ESXi 6.5 HP OEM Purple Screen in HP 380 G7 Server

[crazycen]

I installed ESXi 6.5 HP OEM in my lab HP 380 G7 Server.

ISO:VMware-ESXi-6.5.0-OS-Release-4564106-HPE-650.9.6.0.28-Nov2016.iso

after installation ，esxi crash.

but VMware-VMvisor-Installer-6.5.0-4564106.x86_64.iso not have this issue.

[rherold]

and NOW the server boots up starts to load the VMs then goes to the purple screen of the death.  A reinstall over didn't fix it, or even instlal with out losing the VM...

Yes HPE has fucked this up ... and doesn't care about it.

[DavidinCT]

Going to dowload a live Linix disro... and nuke all the ESXi folders (besides the one with my VMs on it) and re-install 6.5a...then update 1...

Sigh... thanks HPE for screwing up my server...

[rherold]

I try to reachout HPE about this issue. Problem is that I have no G7 under contract in the moment.

If someone has please open a case ...

It may also help if more people join the discussion on the hpe forums:

HPE VMWare ESXi 6.5a Custom Image - Hewlett Packard Enterprise Community

https://community.hpe.com/t5/ESXi/ESXI-6-5-HPE-Image-Upgarde-G7-PSOD-caused-by-hpe-smx-provider/td-p...

[RAJ_RAJ]

Hi ,

As per HCL Gen 7 servers are not compatible with vsphere 6.5  .

Attached the list  , 

you can use 6.0 U2/U3 then do normal patching  with latest versions

[rherold]

As per HCL Gen 7 servers are not compatible with vsphere 6.5  .

hi I think we are all aware of this. But this is normal cause HPE does only certify there current model. If you look a little bit deeper into the HCL

you can see that all major parts from an G7 are supported. The Hardware runs without any problem with 6.5. The only issue is if you installed the

HPE tools (hp-smx) for harwdare monitoring it will end up in an PSOD, cause HPE has fucked up there own tools.

Even if the hardware is not supported I should never end up in an PSOD!!!

[RAJ_RAJ]

Hi ,

Major parts may supported but  if you open case with vmware and there is chance they say this is not supported .

Also  if all the parts supporting then only we use this in production else its risk for infra .

[rherold]

hi,

yes I hope that we not talk here about critical production systems. We for example use the G7 in our testlab and non critical workload.

But I think you knwo how the certification on vmware works. Not Vmware certify the system the vendor does and  trigger it.

No vendor has the interest in certify old hardware....

If the G7 would be a whitelabel box all the hardware like CPU / network/ storage/ chipset ) is supported. You can install it

with the normal 6.5 media without any problem. Ony if you wish to use the deeper hardware monitoring and try to install

the hpe tools for that it breaks...

So for my point of view it is not an vmware problem it is an HPE problem, that they have no interest that customer can run an G7 and deliver

high quality software to their customers with fine error handleing.

[Dieler]

HPE's ‘smx-provider-650.03.12.10.1-4240417.vib still does not fix the problem.. still PSOD.

But what I did, is customize the offline bundle of the latest HPE image instead of customizing the ISO. With the offline bundle, the upgrade via esxcli works fine and is a lot less work than doing it with the ISO.

I did the following steps:

1. Unzip the offline bundle (depot) and and delete the "smx-provider" from the vib folder.

2. Unzip the metadata.zip and edit the vmware.xml and delete the following entry:

-<vib>

<vibVersion>1.4.5</vibVersion>

<vibID>HPE_bootbank_smx-provider_600.03.12.00.17-2768847</vibID>

<name>smx-provider</name>

<version>600.03.12.00.17-2768847</version>

<vendor>HPE</vendor>

<vibType>bootbank</vibType>

<summary>WBEM Providers for ESXi 6.0</summary>

-<systemReqs>

<maintenanceMode>false</maintenanceMode>

<swPlatform productLineID="embeddedEsx" version="" locale=""/>

</systemReqs>

-<relationships>

<depends/>

<conflicts/>

-<replaces>

<constraint name="hp-smx-limited"/>

<constraint name="hp-smx-provider"/>

<constraint name="hpe-smx-limited"/>

<constraint name="hpe-smx-provider"/>

<constraint name="smx-limited"/>

</replaces>

-<provides>

<provide version="1.0.0" name="cim.DMTF.DSP1004"/>

<provide version="1.3.0" name="cim.SNIA.FC-HBA"/>

</provides>

<compatibleWith/>

</relationships>

-<postInstall>

<rebootRequired>true</rebootRequired>

<hostdRestart>false</hostdRestart>

</postInstall>

-<vibFile>

<sourceUrl/>

<relativePath>vib20/smx-provider/HPE_bootbank_smx-provider_600.03.12.00.17-2768847.vib</relativePath>

