Re: Eisner/Encompasserve down

On 8/30/2024 8:12 AM, Simon Clubley wrote:

 I don't know if it's the TCP/IP stack in use, or the applications
running on top of that stack, or something else, but a VMS system
with full exposure to the wider internet does not appear to be a
robust system.
 

It's the design of the filter server in MultiNet. It was clearly never designed to handle cases of extreme, prolonged attacks on a system. Not too surprising; when that was written, such attacks were extremely rare. Now they're not.
I'm working to address that design flaw in the filter server. It currently uses lots and lots of TQEs to manage filter entries and timeouts, which has proven to be problematic for these sustained attacks.
In the meantime, I just stopped trying to filter the bad IPs, and the system has been running fine for weeks now. (I did reboot it 11 days ago, but that was to reboot the underlying host system.)
(I don't check comp.os.vms every day, and the INFO-VAX feed to EISNER is broken right now somewhere along the path, so I didn't see these until now.)
Hunter