Re: [j-nsp] Netflow config for MX204
--- Begin Message --- On Sun, Apr 12, 2020 at 02:45:57AM +0200, Mark Tinka wrote: > > > On 11/Apr/20 08:04, Nick Schmalenberger via juniper-nsp wrote: > > I had the same issue with first trying to export over fxp0, then > > trying with my routing instance, and I ended up making a static > > route in inet6.0 with next-table over to the instance table where > > the route into the LAN for my elastiflow collector is. Flow > > export over IPv6 does also seem to work. > > We just export flows in-band. Just seems simpler, and has been reliable > for close to 10 years. > > Mark. > I am exporting in-band, the next-table is so the default table can access a port in my routing instance that has the in-band ports. What flow collector are you using? Any tips on the under-counting? Thanks! -Nick --- End Message --- ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] Junos Telemetry Interface
Do you have any information on the license part numbers and the MSRP for the licenses? Usually this is in the datasheets, but I don't see it anywhere. Any clarification on the free version would be appreciated. On Tue, Apr 14, 2020 at 4:33 AM Roger Wiklund wrote: > Yepp, then you have to go payed version and buy a HB license per device > you want to monitor. > > /Roger > > On Mon, Apr 13, 2020 at 11:43 PM Colton Conor > wrote: > >> Healthbot looks interesting. However, most of our Juniper gear we buy is >> used, so I doubt we have a support contract on it. Is healthbot still an >> option? >> >> On Mon, Apr 13, 2020 at 4:22 PM Roger Wiklund >> wrote: >> >>> Hi >>> >>> Check out Juniper Healthbot. It can consume Netconf, SNMP, Syslog, >>> Native and OpenConfig, also works with Cisco. >>> Still pretty fresh product but has potential. 3.0 should be released >>> soonish. Freemium approach where you can use basic features for free as >>> long as you have a service contract for the device you want to monitor. >>> Payed version has more features. >>> >>> >>> https://www.juniper.net/documentation/en_US/healthbot/topics/topic-map/healthbot-rules-n-playbooks-map.html >>> >>> note: Some errors in the documentation above that I have flagged, >>> OpenConfig, iAgent/Netconfig and SNMP can of course be used both inband and >>> out of band. >>> >>> /Roger >>> >>> On Mon, Apr 13, 2020 at 11:07 PM Colton Conor >>> wrote: >>> Roger, Thanks for the information. It seem like most all of these telemetry and flow solutions require service providers to build their own solutions ontop of InfluxDB or Elasticsearch. Do you know if there are any newer NMS platforms our there that are built ontop of InfluxDB or Elasticsearch? Basically, we don't want to do it ourselves, and would prefer to buy a solution. However, talking to someone like SolarWinds for example, and they said they don't currently support streaming telemetry, and its not on their road map at the moment? On Mon, Apr 13, 2020 at 3:23 PM Roger Wiklund wrote: > Hi > > Native sensors: > > https://www.juniper.net/documentation/en_US/junos/topics/reference/configuration-statement/sensor-edit-services-analytics.html > > OpenConfig sensors: > > https://www.juniper.net/documentation/en_US/junos/topics/reference/general/junos-telemetry-interface-grpc-sensors.html > > Plugins to consume JTI for common open sources tools: > > https://www.juniper.net/documentation/en_US/junos/topics/concept/jti-opensource-plugins.html > > Personally I run OpenConfig with Telegraf + InfluxDB + Grafana. > > Telegraf has a built-in input plugin for Juniper Openconfig, so it > takes like 5 minutes to enable. > > https://github.com/influxdata/telegraf/tree/master/plugins/inputs/jti_openconfig_telemetry > > OpenNTI is good for demo/test but not really suitable for production. > > /Roger > > On Mon, Apr 13, 2020 at 5:46 AM Aaron Gould wrote: > >> You’re welcome Colton. I understand there are 2 different ways to do >> telemetry on Juniper. One called Native and the other called >> gRPC/openconfig. I’ve done the Native form. I think the native form is >> a >> configured form where by which the network device constantly streams the >> sensor objects… and conversely, the gRPC form is subscription based where >> the management app/computer, subscribes to the network device to receive >> telem data objects. >> >> >> >> I understand the native form to be executed in hardware near the >> monitored object….and because of this, highly scalable.And the >> grpc/openconfig form runs on re cpu. >> >> >> >> I