wardmundy
Nerd Uno
- Joined
- Oct 12, 2007
- Messages
- 20,217
- Reaction score
- 5,974
We noticed over the weekend that five of our VPS deployments at SSD Nodes had lost OpenVPN connectivity. Upon further checking, it appeared that the OpenVPN client was running on all of the machines, but none of them were activating a TUN interface with a private VPN address.
We opened a ticket on Sunday and got the usual runaround that it was something at our end, and they hadn't touched anything. I chalked it up to typical Sunday help hoping for a solution today. Instead, I got this:
And my response including a current listing of /var/log/openvpn/status.log showing dozens of functioning OpenVPN Client connections :
BOTTOM LINE: If you're depending upon SSD Nodes, DON'T!
We opened a ticket on Sunday and got the usual runaround that it was something at our end, and they hadn't touched anything. I chalked it up to typical Sunday help hoping for a solution today. Instead, I got this:
These are all KVM virtualization and TUN/TAP will need to be disabled on each VM separately. This is not something we have done on any of the nodes your VMs are hosted.
This is not a problem on our network or our node.
I can only assume this is related to some updates that happened on the OpenVPN software.
And my response including a current listing of /var/log/openvpn/status.log showing dozens of functioning OpenVPN Client connections :
I would agree with you that this might be because of some change on the OpenVPN side except for the fact that we have this running reliably TODAY on Vultr, Digital Ocean, Oracle Cloud, Raspberry Pi, HotLine, Hostodo, RackNerd, HostMediaUK, Virmach, CrownCloud, HostFlyte, 3CX, and even CloudAtCost. And the platforms include CentOS 7, Ubuntu 20.04, Rocky8, and Debian 10. As you can probably surmise we, too, do this for a living and have for about 40 years.
So... either SSD Nodes installed new kernels or did something in the last week or so that broke all of these servers simultaneously. Fortunately, we use your platforms primarily for backups so this didn't put us out of business. As I previously noted, we haven't touched the software on most of these machines in over a year so this is most definitely because of some change at your end. The fact that nobody else has reported it might mean (1) very few of your users rely upon OpenVPN or (2) they haven't yet noticed.
BOTTOM LINE: If you're depending upon SSD Nodes, DON'T!
Last edited: