Mailing List Archive: internal RE communication

Two questons, possible related.

#1. What does the "fxp2" link in an M160 add over other models? On a
router with 4 SFMs but only one RE, fxp1 is up but fxp2 is down. On a
router with two REs, fxp2 is up. Is this a dedicated NIC to the other RE?
If so, why does it show up in an M160 but not an M20 which also has dual
RE capability? And wouldn't the fxp itself be onboard the RE, and aren't
the RE-2.0's all the same between models? I would have at least assumed
that since the M40 w/RE-2.0 still uses rj45 to connect the RE section to
the rest of the system, that the fxp's themselves were onboard the RE and
the wiring was simply included in pins with everything else, rather than
the fxp's being part of the chassis and accessed through PCI. Or did this
change with the models, so that an M160 has an extra fxp which the RE
accesses through PCI (on the MCS?). Perhaps someone who actually knows
would be so kind as to tell me just how far off my guesses are. :)

#2. Is it possible to put a family inet on an internal link, either fxp1
or fxp2, for easy communication (say for file transfers) between the two
REs?

--
Richard A Steenbergen <ras@e-gerbil.net> http://www.e-gerbil.net/ras
GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)

Friday, December 19, 2003, 8:48:12 PM, you wrote:
> Two questons, possible related.

> #1. What does the "fxp2" link in an M160 add over other models? On a
> router with 4 SFMs but only one RE, fxp1 is up but fxp2 is down. On a
> router with two REs, fxp2 is up. Is this a dedicated NIC to the other RE?

we have two ethernet switches located on each MCS. All SFM's,
FPC's and RE's are communicating via this ethernet switch and
they have all a link connection to both of them. This is the
reason why you need three ethernet controller on the RE. one
for fxp0 and two for fxp1 and fxp2 respectively.

> If so, why does it show up in an M160 but not an M20 which also has dual
> RE capability?

there is just one connection towards the switch on the
backplane which has the connection to the active SSB and
towards the redundant RE

> And wouldn't the fxp itself be onboard the RE, and aren't
> the RE-2.0's all the same between models?

yes

> I would have at least assumed
> that since the M40 w/RE-2.0 still uses rj45 to connect the RE section to
> the rest of the system, that the fxp's themselves were onboard the RE and
> the wiring was simply included in pins with everything else, rather than
> the fxp's being part of the chassis and accessed through PCI. Or did this
> change with the models, so that an M160 has an extra fxp which the RE
> accesses through PCI (on the MCS?). Perhaps someone who actually knows
> would be so kind as to tell me just how far off my guesses are. :)

you are right all RE2.0 and/or RE3.0 are identical and depending
on which system the third ethernet controller will or will not be
used.

> #2. Is it possible to put a family inet on an internal link, either fxp1
> or fxp2, for easy communication (say for file transfers) between the two
> REs?

If you perform a show interface terse you see that family tnp is
applied to it ( trivial network protocol ) with specifc addresses
you will see as well when you do show interface fxp1. You should
not touch those interfaces.

you can copy files between RE's like here from re0 to re1
file copy test re1:
This copy from one tnp address to another uses rcp internal.

hope this answers you questions
Josef