this as plenty good enough.

Note: See TracTickets for help on using tickets.