LAGG improvement opportunity

[Constantin]

Good morning,

I have a suggestion for the failover LAGG feature to address what I would consider a bug.

Specifically, my 10GBe interface from iXsystems (i.e. private-labeled Chelsio 520) is currently not functioning after receiving/installing a replacement motherboard for the mini XL (AVR54 bug).

The 10GBe interface is part of a failover LAGG, with its two SFP+ ports being assigned priority 0 and 1 respectively. Using ifconfig at the console, the card is shown as up, the activity lights are green, the OS acknowledges that a twinax is inserted or removed, etc. Yet when I try to ping the gateway, I get 100% packet loss. Remove the TwinAX cable and IGB0 takes over as intended - everything suddenly works.

As best as I can tell, the LAGG doesn't confirm that the interface actually functions, it relies solely on an OK ifconfig status.

I would consider this a bug, not a feature. Instead of solely relying on ifconfig to tell the truth, I would like the failover LAGG code to confirm at build time that the LAGG actually has connectivity via the current interface. If there is no traffic on the LAGG for an extended period of time, it would be great for the LAGG to periodically ping the gateway to confirm whether there is a path to freedom or not.

I would appreciate at the end of the LAGG building process a confirmation message that confirms what is part of the LAGG, whether it's UP or down, and whether it managed to ping the gateway, a name server, etc.

 

[SweetAndLow]

Make a feature request on the bug tracker. The forums are not for feature request.

 

[Constantin]

The failover LAGG actually not functioning as a failover LAGG is not bug?

Isn't detecting a connection issue the very reason that people use a failover LAGG in the first place?

 

[SweetAndLow]

Ok then file a bug it's basically the same thing. And fail over works. I have it setup up and working right now. Has been for years.

You made it sound like you just wanted a way to make the configuration more idiot proof. So that's why I thought a feature request was best.

 

[Constantin]

Bug report has been filed.

The failover LAGG worked in the past for me also.... when 11.0 U4 came about it stopped working but the Chelsio card still worked on its own (without a failover LAGG). Post motherboard-swap the chelsio no longer works at all even though the FreeNAS seems to think it is 100% functional.

The silent failure of a interface combined with FreeNAS 11.1 STABLE not detecting the loss of connectivity and switching to a working interface makes for an unhappy failover LAGG user experience. Doesn't really have to do with idiot-proofing. It appears to be a basic functionality failure in the failover LAGG implementation.