Hi everyone,
Could anyone tell me what the maximum acceptable latency is between Pritunl and MongoDB Atlas for stable performance?
I need to set up several hosts in different locations, and the ~1000ms latency I’m seeing to the database is quite concerning.
It shouldn’t be an issue for the hosts that are not being used for access to the web console. The Ashburn area datacenters are typically best for global deployments.
@zach
I am using Pritunl in the EMEA region. The MongoDB Atlas M10 cluster is in GCP Frankfurt (europe-west3). The host with web console access is in the neighboring Warsaw region (europe-central2), and no issues are observed there.
However, other hosts located at a more significant distance from MongoDB appear to be quite slow, even though identical resources have been allocated to all of them. Connecting to the VPN on these hosts can take up to ~40 seconds.
Additionally, at a certain point, the VPN servers on these distant hosts start behaving strangely, and I see a large number of “Server stopped unexpectedly” logs in the dashboard.
[ERROR] Error in management socket thread
Traceback (most recent call last):
File "/usr/lib/pritunl/usr/lib/python3.9/site-packages/pritunl/server/instance_com.py", line 341, in _socket_thread
self.connect()
File "/usr/lib/pritunl/usr/lib/python3.9/site-packages/pritunl/server/instance_com.py", line 428, in connect
self.sock.connect(self.socket_path)
ConnectionRefusedError: [Errno 111] Connection refused
Typically it’s better to use the Ashburn/DC region datacenters even if it’s further away from the majority of the servers. There are optimized paths to this region from most of the world.
That error is likely from an error in the OpenVPN process this would be shown in the server output.