The ESX host overall hardware status does not update

One of our ESX hosts had a memory failure and the RAM was replaced but the overall hardware status still showing an alert for the memory warning status. We've cleared the alert in both vmware and DRAC software but SAM still's showing the alert. We are polling it through VMware. How can I troubleshoot the hardware status polling?

Find more posts tagged with

esx_status

Accepted answers

All comments

aLTeReGo

There is a known bug in VMware's API where status is not properly reset after a failure. You're certainly welcome to open a case with support and we will walk you through the VMware API so you can see this for yourself. From there you will have all the information needed to open a case with VMware support. I believe this issue is corrected in ESXi v6.0 but we have not as yet been able to confirm. My understanding from at least one other person that encountered this issue was that rebooting the vCenter and ESX host temporarily resolved the issue until the next failure. The easiest workaround to the problem though is to simply change the polling of that ESX/ESXi host from from "Poll through vCenter" to "Poll ESX server directly. This changes the polling mechanism from the VMware API to CIM, which does not exhibit the same behavior.

VMware Poll Directly.png

mmcgallicher

I'm sure you've worked out this issue by now (sorry for bumping an old thread), but for those that come across this later. . .

I used the below KB to reset, then refresh the sensors to avoid rebooting my host or configuring the node to poll ESXi directly. Hardware Health shows happy again.

https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1037330

jbiggley

mmcgallicher, this is a bit of dated reply but check out the wicked awesome reply I just got from chrispaap detailing how to check if this is a VMware issue.

How is hardware status data propagated from VMAN to Orion via Integration?

aabdul

Thanks for sharing the workaround. The Vmware KB does not mention which Vcenter version this applies to. Seeing similar issue in 6.5

NightEdition

Sorry for bumping old thread, however I'm also seeing this issue when polling via the VMWare API.

I can manually refresh/reset the sensors in VMWare, SolarWinds then picks up the new stats, but the sensors then never automatically refresh and will remain at the same values. I have this same issue on all my ESXi hosts in SolarWinds.

There also no longer appears to be the option in SolarWinds to poll the ESXi host directly (i.e. via CIM protocol), VMAN/Orion now only appears to interact with the VMWare API which is plagued with these issues.

Has anyone encounter similar issues and been able to overcome?

tobyw_loop1

We potentially have a problem with a client and are getting them to trial the workaround KB to see if this gets SolarWinds to update. All looks fine in vCenter but ESX Host in SW showing Memory Red.

benleni0

Sorry for bumping old thread, however I'm also seeing this issue when polling via the VMWare API.

NightEdition

I was able to implement a change on each of my ESXi hosts in order to "fix" this....

In order for up-to-date hardware health status for the hosts to be reported via the VCentre VMWare API, a "throttleInterval" entry needs to be added on each ESXi host VCentre agent config file.

This will force the hardware health to update from the host back to VCenter (which is being polled in SolarWinds) on a regular interval (i.e. 600 seconds). This needs added on the "\etc\vmware\vpxa\vpxa.cfg" file on each ESXi host:

<config>
....
<vpxa>
....
....
<healthSystem>
<throttleInterval>600</throttleInterval>
</healthSystem>
</vpxa>
<workingDir>/var/log/vmware/vpx</workingDir>

To apply the change, the vpxa service on the ESXi host will need a restart:
Run "services.sh restart vpxa" on the host or use the GUI and restart the "VMware vCenter Agent" service.

Not ideal having to make a change on each host, but got things updating for me!

More info here, https://www.veeam.com/kb1654