don’t think we’ve gotten native telemetry to work on ACX. But I >> have it running on MX960’s. >> >> >> >> I understand the grpc/openconfig method requires you to download some >> code/software to the network device. >> >> >> >> Collector I use is the OpenNTI project. Grafana (web ui) (or a less >> known Cronograf, which I actually use and like), InfluxDB (TSDB), >> fluentd, >> and other components. I must credit Dave, my coworker and resident Linux >> genius in assisting my with the server side collector setup. Some >> helpful/related links below…. >> >> >> >> >> >> https://puck.nether.net/pipermail/juniper-nsp/2018-October/036602.html >> >> >> >> >> >> >> https://openeye.blog/2017/06/26/using-opennti-as-a-collector-for-streaming-telemetry-from-juniper-devices-part-1/ >> >> >> >> >> >> >> https://www.juniper.net/documentation/en_US/junos/topics/concept/junos-telemetry-interface-oveview.html >> >> look under “telemetry sensors and data models” >> >> >> >> >> >> >>
Re: [j-nsp] Junos Telemetry Interface
On Tue, Apr 14, 2020 at 2:01 PM Martin Tonusoo wrote: > > Hi Dario, > > > This looks really useful, thanks for sharing. Just checking, do I only need > > this script, InfluxDB and Grafana to get traffic graphs? > > That's correct. Thanks - it's working nicely. So I can get the correct values in Grafana in bps, what reporting-rate do you have configured on the Juniper? Also in Grafana, what should the derivative value be in seconds? If I change that I get different results on the graph for bps.. ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] Junos Telemetry Interface
Hi Dario, > This looks really useful, thanks for sharing. Just checking, do I only need this script, InfluxDB and Grafana to get traffic graphs? That's correct. WBR, Martin ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] Junos Telemetry Interface
On Mon, Apr 13, 2020 at 4:46 AM Aaron Gould wrote: > > You’re welcome Colton. I understand there are 2 different ways to do > telemetry on Juniper. One called Native and the other called > gRPC/openconfig. I’ve done the Native form. I think the native form is a > configured form where by which the network device constantly streams the > sensor objects… and conversely, the gRPC form is subscription based where the > management app/computer, subscribes to the network device to receive telem > data objects. > > I understand the native form to be executed in hardware near the monitored > object….and because of this, highly scalable.And the grpc/openconfig form > runs on re cpu. I'm also deploying native JTI at the moment on some MX devices. I have got OpenNTI working, and managed to get traffic graphs working in Grafana using it. I'm now trying to set each part of this up individually without Docker, for a more production ready setup. I've been following this blog: https://openeye.blog/2017/10/31/serving-up-opennti-deconstructed/ I'm trying to get my head around the difference between fluentd and Telegraf, what does each one do, and why are both of these required? ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] Junos Telemetry Interface
On Tue, Apr 14, 2020 at 12:13 AM Martin Tonusoo wrote: > > Telegraf has a built-in input plugin for Juniper Openconfig, so it takes > > like 5 minutes to enable. > > there also seems to be a patch for native sensors: > https://github.com/influxdata/telegraf/pull/6365. Unfortunately, it's not > yet merged. In addition, in order to better understand how the native > sensors work, I wrote a small script for lab usage, which pushes native > sensors data to InfluxDB: https://github.com/tonusoo/jti-to-influxdb Maybe > someone finds this useful. Hi Martin, This looks really useful, thanks for sharing. Just checking, do I only need this script, InfluxDB and Grafana to get traffic graphs? So no fluentd, telegraf, etc? ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
Re: [j-nsp] Junos Telemetry Interface
