Re: Issue with smx-providers on VMware 6.5 U3

str1ker · ‎09-04-2020

Hi VMware-Folks,

I ran into the following issue and want to try to get the VMware-View of the problem.

Maybe someone has run into the same issue:

We have serveral HPE DL360 Gen10 Servers, installed with the HPE cursomized Image: VMware-ESXi-6.5.0-Update3-14990892-HPE-Gen9plus-650.U3.10.5.0.67-Dec2019.iso

All share the same Issue: If we try to poll Status via WBEM, nothing returns, as if the package is not installed.

I could narrow it down to the following:

The WBEM-Provider List shows a stange ".#vmw_smx-provider" entry:

esxcli system wbem
provider list

Name Enabled Loaded

------------------ ------- ------

.#vmw_smx-provider true false

sfcb_base true true

vmw_base true true

vmw_hdr true true

vmw_hhrcwrapper true true

vmw_iodmProvider true true

vmw_kmodule true true

vmw_omc true true

vmw_pci true true

vmw_smx-provider true true

vmw_vi true true

Looking at /var/lib/sfcb/registration gives:

-rw-rw-rw-    1 root     root           112 Sep 3 13:38 loaded
drwxr-xr-x    1 root     root           512 Sep 3 13:37 repository
-r--r--r--    1 root     root          3513 Nov 1 2019 sfcb_base-providerRegister
-r--r--r--    1 root     root          4307 Nov 1 2019 vmw_base-providerRegister
-r--r--r--    1 root     root           374 Nov 1 2019 vmw_hdr-providerRegister
-r--r--r--    1 root     root           897 Nov 1 2019 vmw_hhrcwrapper-providerRegister
-r--r--r--    1 root     root          4220 Nov 1 2019 vmw_iodmProvider-providerRegister
-r--r--r--    1 root     root           164 Nov 1 2019 vmw_kmodule-providerRegister
-r--r--r--    1 root     root         15650 Jun 21 2019 vmw_omc-providerRegister
-r--r--r--    1 root     root          1550 Nov 1 2019 vmw_pci-providerRegister
-r--r--r--    1 root     root          1308 Sep 3 13:38 vmw_smx-provider-providerRegister
-r--r--r--    1 root     root          2898 Mar 15 2018 vmw_vi-providerRegister

However there is the hidden file mentioned above:

ls -l /var/lib/sfcb/registration/.#vmw_smx-provider-providerRegister
-r--r--r-T 1 root root 77881 Aug 9 2019 /var/lib/sfcb/registration/.#vmw_smx-provider-providerRegister

This file shares the same bytesize on Gen9 Servers and obviously contains the HPE specific data.

If I overwrite the vmw_smx-provider-providerRegister with the contents from .#vmw_smx-provider-providerRegister the polling works as intended.

Until the next reboot or even just until restart of the mangagement network ...

I found: https://www.virtuallyghetto.com/2011/08/how-to-persist-configuration-changes-in.html

.. and tried to do use the Sticky Bit, but nothing kept the file from beeing overwritten with the default(?) version.

I also tried to downgrade the smx-provider package up to three versions and downgraded the whole management bundle in a separate attempt.

Nothing helped.

Gen9 Systems with the same image do not have that Issue at all. (as mentioned: there is just one vmw_smx-provider-providerRegister with the corrent and same bytesize as the .#vmw_smx-provider-providerRegister Version on the Gen10 systems)

Does someone have an Idea how to replace the vmw_smx-provider-providerRegister with the correct version and stick that version into the bootbank ?

Upgrading to 6.7 Ux will happen, but is currently not an option for a quick fix.

The mentioned systems are the only Gen10 servers in this environment.

str1ker · ‎09-30-2020

As an update:

I could not find a solution to this. As mentioned before, I tried downgradind to the following versions:

HPE_bootbank_smx-provider_650.03.14.00.3-4240417
HPE_bootbank_smx-provider_650.03.15.00.4-4240417

.. without any change.

The Version from the Image was

HPE_bootbank_smx-provider_650.03.16.00.4-4240417

I then tried the HPE_bootbank_smx-limited_650.03.16.00.1-4240417

and the .#* Entry is gone:

esxcli system wbem
provider list

Name Enabled Loaded

------------------ ------- ------

sfcb_base true true

vmw_base true true

vmw_hdr true true

vmw_hhrcwrapper true true

vmw_iodmProvider true true

vmw_kmodule true true

vmw_omc true true

vmw_pci true true

vmw_smx-limited true true

vmw_smx-provider false true

vmw_vi true true

The small / default vmw_smx-provider-providerRegister is this there, but is not used in this scanario:

ls -l /var/lib/sfcb/registration

total 100

-rw-rw-rw- 1 root root 128 Sep 7 13:39 loaded

drwxr-xr-x 1 root root 512 Sep 7 13:38 repository

-r--r--r-- 1 root root 3513 Nov 1 2019 sfcb_base-providerRegister

-r--r--r-- 1 root root 4307 Nov 1 2019 vmw_base-providerRegister

-r--r--r-- 1 root root 374 Nov 1 2019 vmw_hdr-providerRegister

-r--r--r-- 1 root root 897 Nov 1 2019 vmw_hhrcwrapper-providerRegister

-r--r--r-- 1 root root 4220 Nov 1 2019 vmw_iodmProvider-providerRegister

-r--r--r-- 1 root root 164 Nov 1 2019 vmw_kmodule-providerRegister

-r--r--r-- 1 root root 15650 Jun 21 2019 vmw_omc-providerRegister

-r--r--r-- 1 root root 1550 Nov 1 2019 vmw_pci-providerRegister

-r--r--r-- 1 root root 31636 Aug 9 2019 vmw_smx-limited-providerRegister

-rw-r--r-- 1 root root 1308 Sep 7 13:39 vmw_smx-provider-providerRegister

-r--r--r-- 1 root root 2898 Mar 15 2018 vmw_vi-providerRegister

As a result, only a subset of WBEM date is available, which is better then nothing.

According to the Date, the File just gets overwritten, and the (correct) version from HPE is backed up on every reboot. Isn't there a way to forcefully prevent that ?

I have beed digging through the change logs of all update released, but this issue is not mentioned (just memory leaks etc.).

rooter344 · ‎12-21-2020

In the recently released VMware ESXi 6.7, Patch Release ESXi670-202008001, in the Resolved issues sections, you will find: "PR ... at least 24 hours: To disable a provider, e.g. the smx-provider, run the following command ...