Eugene
Winbox support for routing-test package
Since 2.9.20 Winbox has support for routing-test package. Happy routing.
Eugene
Eugene
Ok, just downloaded and installed on a non-production router. Disabled the routing package and enabled the routing test package. Rebooted and now I see the new BGP module. However, this router is not configured for BGP.some things are completely rewritten, like BGP for example.
Before we begin upgrading our routers that are running BGP is there anything we should know? Will v2.9.7 BGP settings transfer? Will v2.8.28 BGP settings transfer?
Many thanks,
Brad
-
-
NetTraptor
Frequent Visitor
- Posts: 76
- Joined:
- Location: Athens Greece
- Contact:
How do we declare the subnet(s) we are going to announce through BGP… in routing-test
i.e. / routing bgp network add network=10.1.1.0/24 In the mainstream routing package syntax
I found that with a static route 10.1.1.0/24 to one of the Intrefaces does the same thing but only if I redistribute static routes… which is no good at all because I don’t want other routers to know about them …
Well one would say that I could set up some filters about what should I announce and what not… but it’s the long way around something so simple! ?????
Am I doing something wrong or is there something I do not see…????
i.e. / routing bgp network add network=10.1.1.0/24 In the mainstream routing package syntax
I found that with a static route 10.1.1.0/24 to one of the Intrefaces does the same thing but only if I redistribute static routes… which is no good at all because I don’t want other routers to know about them …
Well one would say that I could set up some filters about what should I announce and what not… but it’s the long way around something so simple! ?????
Am I doing something wrong or is there something I do not see…????
-
-
NetTraptor
Frequent Visitor
- Posts: 76
- Joined:
- Location: Athens Greece
- Contact:
I think that the network command is in the specs of BGP. It is not wise to set static routes and announce them instead of setting the nework announced.BGP will only annouce routes in the routing table...How do we declare the subnet(s) we are going to announce through BGP… in routing-test
Sam
For example. In or community network we use the 10.0.0.0/8 and about 300 subnets like this are announced through BGP. Now lets say that I want to route my inet connection so I add a route to 0.0.0.0/0 from 192.168.0.1 which is in my local net (why? because I want to, I can and I should be able to). Or someone has other static routes which refer to routing that needs to be done locally.
Now in this version I have to set a static route lets say 10.21.127.0/24 to an interface in my router preferably one that has an IP in the range of 10.0.0.0/8 (I don’t know if that makes sense but just in case) so that by using static routes redistribute in the bgp I can announce my subnet as well! AND not only… I will pass to all the other nodes, which I do not want to, all my static routes about my net connection and other local static routes that I don’t want others to know! That is unless I set up filters as to what I announce and what I can receive from others! Now that is not very clever!
Hope someone comes up with a neater solution!
Hello,
I had to install 2.9.20 on our production router (tops about 230 mpbs - together both ways), because old 2.8.26 with old kernel generated packet loss with such heavy traffic.
We have really simple configuration, transit connection to the world and local connection in peering center (http://www.nix.cz), about 100 peers, works great for first touch, packet loss is away with the 2.9.x. No load balancing or such things, only routing to the right gateways according BGP.
Hardware configuration: 2.2 GHz celeron, 3 x GE intel NIC, msi motherboard with intel chipset on the border router.
Amazing
, thanks to mikrotik, good work.
I had to install 2.9.20 on our production router (tops about 230 mpbs - together both ways), because old 2.8.26 with old kernel generated packet loss with such heavy traffic.
We have really simple configuration, transit connection to the world and local connection in peering center (http://www.nix.cz), about 100 peers, works great for first touch, packet loss is away with the 2.9.x. No load balancing or such things, only routing to the right gateways according BGP.
Hardware configuration: 2.2 GHz celeron, 3 x GE intel NIC, msi motherboard with intel chipset on the border router.
Amazing
, thanks to mikrotik, good work.i agree with NetRaptor about BGP
Why don`t use some useful features from "old" routing package?
