As simple as UCS might look from the outside, the configuration is a piece of hell! 6 tabs (Equipment, Servers, LAN, SAN, VM, Admin) where each contains a lot of different settings show up - and in the first few minutes you look puzzled at all the names like Service Profiles, Policies, Pools...
After some time of configuration with an expert, the settings - where to configure what - is now kinda 'ok'. Meaning, I now know the most important configuration stuff.
But: The whole chassis and therefore all the blade servers are not connected to ethernet. It seems that the uplink (a configured port-channel) doesn't work. The following error message proves it:
lan port-channel 1 on fabric interconnect A oper state: failed, reason: No operational members
lan port-channel 1 on fabric interconnect B oper state: failed, reason: No operational members
So we decided to destroy and recreate the port-channel. Since we have changed the SFP's the configuration might have had some problems with the change. We deleted the port members of the port-channel configuration and re-created the port-channel group, making sure we don't do any mistakes by using the official Cisco howto.
The funny thing is, that even in the official documentation the screenshot shows error messages. Take a look at this:
Cisco states here: 'As shown, the port channel that was created is successfully enabled.'. Even though there are errors shown that the port-channel has no operational members. As of now, our UCS is still down... and no solution could be found so far.
Thanks to Fabien to have found this "faulty" Cisco documentation :-).
Update 12:27: The ethernet issue has been fixed. It was due to the protocol which was used for the port-channel between an ethernet switch and the UCS Fabrics. It was decided to use the LACP protocol which now solved the problem.
Claudio from CH wrote on Jun 5th, 2011:
Hmm.. I guess we both can agree that the Cisco documentation misses a lot of important information. Let me know about your 'fight' :)
Philippe from FR wrote on Jun 4th, 2011:
Our setup is the same.
During our test, I saw the status of port channel \"up\", but the information \"No Operational member\" was always present.
I would do another test this week.
Claudio from CH wrote on Jun 4th, 2011:
On the Nexus side we only specified the ports which should be used for the port channel. No further configuration was done on UCS side. On our main switch (a Cisco 3750) we defined the port-channel like seen in my previous comment. The Nexus switches are connected kinda like this:
Cisco 3750/Stack 1/Port 31 -> UCS Fabric B Port 4
Cisco 3750/Stack 2/Port 31 -> UCS Fabric B Port 5
Cisco 3750/Stack 1/Port 30 -> UCS Fabric A Port 4
Cisco 3750/Stack 2/Port 30 -> UCS Fabric A Port 5
So its important that the port-channel was defined in the same switch (Cisco 3750 in our case). But I guess you already have done that, too?
philippe from FR wrote on Jun 3rd, 2011:
hello, thank you for your prompt response.
Even with "active mode" enabled, it does not work.
Our two Fabric Interconnect are connected to two Nexus 5000 (in VPC mode ).
Each FI connected to both N5000.
FI cluster is in "end host" mode but we also tried to switch mode, without results.
If you have an idea, do not hesitate to let me know.
Claudio from Switzerland wrote on Jun 1st, 2011:
Actually it's the channel-group mode which you have to set to active on an interface:
description *** UCS Fabric B ***
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 100,716,723,725,731,751-753,760
switchport mode trunk
channel-group 4 mode active
Important is that you use mode active for both ports of the port-channel.
philippe from France wrote on Jun 1st, 2011:
Hello, we have the same problem with port channel between UCS Fabric Interconnect and Nexus 5000.
What is the LACP mode configure on the Ethernet Switch?
AWS Android Ansible Apache Apple Atlassian BSD Backup Bash Bluecoat CMS Chef Cloud Coding Consul Containers CouchDB DB DNS Database Databases Docker ELK Elasticsearch Filebeat FreeBSD Galera GlusterFS Grafana Graphics HAProxy HTML Hacks Hardware Icinga Icingaweb Icingaweb2 Influx Internet Java KVM Kibana Kodi Kubernetes LXC Linux Logstash Mac Macintosh Mail MariaDB Minio MongoDB Monitoring Multimedia MySQL NFS Nagios Network Nginx OSSEC OTRS Office PGSQL PHP Perl Personal PostgreSQL Postgres PowerDNS Proxmox Proxy Python Rancher Rant Redis Roundcube SSL Samba Seafile Security Shell SmartOS Solaris Surveillance Systemd TLS Tomcat Ubuntu Unix VMWare VMware Varnish Virtualization Windows Wireless Wordpress Wyse ZFS Zoneminder