Probably BUG in vCenter Server 7.0.2.00400-18356314 - API

[Hr_Ross76]

Hi all,

When using API to delete an existing .iso file from an FC SAN storage, the "VMware vCenter Server" service crashes and has to be restarted manually. (Browser window "no helathy upstream")

##### STEPS TO REPRODUCE

We are using a vSphere cluster with attached FC SAN storage.

When calling to the API with a 'DELETE' request vpxd crashes.

1. Upload a test.iso file to data storage

2. Use API to delete test.iso 

curl -u user:password -X 'DELETE' -k 'https://192.168.0.123/folder/_ISO/test.iso?dsName=EXAMPLE_SAN0&dcPath=example_cluster'

3. "VMware vCenter Server" service crashes

Ist this a known issue?

Cheers

Rossi

[Ajay1988]

Interesting.. Let me check on this. 

Do you use SDRS or any other cluster features ? 

Have you raise a VMware Support Ticket? If not please do so ? 

[Hr_Ross76]

HI AJ,

yes please check on this.

We are doing automated installation with ansible and OVAs.
After the successfull deployment we have to delete the ISO file, but that wont work.

Our workaround is to copy an empty fle over the installer iso, because there are some passwords in there.

We are doing HA feature on the 2 node Cluster, SAN is direct coonected without FC switch.

We are still in evalutaion mode with our enviroment, regular license is ordered,so no support.

Thanky u for helping,

cheers

Rossi

[depping]

Tried to reproduce this with the same build in my lab, but not successful so far..

[pflu]

Hi, 

we found this bug trying to use Ansible to remove said .iso file.
And then narrowing down we found that it crashes if we do the call directly with curl.

The Ansible task was looking something like this:

- name: Modify CentOS virtual machine '{​{​ hostname }​}​' (remove cdrom)
community.vmware.vmware_guest:
hostname: '{​{​ hostname }​}​'
username: '{​{​ username }​}​'
password: '{​{​ password }​}​'
validate_certs: 'no'
datacenter: '{​{​ datacenter_name }​}​'
datastore: '{​{​ datastore_name }​}​'
name: '{​{​ hostname }​}​'
folder: '{​{​ datacenter_name }​}​/vm'
cdrom:
state: absent
controller_number: 0
unit_number: 0
delegate_to: localhost

Note that "state: absent" was creating the crash in our case.
So we inspected the Ansible module and found that what it does is basically a curl call. (The one posted above.)
If we run this in our environment we could provoke said crash.

If that isn't happening on your site I assume your setup is different from ours.

What is your current setup looking like?
Also do you know of a good way we could export our vSphere Setup, so we can provide you with more details?

--
Marc

[pflu]

Hi 

My answer was flagged as spam. I requested moderation.
Until then we have to wait I'm affraid...

[Ajay1988]

Does vpxd crash just by running curl delete command ?

I tried curl delete command but vpxd is stable for me.

VCSA 7.0 U2c (Upgraded from 7.0 U2b) + ESXi 7.0U2(fresh install) + No Cluster Features enabled + FC Storage

[pflu]

That's exactly the  case, above curl command crashes vpxd.