I'm facing a strange problem and checked already a lot but maybe u will give me some good addittional thoughts.
We have a server in an openstack enviroment virtually running quiet fine. After a while, today it was after 72h, sometimes after eben less (24h e.g.). The External SAP ist not able to connect to that Servers DB to pull the data.
When I ping the SAP from that DB Server, SAP is again able to pull immediately..
I have a small business switch SG550xg between Openstack and network. The problem I'm not able to check the customer network means router and SAP cloud connection.
Before where the DB was baremetal installed, it was working quiet fine, but also there we hade the application server and DB server on the same system. Now it is separatet...
I expect ARP issues, but don't really imagine why..
I guess it can't be a routing issue..
maybe a state issue on the customer firewall, but why is ping triggering the state for another protocol.
with the ping I was able to have a stable situation for days..
maybye you have some good thoughts..
what makes you think this is a network issue and not a server one?
it could be that for some reason these server nics go into a hibernation mode have you checked for that?
so just to clarify the application server which is the sap and DB server were on the same system but now they are separate and since then your experiencing intermittent connection problems?
have you been able to capture the traffic between these to servers?
What does the interface statistics show for these devices?
The topology here is not clear to me. You describe a server and an external SAP. Can you clarify how they are connected and which parts of that connection you have visibility to and which parts you do not?
The symptom of failing connectivity which is resolved when the server pings the SAP suggests that something is timing out which prevents access from SAP to server and is restored when server pings SAP. Perhaps it is an arp entry or perhaps a mac address table entry?
The mention of VPN is interesting. Can you tell us more about it? Is it possible that the VPN is configured in a way that requires that the VPN be initiated from your end and not from their end? (That can be the result when one of the VPN peers has a dynamically assigned IP address and there are some other reasons but dynamic IP is probably the most common)