SOLVED Troubles with lagg failover not recovering

[andrewt81]

Good morning to all the community,
i'm trying to setup failover between 2 NICs for management purposes.

My Configuration:
Asus P9D-X
Xeon E3-1230v3
16Gb Ram
Intel i218 embedded Dual Gbe NIC
Broadcom BCM Dual 1Gbe
FreeNAS-11.3-U5

For redundancy i need 2 ports connected to 2 different switch (both GS1900-24), if a switch (or a NIC) will fail connectivity must be provided throu the other NIC.
I followed the guides but it seems to not working. Strange thing: when selecting the interfaces aer marked to be members of lagg0 interface but it did not exist, never created before to my memory and i cannot find how to clean this fake interface.

Steps for configuration:
1) I removed IP configuration for both interfaces (igb1 + bge1) so they are available for lagg
2) created lagg interface (name lagg888) with igb1 and bge1 interface (settings attached) and IP configured
3) lagg failover seems to not work (pinging the IP 10.10.10.112 is ok but fails and not recovering if i disconnect the main NIC.

I already has several attempts with vlan configuration (untagged on these specific ports) on switches and also switched both to the same switch to exclude misconfigurations but always same result. Ping fails if main interface is disconnected.

Feel free to ask any other useful information you need.

 

[andrewt81]

i attach interfaces screen

 

[andrewt81]

Update:
my mistake, "member of lagg0" was just interface description, this is not a "fake interface" as stated before.

 

[andrewt81]

Any help to troubleshoot this problem?

 

[andrewt81]

none in this community has skill enough to help me?

 

[Samuel Tai]

Simplify things first. Try the LAGG with both links on the same switch. If it works then, you have a spanning tree issue between the 2 switches.

 

[jgreco]

Yes, lots of us have "skill enough to help" you, but being insulting isn't going to convince anyone to do so, and it's unlikely anyone with "skill enough to help" you has the same unusual hardware setup, which means that your minimalist description is all we have to go on.

The first time I noticed your post, the only comment that came to mind was that you should try using *only* the igb interfaces for a LAG group. There used to be issues with mixing and matching interface types. It's possible that there could still be.

It would be helpful if you would actually try to debug your problem to see if you can come up with a specific set of events that is problematic. This is much more likely to result in someone being able to have an "aha" moment, since we can't experiment on your system directly.

 

[andrewt81]

This is not insulting.. Since it's not a paid support it could be possible.. It happens everyday on technet forums, better the silence then useless replyes :)

But your reply was really useful.. checking again my notes i noticed that only this scenarios were tested:
- igb1+bge1
- bge0+bge1
- connected on same switch (Zyxel)
- connected on 2 different switch (both Zyxel)
- connected to another switch brand (HP)

So the scenario igb0+igb1 was not tested since igb0 was tied to webconfigurator, as you pointed out both interfaces in lagg on intel adapter was not tested, i'll do further investigation when offices are empty.

Thanks for your help

 

[andrewt81]

A bit late but i was able to test the combo igb0+igb1 and i can confirm the problem was related to the Broadcom NICs (mixed with intel and both broadcom),
with 2 Intel adapters failover worked immediatly without any kind of hiccup..

Solved for me, thanks for your help

 

[jgreco]

A bit late but i was able to test the combo igb0+igb1 and i can confirm the problem was related to the Broadcom NICs (mixed with intel and both broadcom),
with 2 Intel adapters failover worked immediatly without any kind of hiccup..

Solved for me, thanks for your help

No problem. The Broadcoms are known to be problematic in various ways, alas.