This is correct, they exist for the bypass LSPs. I wouldn't characterize it as a dirty hack though. RFC4090 fast reroute requires the backup pathways to be pre-computed for a sub-10ms switchover. You put an export policy in place to make sure all labels (including bypass) are in the FIB already. Once a tear event occurs, the hidden RSVP route is just flipped to active, and LSPs using that /32 start pushing the bypass label on the stack. Since that label is already in the FIB, it just works from there.
On Mon, Dec 11, 2023 at 9:27 AM Michael Hare via juniper-nsp < juniper-nsp@puck.nether.net> wrote: > Hi Misak, > > I think what you're seeing is normal for protection LSPs, "dirty hack on > the control plane side", but I'm looking forward to be humbled on this list > that my conclusion is incorrect. > > We use "ldp interface link-protection dynamic-rsvp-lsp" and for all my > bypass LSPs, 'show route hidden table inet.3 detail' tells me > > Label-switched-path et-0/1/0.3402:BypassLSP->143.235.32.2 > ... > State: <Hidden Int ProtectionLSP> > Inactive reason: Unusable path > > I agree this is disconcerting if you are trying to get hidden routes to be > zero, but there are other normal reasons for routes to be hidden such as > rejection by bgp import policy. Better IMHO to focus instead [or > additionally] on " show route resolution unresolved " > > -Michael > > > -----Original Message----- > > From: juniper-nsp <juniper-nsp-boun...@puck.nether.net> On Behalf Of > > Misak Khachatryan via juniper-nsp > > Sent: Monday, December 11, 2023 7:03 AM > > To: juniper-nsp@puck.nether.net > > Subject: [j-nsp] RSVP hidden routes in inet.0 > > > > Hello, > > > > Recently I implemented RSVP in my network, nothing so fancy - automesh > and > > autobandwidth with node-link protection. > > > > By doing final review i saw output of show route summary: > > > > inet.0: 296 destinations, 298 routes (275 active, 0 holddown, 21 hidden) > > Direct: 6 routes, 5 active > > Local: 5 routes, 5 active > > OSPF: 265 routes, 264 active > > RSVP: 21 routes, 0 active > > LDP: 1 routes, 1 active > > > > It is very curious for me why I see hidden RSVP routes in inet.0. It > seems > > somehow related to bypass LSP's and how Juniper organises it. Here they > are: > > > > > show route protocol rsvp table inet.0 hidden > > > > inet.0: 296 destinations, 298 routes (275 active, 0 holddown, 21 hidden) > > @ = Routing Use Only, # = Forwarding Use Only > > + = Active Route, - = Last Active, * = Both > > > > 10.255.0.21/32<http://10.255.0.21/32> [RSVP] 01:11:54, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.222->10.255.0.21 > > 10.255.0.29/32<http://10.255.0.29/32> [RSVP] 1d 10:26:25, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.0.29 > > 10.255.0.33/32<http://10.255.0.33/32> [RSVP] 1d 10:26:25, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.0.33 > > 10.255.0.38/32<http://10.255.0.38/32> [RSVP] 1d 09:32:03, metric 1 > > > to 10.255.0.230 via ae4.7, label-switched-path > Bypass- > > >10.255.0.222->10.255.0.38 > > 10.255.0.70/32<http://10.255.0.70/32> [RSVP] 04:53:42, metric 1 > > > to 10.255.0.230 via ae4.7, label-switched-path > Bypass- > > >10.255.0.226->10.255.0.70 > > 10.255.0.73/32<http://10.255.0.73/32> [RSVP] 1d 10:26:21, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.0.73 > > 10.255.0.122/32<http://10.255.0.122/32> [RSVP] 1d 10:26:21, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.0.122 > > 10.255.0.126/32<http://10.255.0.126/32> [RSVP] 1d 10:26:41, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.0.126 > > 10.255.0.134/32<http://10.255.0.134/32> [RSVP] 1d 05:27:20, metric 1 > > > to 10.255.0.230 via ae4.7, label-switched-path > Bypass- > > >10.255.0.222->10.255.0.134 > > 10.255.0.174/32<http://10.255.0.174/32> [RSVP] 1d 07:19:25, metric 1 > > > to 10.255.0.230 via ae4.7, label-switched-path > Bypass- > > >10.255.0.222->10.255.0.174 > > 10.255.0.181/32<http://10.255.0.181/32> [RSVP] 1d 10:26:19, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.0.181 > > 10.255.0.185/32<http://10.255.0.185/32> [RSVP] 1d 10:26:19, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.0.185 > > 10.255.0.201/32<http://10.255.0.201/32> [RSVP] 1d 10:17:37, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.222->10.255.0.201 > > 10.255.0.214/32<http://10.255.0.214/32> [RSVP] 03:16:59, metric 1 > > > to 10.255.0.222 via ae0.7, label-switched-path > Bypass- > > >10.255.0.226->10.255.0.214 > > 10.255.0.222/32<http://10.255.0.222/32> [RSVP] 1d 10:17:34, metric 1 > > > to 10.255.0.230 via ae4.7, label-switched-path > Bypass- > > >10.255.0.222 > > 10.255.0.226/32<http://10.255.0.226/32> [RSVP] 02:45:52, metric 1 > > > to 10.255.0.222 via ae0.7, label-switched-path > Bypass- > > >10.255.0.226 > > 10.255.0.230/32<http://10.255.0.230/32> [RSVP] 1d 10:26:17, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230 > > 10.255.25.69/32<http://10.255.25.69/32> [RSVP] 1d 10:26:17, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.25.69 > > 10.255.25.73/32<http://10.255.25.73/32> [RSVP] 1d 10:26:15, metric 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.25.73 > > 10.255.25.150/32<http://10.255.25.150/32> [RSVP] 1d 10:26:50, metric > > 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.25.150 > > 10.255.25.158/32<http://10.255.25.158/32> [RSVP] 1d 10:26:50, metric > > 1 > > > to 10.255.0.226 via ae1.7, label-switched-path > Bypass- > > >10.255.0.230->10.255.25.158 > > > > The /32 routes here are the IPs of adjacent routers. The only thing I > found in > > inet is that Juniper does something similar with LDP over RSVP. > > > > It seems like some dirty hack on the control plane side, but I will be > very > > grateful if someone can explain. > > > > Best regards, > > Misak Khachatryan > > _______________________________________________ > > juniper-nsp mailing list juniper-nsp@puck.nether.net > > https://puck.nether.net/mailman/listinfo/juniper-nsp > _______________________________________________ > juniper-nsp mailing list juniper-nsp@puck.nether.net > https://puck.nether.net/mailman/listinfo/juniper-nsp > _______________________________________________ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