In my situation i have multihomed BGP with two peers. In other side have cisco routers and their administrators told me "you don`t have filters, you send to us all from routing table" and "your networks are missing"! where and how to filter, where is network=xx.xx.xx.xx ?
Please be kind to wrote whats goin` on with filters and network?
Sometime things need to be KISS (keep it simple stupid) ?
I saw mikrotik developers have own interpretation for implementing BGP management. Why ?
I`m not mikrotik developer and i`m in trouble with your product.
Nevermind... just tell here when mikrotik will be ok?
sorry for bad engRish (engRish as a third language)
In my situation i have multihomed BGP with two peers. In other side have cisco routers and their administrators told me "you don`t have filters, you send to us all from routing table" and "your networks are missing"! where and how to filter, where is network=xx.xx.xx.xx ?
Please be kind to wrote whats goin` on with filters and network?
Sometime things need to be KISS (keep it simple stupid) ?
I saw mikrotik developers have own interpretation for implementing BGP management. Why ?
I`m not mikrotik developer and i`m in trouble with your product.
Nevermind... just tell here when mikrotik will be ok?
sorry for bad engRish (engRish as a third language)
"you don`t have filters, you send to us all from routing table"
You should use /routing filter to specify what to send them ... Routing-test is a huge improvement over regular routing package.
Here is an example:
add chain=test-in invert-match=no action=discard comment="" disabled=no
The above rule discards all incoming routes, probably not wanted, just an example.
add chain=test-out prefix=21.0.0.0/8 prefix-length=24 invert-match=no \
action=accept comment="" disabled=no
add chain=test-out prefix=20.0.0.0/8 prefix-length=24 invert-match=no \
action=accept comment="" disabled=no
add chain=test-out invert-match=no action=discard comment="" disabled=no
The above rules allow outbound annoucements of 20/8 and 21/8 in /24 increments and filters the rest.
Sam
You should use /routing filter to specify what to send them ... Routing-test is a huge improvement over regular routing package.
Here is an example:
add chain=test-in invert-match=no action=discard comment="" disabled=no
The above rule discards all incoming routes, probably not wanted, just an example.
add chain=test-out prefix=21.0.0.0/8 prefix-length=24 invert-match=no \
action=accept comment="" disabled=no
add chain=test-out prefix=20.0.0.0/8 prefix-length=24 invert-match=no \
action=accept comment="" disabled=no
add chain=test-out invert-match=no action=discard comment="" disabled=no
The above rules allow outbound annoucements of 20/8 and 21/8 in /24 increments and filters the rest.
Sam
thanks for your post
i`ll setup two brand new 2.9.23 to test this rules example
but still suggest mikrotik developers do not give us links to cisco
we need mikrotik syntax to setting up router properly, is not only experience with bgp or cisco
but still suggest mikrotik developers do not give us links to cisco
we need mikrotik syntax to setting up router properly, is not only experience with bgp or cisco
Ok, just downloaded and installed on a non-production router. Disabled the routing package and enabled the routing test package. Rebooted and now I see the new BGP module. However, this router is not configured for BGP.some things are completely rewritten, like BGP for example.
Before we begin upgrading our routers that are running BGP is there anything we should know? Will v2.9.7 BGP settings transfer? Will v2.8.28 BGP settings transfer?
Many thanks,
Brad
BGP configuration from regular routing package is ignored in routing-test.
Eugene
BGP routing-test problem
So, i setting up one BGP link with routing test and saw what happened:
work only with "redistribute connected" - on mikrotik documentation for BGP command reference is connected and static... WHY ?
Again i recommend those people who wrote documentation to provide us
useful primer to configure , this links to cisco make me fit of nerve!!!
Commands to manage BGP IS NOT like Cisco!!!
Filtering out update message on peer connection with other side - NOT WORK. If i setup filter - i CANT get prefixes , just get state established.
Filter OUT is in this order :
my networks accept
other - reject
i don`t see nothing that works for me ... just click to check some options ...and see work or not ...
