RARE user and assistance email list

[mc36 <>]

hi,

thanks for reaching out, i have bad and good news....

i personally have almost zero junos operational experience, we're a cisco 
shop here... but we also
have some bugs and feature requests that will never get satisfied so we can 
say we're also unhappy...

i personally heavily interop test with junos too, take a look on a junos' 
not-yet-released bier feature:
http://www.freertr.net/present/ietf110.pdf it have some screenshots you'll 
hopefully have in the upcoming years...
i just mention that because all that we're doing can happen over whitebox 
switches who have 32x100g, 64x100g, 32x400g, etc...
regarding the control plane, we had the fortune to become part of your core 
isis, from that point we have some info (below)

and yess, rare/freertr have a streaming telemetry exporter too, so i have 
some idea how it have to work:
first of all, you define what you're interested in, and the interval... you 
mentioned snmp so i assume
now you subscribed to the interface stats... in rare, the dataplane's 
middleware reports these in every
second, the route, label, nat, acl, gre, pppoe, etc  counters in every 30 
seconds. from that, the control
plane generates the telemetry reports, and that's all. one can generate the 
reports from the line cards
but we need the asic space for new features.. :) formatting the reports into 
various formats...
at the moment we do xr's streaming telemetry, consumed by influx's xr plugin, 
prometheus, or netconf...
is not a big deal for a proper imlementation, having the whole report takes 
less than 300ms
and you reported almost a million counters... protobuf is a nice binary 
protocol...

if you have interest, we can already export you your igp stats in those 
formats... if you provide us
some bmp or bgp feeds, we also can deal with that and give back some stats 
from those too... and if you
miss a stat, it'll arrive soon if it's feasible... we're more than happy to 
fulfill a feature request....
back in time, niall expressed that you're interested in a per/pop per/node 
per/peer prefix drop stats,
a week passed and since then we're ready to stream that info too... :)

and finally, if you're brave enough, you can implement a rare/freertr box to 
have traffic stats too... :)

regards,
cs







freertr#show logging last 1000 | include establ
2021-03-26 11:43:35 warning util.shrtPthFrst.doCalc:shrtPthFrst.java:535 node 
mx1.lon2.uk.re0 established connection to mx1.dub2.ie.re0
2021-03-26 11:43:35 warning util.shrtPthFrst.doCalc:shrtPthFrst.java:535 node 
mx1.dub2.ie.re0 established connection to mx1.lon2.uk.re0
2021-03-26 12:58:47 warning util.shrtPthFrst.doCalc:shrtPthFrst.java:535 node 
mx1.poz.pl.re0 established connection to mx1.fra.de.re0
2021-03-26 12:58:47 warning util.shrtPthFrst.doCalc:shrtPthFrst.java:535 node 
mx1.fra.de.re0 established connection to mx1.poz.pl.re0
2021-03-26 17:39:55 info user.userReader.cmdEnter:userReader.java:745 command 
freertr#show logging last 1000 | include establ  from telnet #153556583 23 -> 
10.255.255.1 38482

freertr#
freertr#show ipv4 isis 1 tree 2
`--freertr
   `--mx1.par.fr.re0
     |`--mx1.gen.ch.re0
     |  |`--mx1.mil2.it.re0
     |  |  |`--mx2.lju.si.re0
     |  |  |   `--mx2.lju.si.re0-VRR
     |  |  |`--mx2.ath.gr.re0
     |  |  |   `--mx1.ath2.gr.re0
     |  |   `--mx1.vie.at.re0
     |  |     |`--mx2.zag.hr.re0
     |  |     |`--mx2.bra.sk.re0
     |  |     |`--mx1.sof.bg.re0
     |  |      `--mx1.bud.hu.re0
     |  |         `--mx1.buc.ro.re0
     |  |`--mx1.mar.fr.re0
     |   `--mx1.fra.de.re0
     |     |`--mx1.ham.de.re0
     |     |   `--rt2.tal.ee.re0
     |     |     |`--mx2.tal.ee.re0
     |     |     |   `--mx1.tal.ee.re0
     |     |      `--rt1.tal.ee.re0
     |     |         `--rt2.rig.lv.re0
     |     |           |`--rt1.rig.lv.re0
     |     |            `--mx2.rig.lv.re0
     |     |`--mx1.pra.cz.re0
     |      `--mx1.poz.pl.re0
     |         `--rt2.kau.lt.re0
     |           |`--mx2.kau.lt.re0
     |           |   `--mx1.kau.lt.re0
     |            `--rt1.kau.lt.re0
     |`--mx1.mad.es.re0
     |   `--mx2.lis.pt.re0
     |     |`--mx1.lis.pt.re0
     |      `--rt1.por.pt.re0
     |`--mx1.lon2.uk.re0
     |  |`--mx1.lon.uk.re0
     |  |  |`--mx2.bru.be.re0
     |  |  |`--mx1.ams.nl.re0
     |  |   `--rt2.ams.nl.re0
     |   `--mx1.dub2.ie.re0
     |      `--mx1.dub.ie.re0
     |`--rt1.bil.es.geant.net
      `--mx1.par.fr.re0-VRR

