Intermittent Connection issue usng Hanwha Wave Cloud proxy

We have issues below as best I can describe and are looking for some suggestion and ideas where or what might be the issue.

We have integrated Hanwha into our CMS platform and connect via the cloud proxy link https://xxx-xxx-xxxx-xxxxxxx.relay.vmsproxy.com   for example.

The API we have been using until very recently was V2 but have added support for V3 and V4 where possible and that is being released in the next few days. To connect to the systems and get all cameras, the NVR is added to our Azure environment, and connection tests can be carried out from there. 

The NVR ‘s on site are 99% v6.05 with a couple on 6.03 and 6.02.

The issue appears in the following 2 scenarios.

1. Verified Connection Loss

• We create a generic alarm every 2 mins from a windows task that posts the word “ Heartbeat “ as the caption , and we then send an email via our secure SMTP to our dedicated receiver and process in our backend as a “keep alive / Heartbeat” 
• When we fall to get heartbeats for 5 min , we start a process of checking 3 times with a interval and then raise this as verified connection loss to our specialists.
  • So at that point we have not heard from the box , nor can we connect in via the API .

• When specialist then tries to connect using out from our platform we fail to connect , when I then try from our Azure integration , that fails , when I add the Url to a local web browser on my PC it fails with a error 502.

1. Alarms come but are blank meaning we cannot reach back and get images.

So what we know and can test currently are 

• When we stop getting Heartbeat alarms , this starts the check system and this lasts for between 5-10 mins 
• During this point it seems we cannot connect via the cloud proxy on our Azure integration , Postman tests, web browser .
• The connection always restores itself after several minutes .

I did see s p2p failure notice on one of the NVR logs this morning , we did also run the test tool to show open ports . I have attached a before and after .

I am wondering if there are any known improvements that might explain this in either firmware 6.1.2 or the V4 API