We have problem with ospf. When we enable routing-test package, routers have 100% utilization of CPU(RB112 completely die), When we enable standard routing package routers works very good. In /ip route we have about 700-800 routes(we have pppoe). Problem is on the every version of MT (on the latest too). I sent suppout.rif file (31.08.2006) to the support but I don’t have any answer. Does anybody have the same problem as me ?? Its very big problem for us, because we have many mikrotiks(legal!!!) and when routing die, our clients are furious 
We have routing filter that doesn’t work…more then 600 /32 route not filtered !!
Also 2.9.30 will ignore framed-route radius attribute
dude, running routing daemons on an RB112? Are you kidding me?
Dynamic routing protocols on end user equipment! That’s something that will keep hurting, again and again.
Dynamic routing protocols are synonymous with memory-hogs.
The route table requiring memory is bad, but the information on what is going to be put in the routing table is even worse. The routing daemons need to keep everything just to feed the route table with a small selection.
Stick to default routing on edge and keep dynamic routing to the core network. (Core vs. distribution vs. access networks)
600 /32 routes? Which routing protocol are you using? RIP or BGP?
Framed-route bug is serious, it prevents us from using that version.
This was just example of the CPU utilization problem with routing-test.
We have just bought 5 of them for tests, and everyting that is needed to RB112 die is just enable OSPF on it and connect to the network 
sten, so what mikrotik team is produce this version of routerboard(112 is shit for my at this time) ??? this can’t work with any of dynamic routing protocols ?? as AP or APC i can buy ovislink, planet, etc. in this sittuation, but i have mikrotik, I pay and I WANT! When I pay for RB or MT, i think that I pay for SUPPORT, too!
Using BGP/OSPF on a 112 is just plain stupid - they are 175mhz cpu’s designed to be CPE side. Nstream with PPP* and some firewall stuff.
It was never meant to be a AP or router as is clearly shown with a single ether port.
Even using a rb532 with Nstream anf BGP/OSPF would be pushing it
OK, but I writed on my first post about problem with routing-test (high utilization of CPU - x86, too). RB112 was as example. I writed on the forum, because I don’t have any answer(standard) from support. Standard routing package is good, but routing-test do 100% utilization of CPU(celeron, pentium 4, AMD, which have many GHz).
100% CPU is not a bad thing, it shows the system is working. Leave it for an hour or so to get the initial routing stuff out of the way and it should drop
We are using OSPF and the same problem from 2.9.28 to “now” 
IN filters OK, but OUT filtering is broken, after reboot all routes go out :-\
yes, You have a right, but when routers have 100% CPU routing die, LAN don’t work correctly. On a few seconds www and etc. don’t work. I think that routing-test have some bugs, so I writed on the forum. Standard routing package work rather correctly(in my opinion).