[ixpmanager] LAG inter-switch connections

[Rowan Thorpe]

Until now I have seen LAG connections working OK for customer connections
monitored by IXP-Manager (after configuring them through the interface as
multiple physical connections under a single logical connection it seems to
work it out automatically), but I've struck a new peculiarity. We now have a
LAG core connection (between two of our switches - at present it is only one
port but is nevertheless configured as part of a LAG, because more ports will
be added soon). We have had switch-polling cron errors and our inter-switch
graphs have dropped to zero since then, and it seems the only thing
architecturally different from when it was working is the LAG. I checked they
had configured all the switch ports in the interface correctly (no
active-but-unknown ports, no incorrect port-types, inter-switch ports
configured as "core", etc), so I presume it is that IXP-Manager is presuming
that all inter-switch connections are not LAGs. Is there some "trick" I don't
know about to make this work? Or is it presently just not possible to monitor
with IXP-M?. Both switches are Juniper EXes.

NB: I haven't had a chance to look at IXP-M for a while so I know we are using
a couple of versions behind the latest release, but I have scanned the latest
release notes and commits, and *superficially* haven't seen any changes related
to this problem...

--
Rowan Thorpe
PGP fingerprint:
 BB0A 0787 C0EE BDD8 7F97  3D30 49F2 13A5 265D CCBD
----
"There is a great difference between worry and concern. A worried person sees
a problem, and a concerned person solves a problem."
 - Harold Stephens