Cisco 7K Port Power Errors NXOS 7.2(2)D1(2)

Recently we upgraded the NXOS on our 7Ks and now we are getting reports of power issues on individual ports. Performed a rediscover and repolled with no change. This is happening on all the 7Ks that were upgraded.

Ethernet4/17 Lane 1 Transceiver Bias Current Sensor Major 350 µA

Ethernet4/17 Lane 1 Transceiver Receive Power Sensor Major -40

Ethernet4/17 Lane 1 Transceiver Transmit Power Sensor Major -14.71

Find more posts tagged with

hardware sensor alert

cisco nexus 7k

cisco nxos

hardware health

cisco hardware

Accepted answers

All comments

rschroeder

After researching these types of errors with TAC and discovering they were virtually all false positives reported by NPM, and also after researching the errors within Thwack, I've swapped MIBs for my pollers from CISCO-ENTITY-SENSOR-MIB to CISCO-ENVONMON-MIB. The warnings in NPM went away.

The process to accomplish this:

kantort

rschroeder,

Tried this without success. I made the change, and repolled with no change. I then did a rediscover and repolled again without any change. I will do more research when I get a chance. Thanks.

rschroeder

It's always possible NPM is reporting a valid problem, as opposed to a false positive. First verify with TAC that there is NOT really a problem present. After that, you can spend time tweaking NPM to stop reporting a false positive.

foonly

We saw this as well after uprgrading from 6.2(10) to 7.2(2)D1(2) and then on to 7.3(1)D1(1), but only on one set of switches.

For us, the ports that are reporting it had been configured on 6.2(10), then the config was manually cleaned out, and the ports were disabled. The majority of our ports did not do it.

As I mentioned on post:Re: 10.4 hardware health monitoring - anybody know what a bias current sensor is?

It appears that NX-OS does not support CISCO-ENVMON-MIB, so I think you would lose all hardware sensing supported by CISCO-ENTITY-SENSOR-MIB.

So I've disabled the individual sensors in Orion.

Another thing we noticed with this switch upgrade, but not with others, is that the SNMP ifindex values changed for many ports. According to this docwiki, ifindex persistence is on by design in NX-OS, and cannot be disabled:

Cisco NX-OS/IOS Software Default Configuration Differences - DocWiki

So there must be some sort of Cisco bug, I think.

It may have related to us going from one line of code to another on this switch:

old: n7000-s2-dk9-npe.6.2.10.bin
new: n7000-s2-dk9.7.3.1.D1.1.bin

=Foon=

bmallon

I'd claim to have answered this myself, but it's on a Cisco forum that I found my pseudo answer. I kinda don't like the 'solution', but i had to stop the false alerts. The 'solution' is at the very bottom.

Here's the article link: Solved: Te1/1/X Bias Current Sensor Solar Wind ... - Cisco Community

Cisco TAC told me the Bias Current value expressed in mA is not the most interesting one to monitor the health of fiber links. This value has been removed from the recent Cisco IOS XE images that's why it is not displayed on the result of the command # show int tex/x transceiver detail.

You can use the Optical Transmit/Receive Power values to monitor the health of fiber links.

By checking on the Distribution switch (Cat6800 running another IOS flavor), the Bias Current value is displayed and is between thresholds.

So the problem is located on Solarwinds side.

Solarwinds didn't provide any hotfix yet.

As a workaround, you can disable the polling of the Current Bias sensor:
Settings > All Settings > Manage Hardware Sensors > Group By Node, and disable the sensor.

rschroeder

I deployed this same sensor disabling solution after discovering similar problems reported on some new Cisco ASR 920 routers. I suspect their specific issue might be associated with having Interfaces with GBICs inserted and that have configurations present, but no link has been attached to them yet.

Once I get a link connected to them I'll re-enable the sensor pollers and confirm the problem is gone.

Thank you, bmallon for sharing the process to individually disable them. It's much easier to identify all problem sensors in all nodes and to disable them in bulk via the SolarWinds 12.x code train than it was in previous versions.