[German]Quick note for HPE server users who are running VMware ESXi 6.5 Update 3 and ESXi 6.7 Update 3 with a custom image. There it comes to 'hangs', VMware has now confirmed this.
This is the issue
VMware ESXi is a bare-metal hypervisor, i.e. a minimal operating system that can be installed directly on your physical server. Then virtual machines can be installed on the server.
For HP servers (ProLiant), customized VMware ESXi-ISO files are offered for installation. These include all necessary drivers and management software to run ESXi on HPE servers. In addition, according to this HP announcement, this customized installation should work seamlessly with Intelligent Provisioning.
For a successful installation, the customized image of HPE is required for certain ProLiant servers. The drivers for the new network and storage controllers in the servers are integrated in the custom image of HPE and are not part of the general ESXi image distributed by VMware. To use ESXi, the drivers for these controllers must be integrated because you cannot add the controllers during installation.
So far the theory, in practice there are problems with ESXi solutions that hang in the constellation with HP custom images. I had already pointed out the problem in the blog post Caution when updating to ESXi 6.5 Update 3 HPE Custom at the end of July 2019. A user wrote within a email to me:
With a Clean Install on two differently configured ML350 Gen10 the management interface doesn't react about 2 minutes after startup. Neither via the WebUI, nor via the console via F2 (also not via Alt F1). The only way to shut down or restart the server is via F12.
Other users confirmed these problems also on IBM/Lenovos.
VMware confirms the issues
In this comment (thanks) German blog reader Christoph points out that VMware has published a support article On ESXi host running 6.5 U3 and 6.7 U3 certain tasks like entering/exiting maintenance mode and connecting to VC are getting hung (74966) on 9/22/202020, which confirms the problem. There it says:
- On ESXi host running 6.5 U3 and 6.7 U3 certain tasks like entering/exiting maintenance mode and connecting to VC are getting hung.
- esxcli command hangs after restarting wbem service.
- On avg, a host with the problem takes 5 to 10 minutes from the time the ESXi DCUI login screen fully loads until it connects to the VC
- ESXi host randomly disconnects from vCenter server.
- HA/DRS task getting stuck and vMotion may time out.
In the support article log files are listed, which show the error case. This problem occurs when CIMSVC is enabled and attempts to periodically update the sensor data. This causes other hostd threads to wait for the update data to complete, causing the slowdown. However, the post must have been published in 2019, as blog reader Boris already mentioned the support article in this comment in October 2019. Workarounds are shown in the support post. This issue is also fixed in VMware ESXi 6.5 patch release ESXi650-201910001 and VMware ESXi 6.7, patch release ESXi670-201912001.
Cookies helps to fund this blog: Cookie settings