Spark 2.8.3 Windows 10 and VPN and disconnecting sessions (while Windows 7 and 8 clients do not and connect normally)
For months we had troubles with Spark 2.8.3, Openfire 4.2.3 and Windows 10 clients. Clients were connecting via VPN to our datacenter to the Openfire server. While Windows 7 and 8 clients did connect normally and had a stable Saprk Openfire connection, Windows 10 clients disconnected without warning. Once these clients were sent a message or tried to send a message with Spark, it appeared disconnected.
Solved temporarily by entering a line to our Windows hosts file:
This can be done by editing the file with Notepad or any text editor (as Administrator)
Situation, all with sample data:
Openfire Domain: im.yourdomain.com
Openfire internal LAN IP in datacenter: 192.168.114.115
Openfire external WAN IP in datacenter: 184.108.40.206
VPN: Clients connect via VPN and they get a random ip in the range of 10.10.9.1 - 10.10.9.254
Strange behaviour seen in Openfire sessions:
Windows 7 and 8 clients connected via Spark to Openfire, appeared in the sessions list with an IP address 10.10.10.x (VPN).
Windows 10 Clients arriving at Openfire, were listed as having an external/WAN IP of our office, e.g. 97.120.130 instead of the VPN IP address e.g. 10.10.10.4.
Line added to ‘C:\Windows\System32\drivers\etc\hosts’:
# Openfire internal LAN IP in datacenter / (sub)domain Openfire
Now suddenly Windows 10 clients started to arrive at Openfire with their VPN IP addresses.
Spark 2.8.3 didn’t disconnected anymore and stayed as stable as it was under Windows 7 and Windows 8.
As said, it took us a long and frustrating time to find anything that fixed the major issue. I’m not technical enough to relate this behaviour to Windows 10 or to Spark 2.8.3 or to Openfire 4.2.3 or to the combination. Since the discovery of the external/WAN ip versus VPN assigned IP, I have seen that Outlook 2016 has similar problems with IMAP and syncing/slowliness.
Hope this will be useful to anyone.