On Thu, 21 Apr 2016, Martin Winter wrote:
This does not seem to be the case. In this case, QUAGGA seems to be different
and causes the issues.
No need to “migrate” a 9-year old bug.
Let’s just fix it.
The concern is routing loops. The "just fix it" approach need not fix that
risk in mixed environments, but actually /increase/ the risk of them in
more networks - cause commits to upstream Quagga don't just magically
auto-update every Quagga install.
If this issue is serious issue enough to deserve this amount of
teeth-gnashing, then it's serious enough to fix in a way that minimises
the risk of *extending* the problem.
If it's not actually that serious, then it's fine to consider other
issues, like minimising behavioural churn given that there's a standards
track method on the horizon to signal Quagga's "no-transit"
You can't have it both ways. You can't claim this is a serious problem and
then get annoyed at someone for taking it seriously, looking into it,
implementing the config changes (missing from the first patch), and
implementing a safe universal way to at least signal the "transit ok"
desire.
Or alternatively claim it isn't serious and get annoyed at someone for
then looking into whether there are ways to minimise loss of
long-standing, useful functionality.
Personally, I suspect the occurrence of these loops would be very rare.
Stub-router support is not something you would tend to use with routers
that are the only transit point for some paths. You'd tend to only use it
when you have redundancy, with the intention of shifting traffic to other
routers. So in general, how this feature is intended, loops shouldn't
happen regardless (which is probably the RFC disregarded them).
However, if a loop occurs, I agree it can be nasty. So... we shouldn't
"fix" this issue on some networks by creating it on others, if we can
avoid that.
I get the sense part of the problem here is that people havn't looked at
what the patch is trying to do and the matrix of possibilities. "Think of
the RFC Compliance! The Routing loops!" - well, that's exactly what the
patch set is trying to cover - work to improve the former while avoiding
adding more of the latter.
regards,
--
Paul Jakma, HPE Aruba, Advanced Technology Group
Fortune:
QOTD:
"You're so dumb you don't even have wisdom teeth."_______________________________________________
Quagga-dev mailing list
[email protected]
https://lists.quagga.net/mailman/listinfo/quagga-dev
