Mysticism with an internal local network - what's going on?

Question

Windows Server

Hello!
Faced extremely mystical behavior within the same local network.
The crux of the matter is this.
There are 4 hosts on Windows Server 2008 R2 that are running on a Dell r610.
A separate VLAN has been created for them, the number, for example, is 16.
Ports on the Cisco 3750G switch - 44,45,46,47 - in Access mode, 16 VLAN. IPs from the network 192.168.36.0/24 are assigned
to each host : Actually, everything seems to work, but ... it does not work. And it doesn't work weird. 1. Pings go selectively: From host 36.2, ping reaches only host 36.4.
From host 36.3, the ping only reaches host 36.4.
From host 36.4 ping goes to 36.2 and 36.3.
From host 36.5, the ping does not reach any other host.
Ping error - timeout exceeded.
2. At the same time, from any host to any host, you can completely go to RDP o_O, even if there is no ping. I checked the firewall - the problem of not passing pings also occurs after the firewall is completely turned off on all hosts.
3. When you change hosts 36.3 and 36.5 addresses to 36.8 and 36.6, respectively, everything starts working fine, pings fly everywhere, RDP works, everything is ok. But the existence of such a situation is annoying.
What was done:
1. I tried to add a separate subnet on the switch 192.168.36.0/24, and also wrap everything through the gateway 192.168.36.1 - the picture is the same.
2. I looked at the arp table - everything is fine, on each host in the table there were addresses of other hosts, and their poppy addresses were correct.
3. As I already mentioned, I cut down the firewalls - the picture is the same.
4. Switched the subnet to 192.168.40.0/24 - as a result, all hosts began to see each other, except for two - which saw each other after reboot.
5. Switched ports to a trunk with native vlan 16 - the picture is the same.
6. Used another switch - Dell PowerConnect 5524 - the same picture.
7. Added an additional address in the parameters of the network adapter - the ping went to the extra. address, but did not go to the main one.
8. I tried other ports on the switch - the picture is the same.
Actually, the question is - what can it be and where to dig (under Windows, under a switch or somewhere else)?
In general, the rest of the network works fine, the switch is 100% working.
The strange thing is that some specific addresses on the network stop responding and cannot be reached.
Thanks in advance for any advice!

Answer 1

Pings may well be cut by the firewall. It is possible that permissions for ICMP are not correctly configured in the Firewall everywhere. Or the servers see the network as public and ICMP is only allowed on the private/domain network.