Hi

I have 2 of these that I have connected to a stack of 2 x dell 8024f

I have 10G connectivity

I went to LACP bond the interfaces and it wouldn't work, couldn't get LACP active on the switch so I have fallen back to balanced-tlb.

balanced-alb seemed to be failing for half the clients...

so this was on 6.3, I have upgraded to 6.10 was there any LACP fixs. any one else having issues.

Also as a side note for bonded interfaces using balanced do I need to do anything special with the interface on the switch, current have them setup in a port-channel ... ??

Fixed

re tried with 6.10 and all's good

Hi,

I sill have the described issue. An upgrade to 6.10 dit not fix it.
On the CCR1036-8G-2S+ I have bonded both SFP+ ports and connected it to a Cisco 4500X.
I have two CCR that are running without problems and also two that did not work.

This is the log of the Cisco:
*Feb 21 07:11:03.537: %EC-5-UNBUNDLE: Interface Te2/1/8 left the port-channel Po3
*Feb 21 07:11:03.652: %EC-5-UNBUNDLE: STANDBY:Interface Te2/1/8 left the port-channel Po3
*Feb 21 07:11:27.564: %EC-5-BUNDLE: Interface Te1/1/7 joined port-channel Po3
*Feb 21 07:11:27.723: %EC-5-BUNDLE: STANDBY:Interface Te1/1/7 joined port-channel Po3
*Feb 21 07:11:33.531: %EC-5-UNBUNDLE: Interface Te1/1/7 left the port-channel Po3
*Feb 21 07:11:33.650: %EC-5-UNBUNDLE: STANDBY:Interface Te1/1/7 left the port-channel Po3

Cisco Port 1/1/7 is connected to SFP+1
Cisco Port 2/1/8 is connected to SFP+2

I have also seen very strange LED working on the CCR1036-8G-2S+, the LEDs of the SFP+1 and SFP+2
only worked, when Ether8 was connected. It is very strange and I think there is still a software or
hardware problem on this ports.

I haven't used ether 2-8... Ether 1 I have cross connected.

Have you done any throughput tests ?

How to make throughput tests, when the ports leaves and enters into the LACP channel?
Also a ping is not working.

It is very strange that two CCR1036-8G-2S+ does not have this problem and other two
have it.

How to make throughput tests, when the ports leaves and enters into the LACP channel?
Also a ping is not working.

#1 Using iPerf from PC to PC or server to server or PC to server
#2 Using a server stress tool like this one Webserver stress tool but please be careful this tool is very powerful and is also able to freeze hole networks!!!
#3 Using NetIO for a test like described under #1

[quote="iandric"It is very strange that two CCR1036-8G-2S+ does not have this problem and other two
have it.[/quote]
Can it be that this phenomenon depends on different configurations?

Can it be that this phenomenon depends on different configurations?

No, the configuration and software version is identical. I will now clear the configuration and
also install the software again. But I will take the 6.6 or 6.7, the 6.10 have realy some problems/bugs.

Can it be that this phenomenon depends on different configurations?

No, the configuration and software version is identical. I will now clear the configuration and
also install the software again. But I will take the 6.6 or 6.7, the 6.10 have realy some problems/bugs.

Hm this is not really fine as you tell us here now some stuff about the LAG (LACP)
I wanted to buy also some of the models and there fore I am now really hard thinking
about!!! At first I was only thinking that the CCR1036 is not able to handle the load
on the SFP+ ports coming from outside at the WAN interface and there fore a
CCR1072 could or should be better to handle this load.

It is very strange that two CCR1036-8G-2S+ does not have this problem
and other two have it.

Surely I would now also wondering why?
Did you set up the LAG with 802.ad (LACP) on both ends?

Did you set up the LAG with 802.ad (LACP) on both ends?

Yes, but now I have give up and have re-configured it as acive/standby bonding on the MikroTik.
10Gig is enough bandwith in the moment, I wanted to have an interface backup based on the channel
and to use both. I will try it again to make such a configuration but have no time now for troubleshooting...

Have you done any performance testing... I haven't been able to push a single stream of TCP faster that 1Gb...

Have you done any performance testing... I haven't been able to push a single stream of TCP faster that 1Gb...

That is to be expected. If you want more make sure you do more simultaneous sessions, every session with a different source and/or destination IP.