+1It'll be nice if you deploy one CRS as a demo system (similar to demo.mt.lv and demo2.mt.lv) so that the community can see what's the switch management looks like in the UI on a live system.
+1It'll be nice if you deploy one CRS as a demo system (similar to demo.mt.lv and demo2.mt.lv) so that the community can see what's the switch management looks like in the UI on a live system.
Looking good. I think more examples... ... basic examples of what each setting does along with more descriptions. Also complex examples.+1It'll be nice if you deploy one CRS as a demo system (similar to demo.mt.lv and demo2.mt.lv) so that the community can see what's the switch management looks like in the UI on a live system.
Regards,
We have the same problem in 6.9Hello Folks!
We then went further trying to activate port isolation, but it goes back to promiscues all the time, not possible to change.
[Resetting the unit] was only necessary for CRS125 to prevent speed issues and behaviour of a hub.
Alternatively you can enter this command:Code: Select allCode: /interface ethernet switch port set [find] learn-restricted-unknown-sa=yes
This is not just a small bug . This is a huge security vulnerability . Mikrotik should inform users to don't use CRS in production .Hello Folks!
CRS still leaks in RoS6.10, exactly like before.
Tested after resetting CRS, followed by the suggested steps in first mikrotik exampe port based.
For simplicity we used one "trunk port" ether2 and one "access port" ether3.
Running tcpdump on a redhat llinux based server connected to ether3 show arp requests from ALL vlans and other traffic.
Please, can anyone come up with a working non leaking example configuration, how to do it so to say, we badly need going gigabit now ?
Totally agree, CRS without normal documentation and VLAN operation, is useless. Now my two CRS125 will stand on a shelf until bugs are fixed and CRS is normally documented.This is not just a small bug . This is a huge security vulnerability . Mikrotik should inform users to don't use CRS in production .Hello Folks!
CRS still leaks in RoS6.10, exactly like before.
Tested after resetting CRS, followed by the suggested steps in first mikrotik exampe port based.
For simplicity we used one "trunk port" ether2 and one "access port" ether3.
Running tcpdump on a redhat llinux based server connected to ether3 show arp requests from ALL vlans and other traffic.
Please, can anyone come up with a working non leaking example configuration, how to do it so to say, we badly need going gigabit now ?
the original topic in this post asks what features do you wish to be documented in more detail. the original post is about documentation!and CRS is normally documented.
AFAIK you can only route via the embedded CPU, so inter-VLAN routing is achieved by switching the affected traffic through the 1G-CPU-Uplink, forwarding is then done in software and the packet is sent back though the same 1G-CPU-Uplink to the switching silicon. Besides MT claiming otherwise, the CRS isn't a L3 switch is a switch combined with a "router on a stick" in networkers terminology.Hello,
I just would like some more configuration examples such as a basic L3 switch with inter-VLAN routing. The way how a VLAN interface interacts with the switch is pretty unclear for me.
Thanks!
Did you reset to defaults after upgrade to 6.10 or 6.11? ...Oh thank God it's not only me seeing that "leak"!
Using CRS ROS 6.11 - I'm not able to ping, but I can gladly see other MT on other VLAN's with WinBox, also sniffing traffic with Wireshark and I see it leaks traffic.
Seriously.... fucking great! ... I now have (yet again) 8 CRS that are useless
upgraded firmware, and i always "system reset-configuration no-default=yes"
Did you reset to defaults after upgrade to 6.10 or 6.11? ...
-Eric
Yeah, that was my discovery too - Trunk is leaking somehow, like I said, if you disconnect the trunk the problem isnt there (no vlan <-> vlan leak on same CRS).@trn76: using the same tutorial I have discovered that the access ports are still receiving tagged traffic from the trunk port. You can verify that for yourself by running a tcpdump -nei enX on the "access" port, you will be able to see the VLAN IDs affected as well.
I think that the approach described in the configuration example is far away from what is expected from a basic L2 switch, and I hope that this issue will be addressed soon by the dev team.
I just grabbed the latest beta which supposedly has a lot of fixes for the CRS. You may want to test that one.Yeah, that was my discovery too - Trunk is leaking somehow, like I said, if you disconnect the trunk the problem isnt there (no vlan <-> vlan leak on same CRS).@trn76: using the same tutorial I have discovered that the access ports are still receiving tagged traffic from the trunk port. You can verify that for yourself by running a tcpdump -nei enX on the "access" port, you will be able to see the VLAN IDs affected as well.
I think that the approach described in the configuration example is far away from what is expected from a basic L2 switch, and I hope that this issue will be addressed soon by the dev team.
I've tried changing all kind of parameters, and no go.
I guess that is due to the fact that the proposed approach is not a real VLAN encapsulation/decapsulation but a VLAN translation. I would expect a L2 switch to behave in a different manner, but the product is quite young too - perhaps we will see improvement soon.Yeah, that was my discovery too - Trunk is leaking somehow, like I said, if you disconnect the trunk the problem isnt there (no vlan <-> vlan leak on same CRS).
I've tried changing all kind of parameters, and no go.
There are always arguments about whether VLAN tagging or detagging is a decapsulation/encapsulation process. Cisco say it is and technically there is a good argument for saying that it is since the frame and FCS change.I guess that is due to the fact that the proposed approach is not a real VLAN encapsulation/decapsulation but a VLAN translation.
The VLAN table and such are getting fixed in 6.12... We'll have to wait until the documentation gets updated to see exactly how it all works.Agreed. But in the current CRS documentation the tagging/untagging is just a basic process of VLAN translation:
- Trunk port VID xxx mapped to Access port VID 0
- Access port VID 0 mapped to trunk port VID xxx
That allows all other tagged VIDs to be copied on the access port, and I think that's not what you'll expect at this level.
