OpenVPN CPU Usage after WAN outage
-
Running 2.2-BETA (i386)
built on Fri Oct 17 11:29:18 CDT 2014Had an outage on my WAN connection this evening for a few minutes. When the connection came back up again, my OpenVPN client connection spiked the CPU and stayed there. A connection to YouTube (not routed over VPN) was disrupted until I restarted the client. An RDP connection, also not routed over the VPN, recovered before the client was restarted. IMAP and SMTP traffic was disrupted until the restart.
I'm sorry I don't have much in the way of detail. I understand this is a work in progress. Just reporting what happened. I can provide logs if needed - let me know which ones.
-
As and when this happens again, do the unresponsive connections come back when you kill the states to these services?
Diagnostic, States, select the ip address that links to one of your unresponsive connections and delete it.Does the connection come back and behave like normal after a period of time?
This sounds a bit like what happens to asterisk users when the wan ip changes, hence the above.
-
Interesting observation. I do have multiple wan connections, but there is at least one issue with routing traffic out the correct interface / gateway, so I'm not currently configured to failover / switch between WAN links. The observed behavior all happened utilizing a single WAN link.
I will test killing the the states for the frozen connections next time I see this. Thanks for the suggestion.