I need some help about VSAN/VLAN configuration in UCS 2.0 (1s).
Previosly i used to create VLAN dedicated to FCoE traffic, and then create VSAN and map it to previosly defined VLAN to carry trafic. And that works just like it should.
This morning i have upgraded one UCS to 2.0 (1s)
And errors with vsan/vlan configuration begins.
Before update i had created vlans 40 and 41 dedicated to fcoe, and two vsans 10 (mapped to 40) and 20 (mapped to 41).
Erros were as follows :
descr="VLAN default is error-misconfigured because of conflicting vlan-id with an fcoe-vlan"
Code : F0833
Just to continue (browser crashed )
I have deleted vsans/vlans (system is being installed it is not production yet), and tried to recreate, but error occured stating that i cannot map vsan to vlan 40 because it is already assigned to vlan40??
I tried this on my other 1.4 system, it works fine.
After some time i have decided to create vsan, and map it to vlan WITHOUT previously creating matching Vlan.
Now everything seems to be working so far.
Is it possible that matching vlan is automatically created during vsan creation?
If so, why that vlans cannot be listed in LAN manager? This is pretty confusing and can lead to problems.
Tnx in advance/
The information is listed here.
CSCtr74091 (for version 2.0 and all future releases when upgrading from any 1.x UCS release)
Headline: Overlapping FCOE VLAN and Ethernet VLANs no longer allowed with UCS 2.0
Description: In 1.4 and earlier releases, overlapping ids were allowed for Ethernet VLANs and FCOE VLANs. Starting with UCS release 2.0, this is not allowed and a critical fault will be raised on upgrade if overlapping ids exist. Users should ensure that the VLAN ids are not overlapping before attempting the upgrade.
When the Vlan is listed in the LAN then it is defined ethernet VLAN ids
I hope this helps
Thanks a lot.
So proper way is to create VSAN and map it to some VLAN not taken, and it shall work? I have done this, create vsans 10 and 20 and just "map" them to non-defined vlans (40 and 41) and no errors are showing, but i am waiting for storage guys to complete setup to verify.
If this is the case, it is really stupid that vlan IS's assigned to vsans while creating vsans are not showed in LAN manager.
I can't comment on the logic (ref to your last sentence) but here is another way to look at it. I guess one way to look at it is there are SAN ids and then there are LAN ids under 2 different tabs for a SAN admin and a LAN admin.
FCoE VLANs in the SAN cloud and VLANs in the LAN cloud must have different IDs. Using the same ID for an FCoE VLAN in a VSAN and a VLAN results in a critical fault and traffic disruption for all vNICs and uplink ports using that FCoE VLAN. Ethernet traffic is dropped on any VLAN which has an ID that overlaps with an FCoE VLAN ID.
Here's a bit of goofiness. The Default VSAN 1 is required to map to an FCoE ID, you cannot leave it blank. The default is 1, which overlaps with a defined VLAN, so you get an error. You have to go into VSAN 1 and give it a different number for the FCoE mapping (any number you do NOT care about), and the error will clear.
On this same note, I'm upgrading a customer system that just used vlan 1 and vsan 1 mapped together.
What would be the impact of changing the FCoE Vlan in a production environment?
I'm guessing it to be huge, but does anybody have an insight? Machines SAN Storage loss? Total crash?
I'm getting this same error with vlan 1, and as I'm not using vlan 1 anywhere else on the system I'm assuming that it just cannot be deleted (as with any other switch), so until we change the FCOE vlan id... this error will continue...
Thanks for any insight.
Maybe best thing is to power down hosts, reconfigure vlan and vsan mapping and reboot hosts.
I mean if hosts are not Boot from san, you will have only short SAN loss, but in order not to loose data or risk for some problems, maybe it is safer to stop SAN trafic or power down hosts.
Personally, i havent done that just upgrade, change vlan/vsan mapping, everything went OK.
What type of FC switch are the Fabric Interconnects uplinked to? If it is not Cisco then you can create a new VSAN for Fabric A and then change the uplinks and vHBAs in Fabric A to this new VSAN. After Fabric A is done do Fabric B to a different VSAN/FCoE vlan. As long as host SAN multipathing is configured properly then there shouldn't be an outage.
If the uplink switch is Cisco then you will need to take an outage because if you change the FCoE VLAN the vHBAs will logout of both FC fabrics and then log back in. This should only take about 30 seconds though.
The other way is to change each Cisco FC switch to something besides VSAN 1, this can be easily scirpted with a very minor outage and can be done one fabric at a time. VSAN 1 should never be used for production hosts/storage. It is as bad as using VLAN 1 on the Ethernet side.
If your northbound switch isn't Cisco you can do this one Fabric at a time. As long as host multipathing is configured properly then there will not be an outage.
Create a new VSAN for Fabric A
VSAN 20 - FCoE VLAN 920 (make sure this vlan isn't in use, if it is use another non existent VLAN)
For Fabric B
VSAN 40 - FCoE VLAN 940 (make sure this vlan isn't in use, if it is use another non existent VLAN)
the Upstream FC switches are not Cisco. For what I'm seeing even though marked as a Major error, in this particular situation it is not applied. I will consider your suggestions with the customer. Thanks for your help.
Did anyone determine if chaning the fcoe-vlan is a disruptive process?
I am using FC zoning on the FI 6120s
VLAN default is error-misconfigured because of conflicting vlan-id with an fcoe-vlan
Changing fcoe vlan is non-disruptive if you are not using the current fcoe vlan for any storage access.
Since you mentioned, you are using FC zoning, I believe you are not using fcoe. In that case, you are safe to put in any vlan id, which is not used in your fabric as fcoe vlan ID in SAN cloud.