The MPLS WG Archive

["Adrian Farrel" <adrian@olddog.co.uk>]

As my first step in resolving this issue I would like to conduct a brief survey amongst
implementors.

Please feel free to respond to me off-list and in confidence. If you do not want your
information shared more widely, please state that explicitly in your email.

Also, please note that this questionnaire is NOT a discussion of what is the correct
behavior. It is simply collecting information about interpretation of RFCs 2205 and 3209.

Thanks,
Adrian

Questions:

1. When your implementation receives a PathErr message indicating pre-emption for an
established LSP, does it tear the LSP (i.e. release labels) or only release resources
(i.e. leave the LSP in place as 'best effort')?
In other words, is your default behavior hard or soft pre-emption?

2. When your implementation receives any other (non-notify) PathErr for an established
LSP, does it tear the LSP (i.e. release labels) or leave it in place?

3. Is your code carrying live traffic, in lab tests, or in development?

4. If deployed, can you give me a feeling for the number of LSRs in the field (just an
order of magnitude)?

Thanks,
Adrian