I am hoping somebody can help me, i just finished upgrading UCS to 2.2.2c and noticed that my ESXi hosts (B200M3 with 1240 adapters) dropped the speeds from 40000 to 10000. I came across this bug https://tools.cisco.com/bugsearch/bug/CSCun21077 but my firmware and drivers are newer (2.2.2c and 2.1.2.50). Anybody else run into this?
If you are hitting the bug mentioned, then I assume you have QoS policies applied to the vnics? Change the Rate (Kbps) from line-rate to 40000000. The speed presented to VMware will change immediately. Unfortunately appears that line-rate on a few of the 2.2 releases actually means 10GbE.
1240 only, i notice speed dropped once the first FI got upgraded and restarted, all of mine VMware virtual switches were showing one link at 10000 and the other at 40000 or 20000 depending which server it was. Once the second FI was upgraded all virtual switches were reporting two vNICs at 10000. Really bummer with this upgrade so far.
I already verified that your suggestion works, i actually don't even have to restart a server i can just change the QoS policy to not set and speed on virtual switch goes back to normal. However in my environment i have two types of server that share that policy so i need to duplicate them because some servers should have speed of 20000 and some 40000. Thanks a lot for you help.
No worries, I thought this would fix it. You should be OK using 1 QoS policy at 40GbE. To my understanding it is an upper limit so if you are limited by the hardware then it will just step down. Easy to test out as you know, no reboot required for it.
If you are hitting the bug mentioned, then I assume you have QoS policies applied to the vnics? Change the Rate (Kbps) from line-rate to 40000000. The speed presented to VMware will change immediately. Unfortunately appears that line-rate on a few of the 2.2 releases actually means 10GbE.
Learn, share, save
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