freertr#
freertr#show ipv6 segrout geant
prefix                   index  base  oldbase
2001:798:10:20ff::1/128  6707   272   272
2001:798:12:20ff::1/128  6714   272   272
2001:798:13:20ff::1/128  6702   272   272
2001:798:14:20ff::1/128  6712   272   272
2001:798:16:20ff::3/128  6601   272   800000
2001:798:16:20ff::4/128  6602   272   800000
2001:798:17:20ff::1/128  6716   272   272
2001:798:18:20ff::3/128  6713   272   272
2001:798:19:20ff::1/128  6611   272   800000
2001:798:1b:20ff::1/128  6701   272   800000
2001:798:1e:20ff::3/128  6715   272   272
2001:798:1f:20ff::3/128  6605   272   800000
2001:798:1f:20ff::4/128  6606   272   800000
2001:798:21:20ff::4/128  6604   272   800000
2001:798:22:20ff::3/128  6711   272   272
2001:798:23:20ff::1/128  6710   272   800000
2001:798:28:20ff::1/128  6705   272   272
2001:798:2b:10ff::3/128  6619   272   800000
2001:798:2b:20ff::2/128  6620   272   800000
2001:798:2c:10ff::2/128  6621   272   800000
2001:798:2d:20ff::2/128  6608   272   800000
2001:798:2f:20ff::1/128  6616   272   800000
2001:798:2f:20ff::2/128  6617   272   272
2001:798:32:20ff::2/128  6610   272   800000
2001:798:33:20ff::2/128  6704   272   800000
2001:798:aa:1::1/128     6603   272   800000
2001:798:aa:1::2/128     6625   272   800000
2001:798:aa:1::3/128     6626   272   800000
2001:798:aa:1::4/128     6612   272   800000
2001:798:aa:1::5/128     6615   272   272
2001:798:aa:1::a/128     6639   272   272
2001:798:aa:1::c/128     6662   272   272
2001:798:aa:1::d/128     6642   272   272
2001:798:aa:1::17/128    6650   272   272
2001:798:aa:1::18/128    6652   272   272
2001:798:aa:1::19/128    6653   272   272
2001:798:aa:1::1e/128    6658   272   272
2001:798:aa:1::1f/128    6659   272   272
2001:798:aa:1::20/128    6657   272   272
2001:798:aa:1::22/128    6642   272   272

freertr#




On 3/26/21 5:18 PM, Erik Reid wrote:
> Hello Csaba,
>
> I heard from Rick Havern that you have experience with gRPC and maybe in 
> particular Juniper streaming telemetry …?
>
> He may have told you how we’ve worked on replacing our SNMP/polling-based 
> counter/state/properties system with streaming telemetry.
>
> For the most part we were quite successful, and found the technology easy to 
> use. Â But there are a few issues (major and minor) that are preventing us 
> from making a total switch.
>
> The blocking issue for now is that our routers started generating “JLOCK hog” errors.  This was with 9 clients connected to each router, but all streams were configured with only 
> 60s frequency, so I wouldn’t have thought this load to be especially heavy.  Our operations team traced this to the gRPC service and Juniper confirmed this.  They considered this 
> critical and have shut down gRPC for the time being.
>
> Another big issue is that our operations team has defined several of the important peering sessions we’d like to monitor in logical systems.  When I discovered that we were not 
> getting telemetry from these interfaces and sessions, we raised a ticket with Juniper and got a one-line response: “Streaming Telemetry is not support in logical systems”.  We’ve 
> raised an enhancement request but have no feedback so far.
>
> There are a few other less severe issues, and some are on the software side.  For example, we’ve been using telegraf with the jti_openconfig_telemetry plugin as our clients, and 
> it’s relatively stable although there’s an issue with stability when authentication sessions fail.
>
> Do you have experience with any of these situations or tools?  Do you have 
> any advice or lessons learned that you can share?
>
> Regards,
>
> Erik Reid
>
> Solution Architect
>
> GÉANT