RB750Gr3, CRS326 and RB3011 upgraded with no problems (Routerboard firmware also upgraded)
Upgraded CRS317 to 6.41 from rc61. Seemed to be no problems with its upgrade.
I also have a few RB’s and CRS as switches connected to it which I left at 6.40.5.
A little while later my CRS226 @ 6.40.5 stopped responding properly which hasn’t happened in years. Traffic wasn’t being passed properly and a Winbox connection to it would not show interfaces/etc and would disconnect in a minute or so. Rebooted it before Winbox disconnected me all seems to be fine now.
CRS226 has no bridge configuration, all ports master of Ether1, multiple vlans from the CRS317 handled via switch menu with only one internal management vlan off of ether1.
Were user name or password reverted back to default? I cannot visit via web ui or winbox.
Thanks.
By upgrading from 6.40.5, will it automatically and intelligently add the correct rules to switch all the switch-related configurations to bridge ones?
From what I’m seeing, for most older hardware switch vlan configurations the switch menu and settings are still used. Just the Master-Port changes to bridge setups and the vlan parent interface changes from master-port to the bridge. Sometimes the upgrade has to create a new bridge, sometime it tries to convert old bridges. In my case (overblown for fun home setup) an old bridge on my RB2011 didn’t convert cleanly, but most other systems seem to convert fine with hardware acceleration.
My CRS317 which uses the new vlan filtering via bridge menu was already on RC and also upgraded fine.
In total I’ve updated a CRS317, CRS226, CRS-125-2HnD, RB750GL, CCR1009 (older switch model), a couple WaP AC’s and the RB2011.
Only the RB2011 gave me real issues and it seemed to be an invalid configuration issue with a port on the bridge showing “Unknown” bridge. (Was a fairly new configuration so don’t know how that happened). Once I restored to previous version, I fixed the issue, upgraded again and everything went well.
I have multiple trunked vlans on most devices and was expecting more issues, so this went surprising well.
How the conversion works when there are two switches in the device and both are in the common bridge? What if there are multiple switch groups within one switch differently bridged with other interfaces?
My RB2011 (I’m using as a switch at the moment) with two switch chips seems to convert and work fine with just one created bridge, all ports show hardware acceleration.
I believe that two groups in the same switch (not counting individually separated ports) will disable hardware acceleration on at least all but one group, forcing all other traffic to flow thru cpu.
I have thought of this to and if I may speculate:
The new bridge per say will use hardware offload if possible. Add bridge, add all ports from both switch chips then there would be one of these 3 outcomes.
- Nothing is hardware assisted software only.
- Hardware per chip is activated and should be vissible on interface and inter swich chip should be software.
- Hardware all the way (how that would be possible looking at block design of rb’s
Pre 41 with master and slave ports and software bridge, the two master ports made the 2 scenario so why not now…
As per now i recon that 1 is the option and would advocate for a refined implementation towards 2. But if it allready is 2 I lift my hat and say good work MT.
I Posted this question in the 41RC channel but I did not get an answer:
Now Looking at the released version of 6.41 of RouterOS.
if i set:
/interface bridge port
add bridge=bridge1 frame-types=admit-only-untagged-and-priority-tagged interface=ether1 pvid=64
And then look in the switch menu:
The setting is there to and is not reflecting the change. Here there is a change from last rc cli is not outputting the print (winbox does):
[admin@MikroTik] /interface bridge port> /interface ethernet switch port print
Flags: I - invalid
# NAME SWITCH VLAN-MODE VLAN-HEADER DEFAULT-VLAN-ID INGRESS-RATE EGRESS-RATE
0 ether1 switch1 100.0Mbps 250.0Mbps
Winbox switch menu tells me:
Vlanmode=disabled
VlanHeader=leave as is
Default VLAN ID= 0
In RC the cli printed this at print as well but not now.
My questions still are:
Is settings in bridge overriding settings in swith?
Do I need to set both to be sure?
Are there any corner cases where ingres frames with vlan tag would be allowed?
Are there any corner cases when the native untaged vlan would be assumed to be 0 instead of the bridge configured 64
In my case nothing went to default, just new bridge implemented instead of master-ports. Before upgrade I always reboot unit’s (I’m doing it from about a month and I don’t have issues at all, before I didn’t reboot and sometimes I had issues similar like some people writing on forum,like no IP jus login thru MAC address or i loop etc.), then upgrade to RouterOS I want and after that upgrade Routerboard firmware.
This is a complex question:
As of now there is NO hardware support for LACP in 6.41 as far as I know. (Bonding without a protocol is bound to give you problem down the road)
That being said bond interface in RouterOS will and have always been software. Using CRS with weak cpu’s compared to number of avail interfaces this is a problem.
Using a Router Board ROUTER ie CCR this is less of a problem in this case we often go from l3 to l2 domain and are in cpu anyway, and there are plenty of them in the ccr’s.
Looking at your setup The new setup would be One bridge with several Vlan’s. Look at the Router Board block diagram how everything is hooked internally to get a picture of what you are trying to do with the hardware if it is att all possible.
Read the text above with an open mind: Software can do in principal anything but depending on the hardware. If it is software the performance is from nothing to something. Depending on what you are trying to do there may be surprises.
After the conversion on my CCR1009, the DHCP server failed to work if connected to a bridge interface (it worked dough on a single vlan interface). For static IP hosts, everything seemed running normal.
I traced this back to the fact that STP/RSTP was not enabled. After enabling, it worked as expected.
It seems that afterwards, disabling SSTP keeps it running properly.
It was kind of “wise” to post this version right before holidays, with no support (even in expect of huge problems), with no smooth way of conversion.
Nice done!
(Hope noone set RB to upgrade authomatically?)
Just upgrade my RB2011UiAS-2HnD, all went smooth, no issues so far.
No problem here upgraded firmware also.
RB1100ahx2,
ccr1009-8G-1S-1S+, Runs Capsman
RB962UiGS-5HacT2Hnt,
RB750gr3, RB750GL,
RB912UAG-2HPNnD,
CRS12524G-1S.
I love the releases on Friday and big weekends so i can play with the new releases on the weekends and then upgrade customers routers later.
Merry Christmas
Heavily depends on configuration 
Hello!
Updated SXT LTE (3 item) - updated normally, but firmware update 6.41 did not come. There are a couple of other devices out there was an update from 3.41 to 6.41, is this normal?
*) firewall - added “tls-host” firewall matcher;
any documents?
Hello!
Home RB951G-2HnD with simple setup was upgraded to 6.41 from 6.40.5.
Bridge1 was created. DHCP server moved to created bridge1. Firmware was upgraded manually to the 6.41 with additional reboot.
All working fine.
Thank you!
Update to 6.41 bricked my RB750Gr3. In goes into continues boot loop and it is undiscoverable to even perform netinstall. 