I have have one M1000e chassis connected to two Cisco 5548's using the B22 FEX modules. All the blades are using NPAR with FCOE for storage connectivity. The chassis has 7 M630's with Emulex OCE 14000 adapters and 3 M640's with 57810 adapters.
Everything worked fine connected to the Cisco 5548's but ours are near end of life. We moved them to the new Cisco 93180YC-FX switces.
The M630's/Emulex servers work fine.
The 3 M640's 57810 servers have high latency to the storage and eventually the switch will disable the ports with a dcxNoAck state.
The 3 M640s are ESXi 7.0u3 w/ latest Dell firmware, ESXi 6.5 with 2 year old firmware, and RedHat Linux with 2 year old firmware. Since I have a mix of OS's and firmware, I don't think it is a firmware/driver issue since all three do the same thing. It is likely a switch setting or HBA setting but there are not that many settings on the HBA.
To sum it it, we moved from a Cisco 5548 to 93180 and now the M640/57810's don't work. Any ideas what could cause this behavior?
Hi, thanks for your question but I'm afraid this would be better addressed by Cisco. Sorry I can't be of help this time. Maybe we could wait around to see what other users could offer. Wish you a good one.
Social Media and Communities Professional
Dell Technologies | Enterprise Support Services
#IWork4Dell
Did I answer your query? Please click on ‘Mark as Accepted Answer’. ‘Thumbs up’ the posts you like!
We actually have been working with Cisco and still have an open case on it but unfortunately they have not yet been able to find a cause and resolution.
This is old but I still have this chassis running with 1 2012 server in it. Turns out Cisco doesn't support the B22 FEX modules connected to the 93180YC-FX switches. The B22 are so old, so it is likely they are not supported on any switch other than the Cisco Nexus 55xx series. That chassis originally started out with Dell IO Aggregator module so I ended up installing that back in. Has worked fine since.
I do still have some B22's connected to Cisco Nexus 55xx switches and will just keep them that way until I can get the old servers out.
VMware’s software FCoE stack (like many BCM57xxx/“FastLinQ 578xx”) can behave unpredictably after an upgrade including PSODs. Becuase Software FCoE was deprecated in vSphere 7 and removed in vSphere 8. ESXi 8 also drops support for several legacy I/O devices/drivers, so mixing older firmware/driver combos (bnx2x/bnx2fc-era gear) with 8.x can lead to link flaps or PSODs
Hope that helps!
DELL-Erman O
Social Media and Communities Professional
Dell Technologies | Enterprise Support Services
#IWork4Dell
Did I answer your query? Please click on ‘Mark as Accepted Answer’. ‘Thumbs up’ the posts you like!
I think it could be related NIC functionality. Although the BCM57810 still appears in the VMware Compatibility Guide (VCG), this listing only confirms support for its Ethernet (NIC) functionality on ESXi 8.x. It does not indicate that FCoE is supported.
VMware removed Software FCoE entirely in ESXi 8.0, meaning any adapter that relied on VMware’s software stack (such as the BCM57810, which does not provide hardware FCoE offload) can no longer perform FCoE operations. This is what I concluded from my research. Of course, you can also consult with VMware.
DELL-Erman O
Social Media and Communities Professional
Dell Technologies | Enterprise Support Services
#IWork4Dell
Did I answer your query? Please click on ‘Mark as Accepted Answer’. ‘Thumbs up’ the posts you like!
DELL-Young E
Moderator
•
5421 Posts
•
37 Points
363
0
Posted September 20th, 2022 20:00
Hi, thanks for your question but I'm afraid this would be better addressed by Cisco. Sorry I can't be of help this time. Maybe we could wait around to see what other users could offer. Wish you a good one.
Social Media and Communities Professional
Dell Technologies | Enterprise Support Services
#IWork4Dell
Did I answer your query? Please click on ‘Mark as Accepted Answer’. ‘Thumbs up’ the posts you like!