miechovy
Enthusiast
Enthusiast

Multiple VMs - network issue

I'm using Vmware Workstation Pro since few years, since v. 14 has been available. Till today I notice same problem (v. 16).

Namely when I run multiple VMs (for example 10 or 12) at the same time, I have a problem with network connections. A command PING has a response, but PING to domain is not resolving. Next, if I close most of VMs, the network is working properly.

It turns VMware has a network problem with multiple VMs runs at the same time. I know - maybe VMware Pro is not dedicated for such of environment, but we have to use Workstation Pro to our project.

My question is, how can I debug or where to search information about this issue? Is it a problem with VMware DNS or something? (my network in VMs are configured as a NAT).

>ping google.com
Ping request could not find host google.com. Please check the name and try again.

Tags (3)
0 Kudos
13 Replies
wila
Leadership
Leadership

Hi,

Is this in the guest or host?

--
Wil

| Author of Vimalin. The virtual machine Backup app for VMware Fusion, VMware Workstation and Player |
| More info at vimalin.com | Twitter @wilva
0 Kudos
miechovy
Enthusiast
Enthusiast

The problem is in the quests. My host is installed on Linux and has a good hardware performance.

0 Kudos
wila
Leadership
Leadership

Hi,

I thought you would say that, just wanted to make sure.

Yes, you are correct that it is likely to be a DNS issue.

Next thing to try is what the result is of this:

nslookup google.com

Note: run in the guest of course (I should also be explicit)

--
Wil

| Author of Vimalin. The virtual machine Backup app for VMware Fusion, VMware Workstation and Player |
| More info at vimalin.com | Twitter @wilva
0 Kudos
miechovy
Enthusiast
Enthusiast

>nslookup google.com


DNS request timed out.
timeout was 2 seconds.
Server: UnKnown
Address: my_ip_address_cutted

DNS request timed out.
timeout was 2 seconds.
DNS request timed out.
timeout was 2 seconds.
DNS request timed out.
timeout was 2 seconds.
DNS request timed out.
timeout was 2 seconds.
*** Request to UnKnown timed-out

0 Kudos
wila
Leadership
Leadership

Hi,

So yes, it is a DNS issue.

You mentioned you used a NAT connection on a Linux host.

But you also said "if I close most VM's then it works normal"... that one is weird.

Are (some of) your VM's using a fixed IP address perhaps?

--
Wil

| Author of Vimalin. The virtual machine Backup app for VMware Fusion, VMware Workstation and Player |
| More info at vimalin.com | Twitter @wilva
0 Kudos
miechovy
Enthusiast
Enthusiast

Hi.

IP addresses are assigned automatically, but a DNS IP direct to local address to the host (my DNS server).

0 Kudos
a_p_
Leadership
Leadership

Some thoughts:

  • What operating systems do you run in the guests/VMs?
  • Have the guests been cloned, or manually created?
  • Can you confirm that all guests have different MAC addresses, and - in case of Linux - different machine-ids?
  • Does the name resolution work with querying the DNS server directty, e.g. nslookup www.vmware.com <DNS-Server-IP>, or dig www.vmware.com @<DNS-Server-IP>?

André

0 Kudos
miechovy
Enthusiast
Enthusiast

Hello Andre.

  • What operating systems do you run in the guests/VMs?

Windows 10.

 

  • Have the guests been cloned, or manually created?

Full cloned and works as an independent instances.

 

  • Can you confirm that all guests have different MAC addresses, and - in case of Linux - different machine-ids?

Yes - the MAC addresses has been randomly generated after cloned.

 

  • Does the name resolution work with querying the DNS server directty, e.g. nslookup www.vmware.com <DNS-Server-IP>, or dig www.vmware.com @<DNS-Server-IP>?

No, we use a localhost DNS (configured to 8.8.8.8 for resolving domain names) - unavailable from external network.

 

What do you mean - different machine-ids?

0 Kudos
a_p_
Leadership
Leadership

The reason why I mentioned "machine-id" is related to DHCP with current Ubuntu version (see e.g. https://kb.vmware.com/s/article/82229).
What you may try to further troubleshoot the issue is to run s.th. like portqry to check whether the DNS server's port 53 (tcp/udp) can be reached.

André

0 Kudos
miechovy
Enthusiast
Enthusiast

I'm not trying to say you are right or not, however I mentioned this issue has a place when I run dozen VMs, not a few. I think it is a problem with performance some thing. But I do not know how to determine, it concerns VM problem or my custom DNS server problem on Linux platform.

I just checked and all VMs has a different IP addresses.

0 Kudos
a_p_
Leadership
Leadership

Please don't get me wrong, it's not about being rigth or wrong. I'm just trying to rule out possible issues.
What I'm actually thinking of at this point, is that VMware Workstation might have an issue with UDP traffic - I understood, that e.g. ICMP (ping) is still working in affected VMs - in your case (many powerd on VMs), but I'm currently not sure how to figure this out.

André

0 Kudos
miechovy
Enthusiast
Enthusiast

The communication between VMs is correctly, I suppose. Command of PING from quests to host and host to quest and quest to quest is working properly. As you know right now the problem is when multiple VMs is running. Ping command probably is not a problem, rather communications to local DNS and back. For example in first post I write that PING is correct, but domain name is not resolved from command line or browser.

I noticed that something is "stuck" - sometimes DNS works, but mostly machines cannot get access to internet.

0 Kudos
a_p_
Leadership
Leadership

Can you please try the following on a VM that cannot resolve names?

Start nslookup in interactive mode, and configure it to use a virtual circuit for requests. After that enter an URL to see whether it gets resolved.
With vc enabled, DNS queries are supposed to use TCP rather than UDP.

Sample:
nslookup
> set vc
> www.vmware.com 

André

0 Kudos