I think all bonding/LACP uses the CPU.....is this a right place to ask if CRS is capable of trunk group (aka bondig) with LACP (or alike)?
if so, how?
Please put them on ebay and PM me your name on there. I'll buy them off of you (depending on how many you have). If no ebay, let me know.Seriously, these switches should never have been released. We've had a few for 6 months now, and they've never even been capable of functioning as basic switches. They're not even useful as paper weights because they aren't heavy enough.
We're throwing ours out, and we're def not going to try Mikrotik switches again or recommend them to any clients. It would have been way cheaper from the start to buy Juniper or Cisco considering the massive amount of time wasted on these...
No documentation, no functionality, no support. :/
The first ones were sold as pre-production test units, that was made clear. But now with software upgrades they have been made fully functional. Did you see our latest newsletter? It clarifies all the new features we have added to CRS: http://download2.mikrotik.com/news_58.pdfSeriously, these switches should never have been released. We've had a few for 6 months now, and they've never even been capable of functioning as basic switches. They're not even useful as paper weights because they aren't heavy enough.
We're throwing ours out, and we're def not going to try Mikrotik switches again or recommend them to any clients. It would have been way cheaper from the start to buy Juniper or Cisco considering the massive amount of time wasted on these...
No documentation, no functionality, no support. :/
Normis: Why are the ACLs only on the 226?The first ones were sold as pre-production test units, that was made clear. But now with software upgrades they have been made fully functional. Did you see our latest newsletter? It clarifies all the new features we have added to CRS: http://download2.mikrotik.com/news_58.pdfSeriously, these switches should never have been released. We've had a few for 6 months now, and they've never even been capable of functioning as basic switches. They're not even useful as paper weights because they aren't heavy enough.
We're throwing ours out, and we're def not going to try Mikrotik switches again or recommend them to any clients. It would have been way cheaper from the start to buy Juniper or Cisco considering the massive amount of time wasted on these...
No documentation, no functionality, no support. :/
I solved it by circle the switches an putting in two cisco switches in the circle that can do STP, no loops all looking good so far.As far as I know, you can't use any STP on a CRS unless you use software bridges, which you for sure don't wan't to.
Maybe this can help you:Is anyone doing a bit more with a CRS than what you would do with a managed switch ?
I'm trying something very basic like putting a DHCP client on a VLAN and don't get it working because the CRS is not untagging traffic on the ports (which is just a simple concept of an access port).
If someone is willing to have a look at this issue... it is discussed here http://forum.mikrotik.com/viewtopic.php?f=13&t=89595 (sorry it took some time to notice that my PC wasn't capturing VLAN info).
I'm rather interested if any CRS can tunnel VLANs in QinQ-VLANs at line-speed, including 10G for the SFP+ models.does anyone has tested QinQ application using this CRS?
/interface ethernet switch egress-vlan-tag
add tagged-ports=ether8 vlan-id=400
I haven't found any evidence of Spanning Tree except for bridging.As far as I know, you can't use any STP on a CRS unless you use software bridges, which you for sure don't wan't to.
i think the confusion started with marketing statements from mikrotik about CRS series establishing it as a layer 3 switch.
All I wanted to do is simple queue to rate limit bandwidth by port. For eg, rate limit apply for port 80 and 443 only. The reset will have unlimited bandwidth. Right now I cannot even get any packets into the queue.
Also, in CRS docs, there are mention of 3 types of queue. But it is not expanded on how these queue works and whats the difference between queue in CRS and in simple queue/tree queue.
sorry no stp support on crsHello Folks!
I need to put up CRS125 switch in a datacenter to replace a bunch of cisco2960 switches, they are connected in a circle's and some other in meshes.
How do I activate spanning tree in them ?
Note: Multiple master-port configuration is designed as fast and simple port isolation solution, but it limits part of VLAN functionality supported by CRS switch-chip. For advanced configurations use one master-port within CRS switch chip for all ports, configure VLANs and isolate port groups with port isolation profile configuration.
Please provide some clarification on the following statement.
Note: Multiple master-port configuration is designed as fast and simple port isolation solution, but it limits part of VLAN functionality supported by CRS switch-chip. For advanced configurations use one master-port within CRS switch chip for all ports, configure VLANs and isolate port groups with port isolation profile configuration.
Using multiple master ports will create port isolation groups automatically (marked as default), so I neither understand why is this bad. I was experimenting much and it seems there are still some bugs here - sometimes adding and removing will mess up and reboot is needed. Documentation should clarify more details.short answer: don't use multiple master ports
i think mikrotik have to take more seriously the CRS topicAll,
The Cloud Router Switch is very nice. Being able to do mac based vlans or protocol vlans is an awesome feature that requires a certain level of complexity. However, the vlan implementation when trying to do simple tagging and untagging or trunk ports is, well, just the worst. Even with examples shown in documents it is challenging and tedious at best and near impossible at worst. It would be very nice to have a wizard to assist with the vlans. RouterOS makes vlans much easier to understand but I have to build bridges which means I can not get non-blocking throughput because bridges use the CPU.
I have to admit that I avoid using mikrotik switches in situations that require even the most basic vlan configuration due to how complicated it is. It would be very nice to get a wizard or a simplified menu.
Thanks.
can you post the errmsg on the cisco side?i am receiving VLAN1 flapping warning msg in Cisco switch and can't ping managment trunk ip address of vlan1.