ESXi host shuts down/restarts when guest does

[AleTib]

Hi everyone,

I recently moved my stand-alone ESXi (was running on a PC) to new hardware; Since that, each time I restart or shutdown my Linux guests (no matter if from vmware or within the host), the host does the same.

Config:

Proliant DL380e Gen8

2x E5-2450L

6x8GB
LSI SAS HBA with:
- 1x300GB disk for VMFS Datastore
- 4x3TB drives passed as RAW disks
4xSATA drives (connected externally through a dedicated PCIe eSATA controller passed to the linux host )
ESXi booting from SD (but logging on A)
ESXi 6.7.0 Update 3 (Build 15160138) installed through HP custom ISO downloaded from vmWare

VMs:

- one windows server host for tape backup (one SCSI and one FC pass-through)
- one multi-puprpose Ubuntu 18.04 guest (mainly serving as NAS)

Had a look at the logs but didn't notice anything peculiar; for example in syslog:

2020-03-25T20:35:00Z crond[2098477]: USER root pid 2139772 cmd /bin/hostd-probe.sh ++group=host/vim/vmvisor/hostd-probe/stats/sh

2020-03-25T20:53:42Z watchdog-vobd: [2097683] Begin '/usr/lib/vmware/vob/bin/vobd', min-uptime = 60, max-quick-failures = 5, max-total-failures = 1000000, bg_pid_file = '', reboot-flag = '0'

there's an hole in between: the machine just restarted in between.

HW seems to be working fine (been running for days without a glitch).

Any idea? Tried to google around but found tons of unrelated stuff about shutting down/restarting guests etc. but nothing related. Beside being quite bothering, that sounds quite a security issue considering the guest doesn't seem to be isolated as it should.

PS: not a production environment so I can run all the required tests.

Thanks for your help/ideas!

[diegodco31]

You could try to install the 6.0U3 version of esxi.

6.7 isn't compatible with this hardware.

VMware Compatibility Guide - System Search

[AleTib]

Hi Diego,

thanks for your answer. Two questions:

1. by "is not compatible": do you mean "is not in the compatibility list" or "there's a well known issue preventing ESXi 6.7 from running correctly on that HW"?
2. Are you suggesting to try 6.0 for any specific reason or just a blind test?

As far as I know there is no way to easily downgrade (and I mean downgrade, not roll back) from 6.7: I should go through vconverter if it still exists/can convert from vmHW from 6.7 to 6.0.

KR,

[diegodco31]

1 - Such compatibility matrix specifies only the list of hardware that was tested and proven compatible.

2 - Because is the version that has been tested.

[daphnissov]

To elaborate on what Diego (correctly) points out:

1. When you stray from the compatibility list, it becomes difficult to reason about what "should" and "should not" occur when problems strike. A common fallacy is often "if it works on 6.0 then it should work on 6.7." In reality there are massive changes between even point versions, so such an assumption is a guess.
2. Going back to a version which is known good (by virtue of it being stated as compatible), you can rule out version drift. If the problem occurs in a known supported posture, then you have eliminated that variable and can try other troubleshooting methods.