Trey Nolen wrote:
First of all, thanks for running a relay.
Based on my experience, what usually happens is that the provider of your VPS observed during a period of time you used more than N mbps constantly and all the time, so they capped your VPS at some KB/s limit. There are performance monitoring scripts that could do this automatically. A virtual private server shares the network card of the host with the other VPSes on that host, so almost all providers do not allow you to use it all by yourself all the time for long periods. You can open a ticket upstream and they will confirm if this is the case or not.
Nothing you can do about this unfortunately, most providers do this, even the ones they say they don't do it :) Only thing you can do is get a dedicated server with guaranteed bandwidth, or try to convince them to at least lift your the limitation for your VPS to 1mbps.
In this case, this is not going on as we *are* the provider. I'm a sysadmin on the network and I'm one of the guys that would be in charge of limiting any machines which violated any rules. :-)
Trey Nolen
Oh, OK. Glad to see not everybody who rents virtual private servers also caps their bandwidth. It happened to me so many times that I could even bet that this was the issue here, but looks like it's not.
Since atlas is down, I have looked at the votes here: https://consensus-health.torproject.org/consensus-health.html
and it looks like your relay has a measured by authority 'bastet' of 355. That is not a big value. The other authorities measured this:
278; 355; 367; 803;
So it looks like the speed was pretty much the same for the measurements performed on your relay by different servers on different networks. If you say you are sure there is nothing automated (at either OS level, hypervisor level, local router/network level or something upstream) that could throttle this in case of continuous high usage, there's not much you can do other than waiting some time to see the next speed measurements.