Rare project developers

[mc36 <>]

hi,
i go inline....
br,
cs

On 12/17/22 18:19, Antoni Przygienda wrote:
> Sorry, I cannot drop you vmx images easily. Product not supported, only internal stuff, images not supposed to go out and I   m not sure you   ll be able to run 22.x stuff anyway, tons 
> stuff has changed since we supported the product.
okk, then i'll abandon this vmx idea... i just hoped if it's listed on the 
public juniper.net then it's something one can use...:)

> Kithara is trivial, just grab a standard ub18 container, throw your stuff onto it, snapshot the image and give it to kithara on a lab. 

there is a project a team member started back in the days at 
https://github.com/rare-freertr/freeRtr-docker hopefully it still works... :)


> To test rift stuff properly you   ll be 
> building pretty soon lots tooling yourself to bring up CLOS networks and then realize you cannot run it with VMs at any reasonable size and end up with containers anyway (or 
> namespaces as Bruno   s code can partially do but that   s also super limited). RIFT only starts to really cranck once you   re at couple hundred nodes at least     
well, i'm not against the kithara nor anything but i like the tooling...
i already have templates like this:
http://src.mchome.nop.hu/cfg/temp-isis.tmpl
then i can simply use a csv with the intended connections, repetitions, etc...


> Very simplest stuff is to run within my Jason environ just like Bruno does, there is no virtualization/data plane at all but you can build very large topologies within seconds. We 
> could easily extend it from    don   t run the passive nodes    to    run only nodes with this simulation-partition-id    if you want and then can mix all implementations in any fashion    
well, freerouter dont have global so i can run 2 rift instances and peer them 
over a hairpin interface...
moreover we have a lightweight dataplane with uses simple udp sockets to pass 
ethernet frames... the
forwarder is the same found in rare's dpdk dataplane, just the packetio is 
swapped out here... as it's
a simple process in linux, one can have 1000s of freerouters with dataplanes 
in a single computer...
for example on a single 2 cpu xeon i run with the test cases with paralell 
100, and we can safely
assume that each test case spans up at least 4 routers soo it's there 
already... :)
but aggreed upon, for rift self testing, the hairpin is more than enough, 
then it's a single java process,
much like with bruno's stuff, but here java hotspot quickly produces native 
executable
(or graalvm can do in at compile time if one needs the quick warmup times 
too, like me:)

sid#show config-differences
hairpin 1
 exit
router rift4 2
 vrf v2
 router-id 12345678
 redistribute connected
 exit
router rift4 3
 vrf v3
 router-id 87654321
 redistribute connected
 exit
interface loopback2
 vrf forwarding v2
 ipv4 address 2.2.2.2 255.255.255.255
 no shutdown
 no log-link-change
 exit
interface loopback3
 vrf forwarding v3
 ipv4 address 2.2.2.3 255.255.255.255
 no shutdown
 no log-link-change
 exit
interface hairpin11
 vrf forwarding v2
 ipv4 address 1.1.1.1 255.255.255.252
 router rift4 2 enable
 no shutdown
 no log-link-change
 exit
interface hairpin12
 vrf forwarding v3
 ipv4 address 1.1.1.2 255.255.255.252
 router rift4 3 enable
 no shutdown
 no log-link-change
 exit

sid#
sid#
sid#show ipv4 rift 2 database
dir  origin    num  typ  seq  left
s    12345678  1    2    2    6d23h
s    12345678  2    3    3    6d23h
s    12345678  3    3    1    6d23h
s    87654321  1    2    4    6d23h
s    87654321  2    3    2    6d23h
s    87654321  3    3    1    6d23h
n    12345678  1    2    2    6d23h
n    12345678  2    3    3    6d23h
n    12345678  3    3    1    6d23h
n    87654321  1    2    4    6d23h
n    87654321  2    3    2    6d23h
n    87654321  3    3    1    6d23h

sid#show ipv4 rift 3 database
dir  origin    num  typ  seq  left
s    12345678  1    2    2    6d23h
s    12345678  2    3    3    6d23h
s    12345678  3    3    1    6d23h
s    87654321  1    2    4    6d23h
s    87654321  2    3    2    6d23h
s    87654321  3    3    1    6d23h
n    12345678  1    2    2    6d23h
n    12345678  2    3    3    6d23h
n    12345678  3    3    1    6d23h
n    87654321  1    2    4    6d23h
n    87654321  2    3    2    6d23h
n    87654321  3    3    1    6d23h

sid#
sid#
sid#show ipv4 route v2
typ  prefix      metric  iface      hop      time
C    1.1.1.0/30  0/0     hairpin11  null     00:01:00
LOC  1.1.1.1/32  0/1     hairpin11  null     00:01:00
C    2.2.2.2/32  0/0     loopback2  null     00:01:14
F    2.2.2.3/32  100/10  hairpin11  1.1.1.2  00:00:07

sid#
sid#
sid#show ipv4 route v3
typ  prefix      metric  iface      hop      time
C    1.1.1.0/30  0/0     hairpin12  null     00:00:29
LOC  1.1.1.2/32  0/1     hairpin12  null     00:00:29
F    2.2.2.2/32  100/10  hairpin12  1.1.1.1  00:00:21
C    2.2.2.3/32  0/0     loopback3  null     00:00:09

sid#ping 2.2.2.2 vrf v3 source loopback3
pinging 2.2.2.2, src=2.2.2.3, vrf=v3, cnt=5, len=64, df=false, tim=1000, 
gap=0, ttl=255, tos=0, sgt=0, flow=0, fill=0, alrt=-1, sweep=false, 
multi=false
!!!!!
result=100.0%, recv/sent/lost/err=5/5/0/0, took 18, min/avg/max/dev 
rtt=0/0.3/1/0.2, ttl 255/255/255/0.0, tos 0/0.0/0/0.0
sid#trace 2.2.2.2 vrf v3 source loopback3
tracing 2.2.2.2, src=2.2.2.3, vrf=v3, prt=0/33440, tim=1000, tos=0, flow=0, 
len=64
via 2.2.2.2/32 100/10 hairpin12 1.1.1.1 00:00:35
1 2.2.2.2 time=0
sid#


> If you need kithara help I can hook you up directly with one of the authors, 
> I know them well, they   ll be more than happy to bring freerouter in

ahhh, hopefully i'll figure it out alone.. i'll let you know how i progress...

thanks,
cs


> --t ony
>
> *From: *mc36 <>
> *Date: *Saturday, 17 December 2022 at 16:19
> *To: *Antoni Przygienda <>,  
> <>
> *Subject: *Re: [rare-dev] FW: rift in freerouter
>
> [External Email. Be cautious of content]
>
>
> On 12/17/22 16:18, mc36 wrote:
> > that way i'm pretty sure will work and i already have a rift build for that 
> > vmx from 2020-10-xx...
> surely a typo, 2022
>
>
> Juniper Business Use Only