<packedSize>8513282</packedSize>

-<checksum>

<checksumType>sha-256</checksumType>

<checksum>49567090f839b8784ae1d6ba84d39d151f99a0aa814fba020c12aba9cbb753e8</checksum>

</checksum>

</vibFile>

</vib>

3.Delete the "smx-provider-291635400.xml" from vib folder inside "metadata.zip"

4. Edit the file "HPE-ESXi-6.0.0-Update3-600.10.1.0.73.xml" in "bulletins" folder and delete the vib entry: <vibID>HPE_bootbank_smx-provider_600.03.12.00.17-2768847</vibID>

5. Pack the metadata folder to "metadata.zip" again and afterwards the offline bundle to create the default folder structure

6. YOUR ESXi 6.5 U1 HPE Offline Bundle is ready

Then login to your ESXi server and do the upgrade via esxcli:

esxcli software vib install -d /path/to/your/custom/offline/bundle.zip

I like this solution (removing the smx-provider vib) more than adding an older version from ESXi 6.0.x, because you would run into problems when upgrading to a new HPE ESXi version, e.g. 6.5 U2.

When an older smx-provider vib is installed and you want to update to HPE 6.5 U2, you will do an update via esxcli and it will overrite the existing smx-provider and you could run into the "PSOD" again. When you use my solution, the vib is not installed and when you update to HPE 6.5 U2, it will not install the "smx-provider" vib, because the update function only updates existing vibs and will not install completely new vibs.

I hope some of you guys are able to get their HP DL380 G7 running with this fix. So far I did not notice any negative effect.

If someone wants to have the customized Offline Bundle, just let me know.

Is there anyone. who knows exactly what the smx-provider does? I could not find any details on the net.

Thanks and Greetings

[rherold]

Thx Dieler​ !

Here some informations about the smx stuff:

Overview of HP Insight Management WBEM providers for ESXi (2001543) | VMware KB

Seems that thsi is needed for monitoring the hardware

[RAJ_RAJ]

Hi ,

I have not tested on 380 G7 , i installed esxi 6.5 u1 build on hp 460 g7 .

Installed esxi 6.0 u3  , added on vcenter 6.5 u1 .

uploaded esxi 6.5 u1 iso to update manger . Using update manager applied the latest version  .

update success .

[nparas5]

Most PTE error codes are Hardware related mostly with CPU, Memory DIMM, or system board. Check with HP for the same, they will suggest you to upgrade the Firmware and BIOS as their usual Practice 🙂 But it works after firmware upgrade. If not HP will provide some commands to be run at SSH to host that will fix the issue for sure proceeding with the reboot.

VMware will also point to Hardware OEM in case of PTE and MCQs so better to check with HP. If you check the stack of the PSOD it will show you hp components there right before host crashed.

[rherold]

VMware will also point to Hardware OEM in case of PTE and MCQs so better to check with HP. If you check the stack of the PSOD it will show you hp components there right before host crashed.

This is def. no hardware related issue. It is a bug in the hpe smx stuff.

Problem ist that HPE has not certified the G7 for 6.5, but all included hardware components are supported.

[nparas5]

Here only HP can suggest, you can share them the screenshot and AHS logs if the hardware is under warranty. After upgrading to 6.5 i also faced same issues approx 6-7 PSODs in a week, Few are resolved by upgrading firmware and BIOS whether for few they gave some hp commands to run on ssh host console proceeding with the reboot.  And it all resolved.

Hope i could give you the hp articles for commands but i recently joined new organization and all was on emails.

[mpevner]

Did the commands they ask of you include removing the smx-provider vib? I'm curious if they were able to make it work with that installed.

[AdilArif]

This server is not supported for ESXi 6.5.

Please refer to the below from the VMware HCL guide.

[larrebarre]

Is there anyone that has tried 6.5 HPE custom Image after update oct 1st?

[rherold]

Yes same problem here at DL385 G7.

We have now startet with deactivating the smartarray controller and boot esxi vom sdcard/usb with normal

vmware image. Thats seems to run fine.

I will be at the HPE Discovery in Madrid and hope to get some of the people which are responsible for

this code in my fingers....

[rherold]

Still not work with the OEM image from 26.Oct

[CelsoNarita]

This one worked for me:

https://blog.monstermuffin.org/fixing-esxi-6-5-hpe-g7-servers/

Pay attention***

"You must do a clean install for this to work, upgrading will not work as the offending driver will not be overwritten."

[rherold]

Yes this will work as long as you don't wan't update to current HPE tools or updates after this.

And it is not a clean future save fix.