Solidfire NVRAM device problems detected

[chinchillaking]

Hi All,

We check the Solidfire display below error, I can found NetApp KB "NVRAM device problems detected, metadata drive will be removed" but the error not same as "error=deviceCapacitanceOffThePack", should I require open support case?

nvramDeviceStatus

NVRAM device problems detected, metadata drive will be removed. Contact NetApp support, error=deviceCapacitanceOffThePack 3.926F

Best rgeards,

Chung 

[elementx · Accepted solution]

Definitively open a support case for all such suspicious errors. Warnings may be temporary, but Errors could become critical.

In this case, after this error maybe one of your nodes lost Drive 0 following this error (check Drives > Available), so although redundancy has most likely been restored by redistributing data on the remaining nodes, you may have one non-functioning metadata drive, which isn't good.

[chinchillaking]

Hello,

If HCI Solidfire in private subnet without send to ActiveIQ outside, but the mNode can reach vCenter only, if Solidfire have warning or error, could vCenter detect the Solidfire problem and send the email notification to outside? 

[elementx]

It works the other way around: mNode/SolidFire push alerts to vCenter

"NetApp HCI alert monitoring forwards NetApp HCI storage cluster system alerts to vCenter Server, enabling you to view all alerts for NetApp HCI from the vSphere Web Client interface."

https://docs.netapp.com/us-en/hci/docs/task_mnode_enable_alerts.html#alert-monitoring-options

I once looked into the possibility of creating notifications on VMware based on these events, but it seems those are only vCenter-focused, and cannot be turned into vCenter notification alerts.

Could you create a custom gateway VM and add a route on mNode to go to that proxy VM in order to reach SF MVIP?

You could also do this in several other ways.

I think the easiest way is to not do it in vCenter, but to forward SolidFire syslog to external destination (even vRealize). Then there filter events and create alerts.

Or, enable SNMP Traps and just forward SolidFire alerts to some destination where customer already has SNMP monitoring. That can be Splunk, Elastic, etc.

You can google it (I wrote about this) or if you need help you can contact me offline.