VMmark

 View Only

 VMmark4 run fails with "Error: STAF Ping machName = '

schris14's profile image
schris14 posted Sep 18, 2025 10:58 AM

VMmark4 run fails with "Error: STAF Ping machName =  '<VM>' : Failed" on all VMs in tile0.

I am running VMmark 4.0.3

I have verified that the Primeclient's /etc/hosts file has all VMs listed correctly. 

I have verified that all VMs are powered on and running and that I can ping from the PC to each VM via IP address and hostname.

I have attached the VMmark4-STAX_Job_3_User-Parsed.log file.

Lisa Roderick's profile image
Broadcom Employee Lisa Roderick

Hello. The harness is attempting to ping the VMs through STAF, which is the benchmark engine the VMmark harness uses to communicate with the VMs. So though ping over TCP/IP works, something isn't right with the STAF configuration. After you've rebooted the prime client and all the VMs, please issue the following and post the output:

  staf ds35weba0 ping ping

Also please zip the contents of the results directory and post here. Thanks. 

Rebecca Yo's profile image
Broadcom Employee Rebecca Yo

In addition to the results directory that Lisa asked for, could you please include these files as well and then zip and attach to the forum: 

From the Prime Client, the files /etc/hosts and /usr/local/staf/bin/STAF.cfg

From any one of the workload VMs, the files /etc/hosts and /usr/local/staf/bin/STAF.cfg

Thank you!

schris14's profile image
schris14

I rebooted the primeclient as well as all workload VMs.

root@primeclient:~# staf ds35weba0 ping ping
Error submitting request, RC: 22
Additional info
---------------
STAFConnectionReadUInt: select() timeout: SSL_read() RC=0 SSL error: error:00000000:lib(0):func(0):reason(0): 22
root@primeclient:~#

Attachment  View in library
Rebecca Yo's profile image
Broadcom Employee Rebecca Yo

Hi, thank you for providing the files. In the VMmark4.properties file, the parameter PrimeClient is set to the Prime Client's public IP address (10.61.60.31) when it should be the private IP (such as 198.18.4.251). As a result, the workload VMs' hosts files had the Prime Client's public IP in them which was unreachable.

The Prime Client has two virtual NICs and two IP addresses: A private IP (such as 198.18.4.251 by default) and a public IP (Yours is 10.61.60.31). The private IP is used to communicate with the workload VMs and clients. The public IP is used to communicate with vCenter Server.

Since your workload VMs use the default IP range starting at 198.18.4.1, I recommend using the prime client default private IP 198.18.4.251.

In VMmark4.properties, please update the PrimeClient parameter to 198.18.4.251 or leave it commented out.  It looks like you provisioned the VMmark tile using the VMmark4.properties file. You will need to delete and reprovision your VMmark workload VMs after you change the VMmark4.properties PrimeClient parameter.

When you set up the Prime Client, did you also run make-prime.sh with the parameter 10.61.60.31? If so you will need to delete and recreate your Prime Client and run make-prime.sh using the private IP address. 

Thank you and please let me know if you have any further questions.

schris14's profile image
schris14

I rebuilt the Primeclient since I used the Public IP address in the make-prime.sh command.  I left the default IP address in the VMmark4.properties file and that got me past that set of errors...I ran into more but was able to get a benchmark run to start for tile0.

Thank you for your help...no other assistance needed at this time.

Rebecca Yo's profile image
Broadcom Employee Rebecca Yo

That's great, thank you for the update.