Networking anomaly??

[Amit]

Hello guys,
I'm quite new to Proxmox but I love it. I am testing a few Hypervisors (7 to be exact) and each is running a few VMs. They all make up a VoIP cluster. I'm running VE 5.1-35 and all VMs are linux containers, CentOS 7.4 to be exact.
The Hypervisors and VMs all have public IPs, protected by cisco firewalls.
I am testing the failover capability of the VoIP cluster, but I am extremely confused about the results.
In short, if I stop a service on one of the machines, I see the cluster uses another VM which provides the same service, and the cluster remains operational. Additionally, if I stop the network interface on the VM (simulating a network outage), I also see the cluster use another VM which provides the same services, and the cluster remains operational. However, if I simply power off that same VM, the failover does not kick in and the cluster breaks down. Is there something about the network bridging that I am missing here?
I am sure this is not nearly enough detail for you to provide any insight, so please let me know what other details you need, but keep in mind that I may need some guidance on how to retrieve that info

PS, I would LOVE to do the 4 day course, but it is never offered in the US or Canada, and it only seems to be offered in German or French.

Thanks in advance for any suggestions.
Amit

 

[wolfgang]

Hi,

do you see with tcpdump something?

 

[Amit]

hi Wolfgang, that was going to be my next step, since all other diagnosis attempts have not yielded much. I was wondering if there was something unique I was not considering using proxmox and vmbr0, since the issue only happens when the VM is actually powered down, and not when the services are disabled. I was thinking perhaps something along the lines of some kind of routing table that is not updated or something?
Very bizarre symptoms!