v7.3 and v7.3.1 [stable] is released!

Hi,
Routing marks now have higher priority.
One way to make rules work.
Separate the route tables (routing marks) the rules check vs the route tables the rules specify as a lookup.

eg. route mark packets with via-alt table

/routing table
add disabled=no fib name=Rvia-alt
add disabled=no fib name=via-alt

/ip firewall mangle
add action=mark-routing chain=prerouting comment=via-alt new-routing-mark=via-alt passthrough=yes

Rule to select Routing Table (Note lookup table != routing mark)
add action=lookup disabled=no dst-address=0.0.0.0/0 routing-mark=via-alt table=Rvia-alt

In ip route table only have Routing table entries with Rvia-alt (I put in R to indicate goes in routing table)
Do not have entries with via-alt in ip Route Table.

Hopefully that works ok for you.

WireGuard in 7.3 have problem . connectivity is ok but icmp not working ! how to sloved that?

No, I’m not, I’m just coming from 7.2.3. I could reproduce the same issue in RB4011 and mAP, so definitely not ARM related. And RIP is not the problem, it acts later, once client has obtained and IP, to create routing table. With or without RIP, DHCP Client should obtain an IP from VLAN 3.

Regards.

Help is F1

HATS OFF for finally posting a warning in the topic for cases of severe breakage. It would be even better when it was shown also on the router itself (i.e. put in the changes list).

i have upgraded RouterBOARD 952Ui-5ac2nD, and the router just died.
it keeps rebooting by itself

RB4011: When having a bridge with VLAN-filtering (and hw accel), containing ports that have a numbered VLAN set to “untagged”, it appears that output going to such a port also “floods” to other untagged ports in the same VLAN, as if it is a HUB instead of a SWITCH. When the second port is on the other hardware switch the leakage is about 100%, when it is on the same switch it is much less than that.
So, when you have had some interval with 20Mbps traffic to port 8 (untagged port on a VLAN), you see the same 20Mbps on port 4 (untagged port in the same VLAN), and like 0.5Mbps on port 7. This is shown in the traffic counters for the interface, and in the graphing. Torch does not show the actual traffic going to the port.

Thanks for the clue, i try but still some ip not working, i decided too add ip address in firewall address, this work for some other is not, my goal is like this my ISP have a few CDN like netflix,GGC,facebook,akamai for this CDN i need to route to my ISP other than that routing go too wireguard warp Cloudflare VPN, that why i add mangle in ip firewall address list nice addres, mangle rule is like this

add action=mark-routing chain=prerouting comment=International \
    dst-address-list=!nice new-routing-mark=Cloudflare passthrough=yes \
    src-address-list=private-lokal

that mangle have conflicting to some ip so thats why i adding to routing rules for some ip to route to my isp not VPN, on ROS 7.2.1 it works but on ROS 7.3 doesnt work like you say it “Routing marks now have higher priority”, this is my config i hope you can give me some advice
dioey.rsc (393 KB)

Upgraded one RB5009 to 7.3, no immediate issue, export diff clean.

Edit : had total loss of connectivity some hours after.
Need further check to see if it is related or not.

As a RB3011 user with an SFP for uplink to my RB260GSP I am glad I held off installing yet another ‘STABLE’ release :frowning:

Succesfully upgraded 7.3 to CCR2004, but I have question about BGP route filtering.

I using input.accept-* filter. exactly input.accept.NLRI where i have some prefixes in address list. Prefiltering works good but .. in routing table i have still /32 ipv4 routes as FILTERED and can not deny storing it in memory. Is there trick how to prefilter ipv4 /32 routes? Other higher prefixes is prefiltered perfectly.

THX

–zs

Presumingly these /32 routes are part of a subnet that is accepted in the input.accept-nlri address list? You can only filter addresses outside that address list, not smaller (e.g. single) addresses inside the listed networks.

