Apr 29, 2020

I am having the same issue as Robert below - HCIBench on segment 1, test vm on segment 2. DHCP enabled on both segments, and HCIBench appliance is receiving a DHCP address on both. Test vm deployed on either aren't receiving any ip addresses, regardless of whether using the segment DHCP or the HCIBench appliance as the DHCP server.

Apr 29, 2020

Furthermore, if both the NICs on the HCIBench appliance are connected - I can't browse to the configuration page. Once I disconnect vmnic2 (the VM network nic), I can browse to it just fine.

If both vmnic are connected, I can't reliably ping either, but if either is disconnected, PING requests are returned reliably... I think there is a problem with the internal gateway settings...

Apr 29, 2020

Hi Tom, this is definitely a configuration that we use in our lab so this behavior is unexpected. Can you reach out to me by email at vsanperformance@vmware.com?

Apr 29, 2020

The issue was a blocked port between HCIBench and the ESXi hosts preventing the deployment of the worker VMDK.

Added a firewall rule to permit ICMP and HTTPS(443) from HCIBench to the ESXi hosts solved the issue.

Apr 24, 2020

I'm trying to use HCIBench on a V6.7 environment with VSAN and every time I run prevalidation I get failed IP tests of one kind or another. It looks like there's always one of the test VMs that fails it's ping tests.

2020-04-24 14:43:28 -0700: IP Assignment failed or IP not reachable

From one of the tvm logs:
networks: Pub-Net-1284 = Bench-VLAN
Deploying VM on 10.252.3.64...
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 14.4M 100 14.4M 0 0 36.8M 0 --:--:-- --:--:-- --:--:-- 36.8M
Powering on VMs ...
PowerOnVM hci-tvm-vsanDatastore-6-1: success
Waiting for VMs to boot ...
green
hci-tvm-vsanDatastore-6-1: 192.168.3.6
PING 192.168.3.6 (192.168.3.6) 56(84) bytes of data.
From 192.168.2.1 icmp_seq=1 Destination Host Unreachable
From 192.168.2.1 icmp_seq=2 Destination Host Unreachable
... several failed ping attempts...
PowerOffVM hci-tvm-vsanDatastore-6-1: success
Destroy hci-tvm-vsanDatastore-6-1: success
Can't Ping VM VirtualMachine("vm-858") by IP 192.168.3.6
254
2020-04-24 14:43:28 -0700: IP Assignment failed or IP not reachable

Apr 27, 2020

Usually this type of problem is a configuration difference if using standard virtual switches and/or a misconfigured trunk port on the physical switch (e.g. VLAN not allowed).

Create 2 VM: one on host 10.252.3.64 and on on another host. Configure the IP on those VM (e.g. 192.168.3.1/24 and 192.168.3.2/24). If the VM cannot ping each other but if you vMotion the VM from host 10.252.3.64 to other hosts and the pings succeeds, this is a strong confirmation of a network configuration problem.

Apr 24, 2020

The other test hosts have ping results like this:

hci-tvm-vsanDatastore-7-1: 192.168.3.7
deploy-hcitvm-7.log-PING 192.168.3.7 (192.168.3.7) 56(84) bytes of data.
deploy-hcitvm-7.log-64 bytes from 192.168.3.7: icmp_seq=1 ttl=64 time=0.291 ms
deploy-hcitvm-7.log-64 bytes from 192.168.3.7: icmp_seq=2 ttl=64 time=0.161 ms
deploy-hcitvm-7.log-64 bytes from 192.168.3.7: icmp_seq=3 ttl=64 time=0.161 ms
deploy-hcitvm-7.log-64 bytes from 192.168.3.7: icmp_seq=4 ttl=64 time=0.196 ms
deploy-hcitvm-7.log-64 bytes from 192.168.3.7: icmp_seq=5 ttl=64 time=0.163 ms
deploy-hcitvm-7.log-
deploy-hcitvm-7.log:--- 192.168.3.7 ping statistics ---
deploy-hcitvm-7.log-5 packets transmitted, 5 received, 0% packet loss, time 22ms

Apr 15, 2020

How does HCIBench determine if the VSAN cluster is All Flash or Hybrid?
Is it possible that the information reported is wrong?
Example, vSAN AF Stretched cluster reported as Hybrid.
Thanks

Apr 15, 2020
Apr 16, 2020

Thanks for your reply.
Om maybe because one disk in the witness appliance is marked as HDD:

vSAN Disks:
HDD: Local VMware Disk (mpx.vmhba0:C0:T1:L0)
VMware Virtual disk 375 GB
SSD: Local VMware Disk (mpx.vmhba0:C0:T2:L0)
VMware Virtual disk 10 GB

Thanks again

Apr 16, 2020

Here is a quick workaround. Run these commands on HCIBench to use modified scripts that force the type to All-Flash.

cp /opt/automation/lib/easy-run.rb /opt/automation/lib/easy-run.rb.bak

cp /opt/automation/lib/get-vsan-info.rb /opt/automation/lib/get-vsan-info.rb.bak

curl https://raw.githubusercontent.com/cleeistaken/hcibench_miscellaneous/master/force_af/easy-run.rb --output /opt/automation/lib/easy-run.rb

curl https://raw.githubusercontent.com/cleeistaken/hcibench_miscellaneous/master/force_af/get-vsan-info.rb --output /opt/automation/lib/get-vsan-info.rb

Mar 11, 2020

Hi Chen,

I tested now the procedure and followed all the steps.
The VM are now created with NVMe Controller but there is an issue as they do not boot.
If I login to the console I can see an error that says "Boot Failed".
I think it may be related with the NVMe Controller on the GuestOS Disk.
I am going to replace it by the orginal controller and see if that helps.

Thanks

Mar 11, 2020

for the template, just want to ensure you only did vm compatibility upgrade and replaced the original one in vm-template folder, right?

Mar 11, 2020

That's right.
Upgraded VM hardware compatibility to 7.0 (vsphere 7), exported to ovf and replaced original disk-0.vmdk and the .ovf file. Kept the original names and permissions. Ruby files also updated.

Mar 19, 2020

Boot Failed usually means the boot disk got some issue.

but we should not touch the boot device since it should be still under pre-built pvscsi controller. could you help to verify that on the guest vm?

Mar 24, 2020

Hi Chen, I checked the controller on the GUEST OS disk and it has not changed. It is the original pvscsi.