I have one stack of S6000-ON switches and one stack of S6010-ON which experience the same issues. I have multiple LACP port channels active, and on the S6000-ON stack there is one host experiencing issues where the host cannot be reached by other hosts on the same stack, except when I shut down the Fo 2/26 interface (when I shut the Fo 1/26 interface it is still broken).
On the S6010-ON stack this was even worse, that why we have migrated all the hosts to the S6000-ON to get a replacement for the S6010-ON.
When I destroy the Port channel the host is on and configure the interfaces as "separate" interfaces on the switch, everything works perfectly.
We have tried swapping the cables for new, swapping the cables from Fo 1/26 to Fo 2/26 and vice versa, updating the NIC firmware on the host, used other interfaces in the same port channel (Fo 1/28 and Fo 2/28), nothing helps. Only when I shutdown one specific interface of the port channel, everything starts working.
We have tried creating a static LAG as well, and that didn't help either.
We see no port errors or discards.
I don't understand it anymore. The S6000-ON stack is running firmware 9.14(2.7) The S6010-ON stack is running firmware 9.14(2.8)
As mentioned before, the S6010-ON stack has had issues with creating a LACP Port channel to an S6000 (non ON) stack as well, and with a LACP port channel to the S6000-ON stack as well. Now we have migrated everything towards the S6000-ON the problems have become less, but now it's only one host experiencing issues and only with the port channel interface (each host has 4 interfaces to the stack, 2 of the 4 interfaces are in a port channel).
Any ideas? Anyone else has seen this behaviour before? It looks like the Port channel just stops accepting/forwarding traffic when I enable a second interface in the port channel.
There is no spanning tree active on the storage network (as it’s supposed to be).
There are 2 interfaces in the port channel: 1/26 and 2/26. We have tried adding 1/28 and 2/28 and that doesn’t change or resolve the issue.
As mentioned, we have had similar issues on the S6010-ON stack, but then across multiple hosts.
Also we have found that some hosts cán communicate with the affected host on the affected address. Also we have found that it doesn’t affect all IP addresses on the host (it has multiple addresses on the same port channel, in the same VLAN and IP segment).
Josh is out today, but it looks like he should be back in tomorrow. We don't have access to each other's PMs, so if you'd like to send me a copy of the PM, I can use that to help make sure the log information is available.
DELL-Josh Cr
Community Manager
•
9801 Posts
•
45350 Points
924
0
Posted December 14th, 2020 13:00
Hi bladman87,
Are there any spanning-tree errors? How many ports are in your port channel? What happens if you stop using 1/26 and 2/26 in the port channel?