Yepp, then you have to go payed version and buy a HB license per device you want to monitor. /Roger On Mon, Apr 13, 2020 at 11:43 PM Colton Conor wrote: > Healthbot looks interesting. However, most of our Juniper gear we buy is > used, so I doubt we have a support contract on it. Is healthbot still an > option? > > On Mon, Apr 13, 2020 at 4:22 PM Roger Wiklund > wrote: > >> Hi >> >> Check out Juniper Healthbot. It can consume Netconf, SNMP, Syslog, Native >> and OpenConfig, also works with Cisco. >> Still pretty fresh product but has potential. 3.0 should be released >> soonish. Freemium approach where you can use basic features for free as >> long as you have a service contract for the device you want to monitor. >> Payed version has more features. >> >> >> https://www.juniper.net/documentation/en_US/healthbot/topics/topic-map/healthbot-rules-n-playbooks-map.html >> >> note: Some errors in the documentation above that I have flagged, >> OpenConfig, iAgent/Netconfig and SNMP can of course be used both inband and >> out of band. >> >> /Roger >> >> On Mon, Apr 13, 2020 at 11:07 PM Colton Conor >> wrote: >> >>> Roger, >>> >>> Thanks for the information. It seem like most all of these telemetry and >>> flow solutions require service providers to build their own solutions >>> ontop of InfluxDB or Elasticsearch. Do you know if there are any newer NMS >>> platforms our there that are built ontop of InfluxDB or Elasticsearch? >>> Basically, we don't want to do it ourselves, and would prefer to buy a >>> solution. However, talking to someone like SolarWinds for example, and they >>> said they don't currently support streaming telemetry, and its not on their >>> road map at the moment? >>> >>> On Mon, Apr 13, 2020 at 3:23 PM Roger Wiklund >>> wrote: >>> Hi Native sensors: https://www.juniper.net/documentation/en_US/junos/topics/reference/configuration-statement/sensor-edit-services-analytics.html OpenConfig sensors: https://www.juniper.net/documentation/en_US/junos/topics/reference/general/junos-telemetry-interface-grpc-sensors.html Plugins to consume JTI for common open sources tools: https://www.juniper.net/documentation/en_US/junos/topics/concept/jti-opensource-plugins.html Personally I run OpenConfig with Telegraf + InfluxDB + Grafana. Telegraf has a built-in input plugin for Juniper Openconfig, so it takes like 5 minutes to enable. https://github.com/influxdata/telegraf/tree/master/plugins/inputs/jti_openconfig_telemetry OpenNTI is good for demo/test but not really suitable for production. /Roger On Mon, Apr 13, 2020 at 5:46 AM Aaron Gould wrote: > You’re welcome Colton. I understand there are 2 different ways to do > telemetry on Juniper. One called Native and the other called > gRPC/openconfig. I’ve done the Native form. I think the native form is a > configured form where by which the network device constantly streams the > sensor objects… and conversely, the gRPC form is subscription based where > the management app/computer, subscribes to the network device to receive > telem data objects. > > > > I understand the native form to be executed in hardware near the > monitored object….and because of this, highly scalable.And the > grpc/openconfig form runs on re cpu. > > > > I don’t think we’ve gotten native telemetry to work on ACX. But I > have it running on MX960’s. > > > > I understand the grpc/openconfig method requires you to download some > code/software to the network device. > > > > Collector I use is the OpenNTI project. Grafana (web ui) (or a less > known Cronograf, which I actually use and like), InfluxDB (TSDB), fluentd, > and other components. I must credit Dave, my coworker and resident Linux > genius in assisting my with the server side collector setup. Some > helpful/related links below…. > > > > > > https://puck.nether.net/pipermail/juniper-nsp/2018-October/036602.html > > > > > > > https://openeye.blog/2017/06/26/using-opennti-as-a-collector-for-streaming-telemetry-from-juniper-devices-part-1/ > > > > > > > https://www.juniper.net/documentation/en_US/junos/topics/concept/junos-telemetry-interface-oveview.html > > look under “telemetry sensors and data models” > > > > > > > https://community.grafana.com/t/how-to-send-juniper-router-telemetry-to-grafana/11071/9 > > > > > > -Aaron > > > > > > From: Colton Conor [mailto:colton.co...@gmail.com] > Sent: Saturday, April 11, 2020 9:05 AM > To: Aaron Gould > Cc: Juniper List > Subject: Re: [j-nsp] Junos Telemetry Interface > > > > Aaron, > > > > Thanks, this is indeed
Re: [j-nsp] Netflow config for MX204
--- Begin Message --- Perhaps you just needed to add (to your original config) routing-instance vrf-name under forwarding-options sampling family inet output flow-server x.x.x..x and not to overdo everything under vrf On 09-Apr-20 10:03, Liam Farr wrote: Seems I cant just drop the forwarding options into the vrf verbatim; # show | compare [edit] - forwarding-options { - sampling { - sample-once; - instance { --- End Message --- ___ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp