Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi
-
So this will definitely impact some if not all of you, but all Gen 10 and newer HPE Servers with hardware RAID, running ESXi will no longer be able to monitor the drive or array's health. Below I'm quoting Mario Glenwin from L2 Technical Solutions Consultant from HPE.
We received a confirmation from engineering team. Smx-provider package is set to retire and we are recommending customers to remove the same going further.
There is no additional development/research happening on smx-provider.Engineering team have recommended that we could monitor drives from ILO/AMS and One View if configured.
Support article for the recommended fix: https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=a00117054en_us
With 3.53 or newer raid controller firmware on Gen10 or newer servers you can expect to see misreported issues like the ones pictured below.
HPE's current officially recommended solution to prevent seeing these errors is to uninstall the SMX-Providers from within ESXi. Unfortunately VMWare and HPE documentation lists the smx-provider as the solely supported way to monitor storage.
https://www.vmware.com/resources/compatibility/pdf/vi_cim_guide.pdf
You could of course move to HPE OneView or potentially use ILO to email alerts out of an individual server but there are far more fault domains then getting directly reported from the tooling on the Hypervisor (in this case VMWare).
I've been in conversations for months now about this with HPE and it appears that this is likely not going to change for the better.
-
Good thing no one here runs ESXi..
-
@dashrender said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
Good thing no one here runs ESXi..
Sure and none of us support it either.
....
-
@dustinb3403 said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dashrender said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
Good thing no one here runs ESXi..
Sure and none of us support it either.
....
Just like HPE. jajaja
-
@scottalanmiller said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dustinb3403 said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dashrender said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
Good thing no one here runs ESXi..
Sure and none of us support it either.
....
Just like HPE. jajaja
That is actually a good one
-
@dustinb3403 who cares. This is what the iLO Amplifier Pack is for. Just use that. It even automatically opens tickets with HPE.
-
@voip_n00b said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dustinb3403 who cares. This is what the iLO Amplifier Pack is for. Just use that. It even automatically opens tickets with HPE.
You clearly don't support customers.
"Hey those new servers that were easy to monitor and support, yeah well now all of that is gone. But you can buy this other thing to get a tiny portion of the functionality back."
-
@dustinb3403 It’s free and has more functionality.
-
@voip_n00b said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dustinb3403 It’s free and has more functionality.
For a standalone environment, it is not msp/itsp friendly.
-
This post is deleted! -
@dustinb3403 Why?
-
@voip_n00b said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dustinb3403 Why?
Because you're required to setup and maintain and additional environment for it.
1 customers won't want to spend more money for something they've previously had included (by hpe) and 2 ESXi needs this functioning to report on the underlying health status.
-
it does suck, but shit changes. That's only natural.
-
@dustinb3403 said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@voip_n00b said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dustinb3403 Why?
Because you're required to setup and maintain and additional environment for it.
1 customers won't want to spend more money for something they've previously had included (by hpe) and 2 ESXi needs this functioning to report on the underlying health status.
Explain.... "Well, this happened because you are overpaying for something that you don't need and now it costs even more because they know you will pay because you are already paying just for the sake of paying. Instead of paying more, you could pay less."
And they will instantly say "oh heck no way do we want to SAVE money, spend spend spend"
-
@scottalanmiller said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dustinb3403 said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@voip_n00b said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@dustinb3403 Why?
Because you're required to setup and maintain and additional environment for it.
1 customers won't want to spend more money for something they've previously had included (by hpe) and 2 ESXi needs this functioning to report on the underlying health status.
Explain.... "Well, this happened because you are overpaying for something that you don't need and now it costs even more because they know you will pay because you are already paying just for the sake of paying. Instead of paying more, you could pay less."
And they will instantly say "oh heck no way do we want to SAVE money, spend spend spend"
What?
HPE is removing a hardware monitoring provider for VMWare (and presumably everything else). The assumption that anyone who has hardware, must be able to monitor it, ideally through their hypervisor.
Sure shifting to monitoring through the hardware interface, such as ILO or OneView but these approaches add yet another administrative panel that must be used and managed and maintained.
-
Unfortunately tech that goes obsolete always causes problems but it's more technically sound to monitor through the OOB management interface.
It's after all independent of the OS running on the hardware, independent of the server's NICs, independent of most hardware failures and can be used for a lot more than just monitoring.
And in any modern installation, the OOB management should have been setup and in use already.
-
@pete-s said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
Unfortunately tech that goes obsolete always causes problems but it's more technically sound to monitor through the OOB management interface.
It's after all independent of the OS running on the hardware, independent of the server's NICs, independent of most hardware failures and can be used for a lot more than just monitoring.
And in any modern installation, the OOB management should have been setup and in use already.
Absolutely I agree with that, except the only OOBM that existed before OneView was Ilo and smtp emailing. Which is hardly reliable.
And I do agree that moving to an OOBM like OneView makes sense, it doesn't make sense for an ITSP to have to use though, as it's setup per customer, and would be running on the same hardware it's monitoring in most cases.
Edits are corrected typos
-
@dustinb3403 said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@pete-s said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
Unfortunately tech that goes obsolete always causes problems but it's more technically sound to monitor through the OOB management interface.
It's after all independent of the OS running on the hardware, independent of the server's NICs, independent of most hardware failures and can be used for a lot more than just monitoring.
And in any modern installation, the OOB management should have been setup and in use already.
Absolutely I agree with that, except the only OOBM that existed before OneView was Ilo and smtp emailing. Which is hardly reliable.
And I do agree that moving to an OOBM like OneView makes sense, it doesn't make sense for an ITSP to have to use though, as it's setup per customer, and would be running on the same hardware it's monitoring in most cases.
Edits are corrected typos
Why can't you have one Oneview hosted centrally and have it communicate with iLO over VPN or whatever?
That's how a centrally managed vCenter is setup isn't it? It's also how you would have to manage a server remotely using iLO.
-
@dustinb3403 said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
@pete-s said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
Unfortunately tech that goes obsolete always causes problems but it's more technically sound to monitor through the OOB management interface.
It's after all independent of the OS running on the hardware, independent of the server's NICs, independent of most hardware failures and can be used for a lot more than just monitoring.
And in any modern installation, the OOB management should have been setup and in use already.
Absolutely I agree with that, except the only OOBM that existed before OneView was Ilo and smtp emailing. Which is hardly reliable.
And I do agree that moving to an OOBM like OneView makes sense, it doesn't make sense for an ITSP to have to use though, as it's setup per customer, and would be running on the same hardware it's monitoring in most cases.
Edits are corrected typos
How are you doing those things today? If you're using a centralized server to manage all of your clients, why can't you manage iLo the same way?
I agree, in this day and age - that's super risky, i.e. you get compromised and all of your customers are now compromised.
though just because you have 100 passwords, one for each client, that info has to be stored somewhere and perhaps it would be compromised as well - and your clients are still compromised...
-
@dashrender said in Goodbye hardware monitoring on HPE Gen10 and newer equipment running ESXi:
I agree, in this day and age - that's super risky, i.e. you get compromised and all of your customers are now compromised.
though just because you have 100 passwords, one for each client, that info has to be stored somewhere and perhaps it would be compromised as well - and your clients are still compromised...
Risk has to be managed but it's not more risky having 100 customers with one server each on-prem than having 100 servers in one location.
