Seeing a LOT of timeout messages and other strange behaviour

Support Relay Operator
1

I am seeing some very strange behaviour on one relay:

So far today 32 of these:

│ 15:11:54 [NOTICE] Your network connection speed appears to have changed. Resetting timeout to 60000ms after 18 timeouts and 943 buildtimes.

Almost daily one or two:
Guard ($hex) is failing a very large amount of circuits. Most likely this means the Tor network is overloaded, but it could also mean an attack against you or potentially the guard itself. Success counts are 93/217. Use counts are 83/83. 93 circuits completed, 4 were unusable, 0 collapsed, and 3 timed out. For reference, your timeout cutoff is 60 seconds.

I’ve had several messages about adding a third guard because two were having problems.

There have also been issues maintaining the stable flag, with strange losses of it that I can’t explain - no down time and no issues with the descriptor.

There is some reason to expect attempts at de-anonymizing hidden service users of the node. I am curious if this could be attempts at that.

2

This is happening on 100% of my restarts. I delete /var/lib/tor/state to force new guard selections, restart tor, and on restart the two guards it selects are showing the same warning right at startup.

The node has significant issues with incoming onion services connections, with about half failing.

I am beginning to suspect my provider is doing something odd with my VPS’s outbound connections. This would be odd, since my provider is a well known Tor friendly operator.

Any other ideas? I am running out.