Like one of users in forum said :
we are your experimental lab or something ?
in every new version something work , something not ...
In my opinion work is done but not very well - missing documentation and primer for configuration is a problem for me.
just want to figure out multi homed BGP router ...
work only with "redistribute connected" - on mikrotik documentation for BGP command reference is connected and static... WHY ?
Again i recommend those people who wrote documentation to provide us
useful primer to configure , this links to cisco make me fit of nerve!!!
Commands to manage BGP IS NOT like Cisco!!!
Filtering out update message on peer connection with other side - NOT WORK. If i setup filter - i CANT get prefixes , just get state established.
Filter OUT is in this order :
my networks accept
other - reject
i don`t see nothing that works for me ... just click to check some options ...and see work or not ...
Like one of users in forum said :
we are your experimental lab or something ?
in every new version something work , something not ...
In my opinion work is done but not very well - missing documentation and primer for configuration is a problem for me.
just want to figure out multi homed BGP router ...
Re: BGP routing-test problem
need some answer here please, same problem!So, i setting up one BGP link with routing test and saw what happened:
work only with "redistribute connected" - on mikrotik documentation for BGP command reference is connected and static... WHY ?
thanks
Re: BGP routing-test problem
I've just finished to set up my MT network using bgp only.. i'd like to help you, but i don't understand the question... what's the problem?need some answer here please, same problem!So, i setting up one BGP link with routing test and saw what happened:
work only with "redistribute connected" - on mikrotik documentation for BGP command reference is connected and static... WHY ?
thanks
Bye,
Ricky
I'm testing 2.9.24 routing-test in a dev environment and having major issues. It seems like MT is keeping routes that it learned even after peer disconnects...
I have 2 external routers running quagga, 2 different AS numbers both with 180,000 routes each. I then have 2 Mikrotik BGP routers, one peering with each of the above. Then a cross connect between MT routers on another interface with a peering setup. I don't know if I am doing anything wrong but I have major problems with CPU usage, leftover routes not being removed, phantom routes in memory even though the peering session stopped 30 mins prior, etc. It's not ready in my opinion.
PS - quagga can dish up 180,000+ routes on a p3 1000 with barely any cpu usage ... mikrotik on a dual core p4 3.2ghz seems to hit 100% cpu for 10 minutes and then gets all squirly afterwards : )
I'm putting together some supouts to have support look at them, although I've sent 2-3 bgp related issues with 2.9.23+ and never hear back. . . If anyone is interested in peering with these 2 quagga boxes for testing feel free to drop me a line.
Sam
I have 2 external routers running quagga, 2 different AS numbers both with 180,000 routes each. I then have 2 Mikrotik BGP routers, one peering with each of the above. Then a cross connect between MT routers on another interface with a peering setup. I don't know if I am doing anything wrong but I have major problems with CPU usage, leftover routes not being removed, phantom routes in memory even though the peering session stopped 30 mins prior, etc. It's not ready in my opinion.
PS - quagga can dish up 180,000+ routes on a p3 1000 with barely any cpu usage ... mikrotik on a dual core p4 3.2ghz seems to hit 100% cpu for 10 minutes and then gets all squirly afterwards : )
I'm putting together some supouts to have support look at them, although I've sent 2-3 bgp related issues with 2.9.23+ and never hear back. . . If anyone is interested in peering with these 2 quagga boxes for testing feel free to drop me a line.
Sam
Since im using policy routes most of the time to shoow traffic out of lower cost carriers switching to routing test and getting a full feed but rejecting all the route injections into the routing tables wouldnt that make more sense ? And routing would continue to run just the full feed would be pretty much useless since I use policy routes.
Yes, we had to do this go get MT to play nice with bgp. 180,000 routes from 2 providers with only 2 possible directions seems kinda dumb. We reject all routes from 1 provider and get only customer gateways from the other - and policy routing / ecmp detemines which way to send the rest. A total of about 20,000 routes on our current 2.9.6 router seems fine ... but seeing that a quagga or other bgp device can handle it no problem is a little strange.