OS v5.0 - VLAN bugs - No GO

Hi,

My 450G is fully operationnal with 4.17.

This are 3 problems that prevent my board to be operationnal on 5.0:

  1. cant modify switch-all-ports, answer is “mirror-source does not belong to switch”, and “none” isnt recognized, so that if you define a valid interface there, you wont be able to “undo”. Bugs not there on v4.17.
  2. pppoe-client (over a vlan on an ether1 non switched interface) isnt working anymore. I get this during past hour:
02:20:00 pppoe,ppp,info pppoe-fibre: initializing...
02:20:00 pppoe,ppp,info pppoe-fibre: dialing...
02:20:03 pppoe,ppp,info pppoe-fibre: terminating... - disconnected
02:20:03 pppoe,ppp,info pppoe-fibre: disconnected
  1. vlan secure mode is quite borked: vlan-id 0 has to be set, ok I’ve seen this on wiki, but the cpu port isn’t recognized anymore as a valid port for declaring vlan 0:
set ports=cpu numbers=0
port does not belong to switch

It’s not a problem unless you have a master port in your interfaces list… which is my case of course.
Even with cpu port declared (like when migrating from 4.17 where it was operationnal), ports concerned by master-port setup aren’t operationnal.
What confirms that cpu port is involved and not taken in account is this:

interface ethernet switch host print
Flags: D - dynamic, I - invalid
 #   SWITCH                                              MAC-ADDRESS       PORTS                                              TIMEOUT DROP MIRROR VLAN-ID
 0 D switch1                                             3E:2D:A1:BC:94:DA ether3-LAN                                         2m30s   no   no
 1 D switch1                                             00:0C:42:7F:40:75 (unknown)                                          2m      no   no

5.0 went stable, ahem…
Bug 1 was there since last RC (which was the first I tried back then).

I revert to v4.17.

Well, after thinking a bit, there may not be any pppoe-client related problem, if it’s simply VLAN management by cpu port which is deficient.
Both problems may have the same cause.

bug has been mentioned here also: http://forum.mikrotik.com/t/routerboard-documentation-and-mikrotik/45243/1

I’m sure there will be a fix. Any time frame on this?

We are looking into these problems. Thank you for the report.

Hello!

Krishteins: i received a mail from you about 3 weeks ago with same content. What is the state of this bug?

This bug is fixed. Currently it needs to be tested internally. This issue will be resolved in v5.1.

Thanks for this clear status report.

Should I open a separate thread for the switch bug on “switch-all-ports=yes/no” command?

We are aware of this bug as well. It will be fixed in next release.

Maybe my problem is related: Since 5.0 on RB750G I can’t add static MAC to switch port (Switch - Host - New Switch Host), neither modify previously set hosts, this error appears: Couldn’t add New Switch Host - vlan-id not supported for this switch (6)

I don’t use any VLAN. Could you check it please?

Thank your for the report. It seems like Winbox bug. Works fine from terminal.

No reference to the 2 bugs mentioned in my OP in 5.1 changelog, is it normal?
Fixes were postponed to 5.2?

those bugs should be fixed in v5.1

WARNING! Please DO NOT install 5.1 if you have Atheros-8316 switch chip in use!
Board ssh communication stopped immediately after you type “interface ethernet switch p” (not able to write after to terminal) (issued on p as port or print, and h as host)
(winbox works)

Adding static MAC to switch port using WinBox now works in 5.1, thanks!

Does this SSH problem only occurs during update or does it persist after a reboot or more of the router?
I’m not too concerned about webfig being broken on 5.1 (I only use CLI), but if working with CLI through SSH is not reliable, I still can’t reasonnably upgrade from rOS 4.17 :confused:

Is the VLAN-bug fixed now? I see no mention of it in the ROS 5.1 changelog.





I cannot reproduce this problem. Is there any addition actions to trigger this problem. What is the configuration in those sections?

I have this same problem (and more with switch chip)
Ticket#2011041166000202 (with supout.rif from 5rc10 - last working for me)

Are you serious? I reset board configuration to default to check this problem exists. The result: problem exists if board in defconfig. How to prove it?