Unfortunately I tried the upgrade before the warning was released. Had to reset, downgrade then I restored from the backup I made before the upgrade.
Waiting now for V7.3.1 :slight_smile:

No, These /32 routes is not parts of accepted nlri addresses

-zs

Successfully upgrade Chateau LTE12, mAP lite and hAP lite.

Installed on CCR2116
2.5Gbps mode still crashes the switch chip
Switch rules from cpu port still not working

SLAAC configuration in CHR works case by case, but if it works, the autoconfigured dynamic GUA is shown in

/ipv6 address print
Flags: D - DYNAMIC; G, L - LINK-LOCAL
Columns: ADDRESS, INTERFACE, ADVERTISE
#    ADDRESS                                   INTERFACE  ADVERTISE
0 DL fe80::20c:29ff:fe8b:2fc2/64               ether1     no       
1 DL fe80::20c:29ff:fe8b:2fcc/64               vlan20     no       
2 DL fe80::20c:29ff:fe8b:2fcc/64               vlan10     no       
3 DL fe80::20c:29ff:fe8b:2fcc/64               ether2     no       
4 DG xxxx:10d:c089:e002:20c:29ff:fe8b:2fc2/64  ether1     no

with DG flags, which is cool.
even a default gateway is installed based on the received RA, but it is not visible under

[admin@l4stik] > /ipv6/route/print 
Flags: D - DYNAMIC; A - ACTIVE; c, s, y - COPY
Columns: DST-ADDRESS, GATEWAY, DISTANCE
#     DST-ADDRESS              GATEWAY                           DISTANCE
  DAc xxxx:10d:c089:e002::/64  ether1                                   0
  DAc fe80::%ether1/64         ether1                                   0
  DAc fe80::%ether2/64         ether2                                   0
  DAc fe80::%vlan10/64         vlan10                                   0
  DAc fe80::%vlan20/64         vlan20                                   0

at all. i was also expecting a route with DA flags there, just as with DHCP.
hopefully this gets fixed soon.

also, it would be nice to have a bit more control over accepting RA messages. say being able to set an interface (list) from where RA messages are accepted under /ipv6 settings
maybe even the possibility to start a script upon successful SLAAC or an expiring RA, just like with DHCP client.

Please keep this forum topic strictly related to this particular RouterOS release.
Not strictly related:
http://forum.mikrotik.com/t/posts-not-strictly-related-to-v7-3-and-v7-3-1-stable/158547/1

Still the same Problems with DualWan Config, on 7.2.1 the Config is running smoothly, upgrade to 7.2.3,
Router not accessible through IP only MAC, upgrade to 7.3 is the same behaviour.
If i contact Support, they recommend to call a Consultant.

I probably know what it is, and it’s not you, it’s RouterOS. I now tested it and it happened between 7.2.1 and 7.2.2. Previously if there were multiple routing tables, local destinations (addresses assigned to router) always had priority and used main routing table, but it doesn’t happen anymore. Simplified (incomplete, but shows relevant parts) example:

/ip address
add interface=LAN address=192.168.88.1/24
/ip route
add dst-address=0.0.0.0/0 gateway=x.x.x.x routing-table=WAN2
/ip firewall mangle
add chain=prerouting src-address=192.168.88.20 action=mark-routing new-routing-mark=WAN2
add chain=forward src-address=192.168.88.20 action=log
add chain=input src-address=192.168.88.20 action=log

If 192.168.88.20 connects to 192.168.88.1, in 7.2.1 and older the log will show packets in input. In 7.2.2 and newer they go in forward. It could be intentional, but since I don’t see anything like that in changelog, it’s probably bug. And annoying one, because it can significantly break some configs.

Edit: Well, it’s not strictly about 7.3, but it’s still very much broken/changed there, so it’s interesting for anyone with such config, because upgrade can easily lock you out.

Thank you for this post! Please forward it to developers! We suffer from this issue and have to stay on 7.2.1.