Getting Annoyed

My Tor Relays are encountering problems, with instances where they disconnect from Tor Metrics seemingly at random. I’ve gone through various troubleshooting steps, such as examining nyx and error logs, and even reaching out to my Host Provider through multiple support tickets. According to them, everything appears to be functioning correctly on their end, including open ports for Tor. I diligently followed the guidelines outlined in the Tor guide to set up the relays.

Interestingly, upon restarting the relays, everything functions normally. I’m quite puzzled by this situation and find it increasingly frustrating. I believe a one-on-one remote support session would be immensely helpful, as I’m becoming quite frustrated and even considering pulling my hair out trying to resolve this.


When you say “disconnect from Tor Metrics”, what does it means? Metric tell that your relay is offline (but in fact, still online and available)?

Does your relay have enough resources? Is it shared with any other service?

One of my relays is running on a free tier of a cloud provider, enough for a small/medium relay and I recently faced some issues with the daemon consuming 100% of CPU. That wasn’t normal as after a few days running CPU would be around 40-60%. Only thing that worked was recreating the instance with a fresh install.

If logs aren’t configured properly you may miss notices/warns about the relay being unresponsive. I’d rather manually check logs than looking through nyx.

Does your relay earn the Guard or HSDir flag after some time? This ramps up traffic and may crash tor or become unresponsive if no enough resources are available to handle it. One way to solve it is by limiting the advertised bandwidth with MaxAdvertisedBandwidth, that will limit the amount of circuits handled.

I suggest you to look at measurement results for your relay by authorities at this page: Consensus Health.
If you see vote from some authority for your relay without Running state, then you can try pinging authority address from address of your relay. More precise check is to try connecting to authority. If such checks fails, then you would have proof for ISP that something is wrong.

i upgraded one of my relays that is running windows 10 and it seems to be running fine till it restarts then i have to go in and restart it

as for the two others they are running on VPSs the HP is
the stats don’t show 100% CPU use good on Memory just shows a solid 100% but at that usage
before i had a reinstall it ran well for more then 30 Days

shows offline

Also here are my nodes

I had the same issues with some of my relays at Ionos.
Problem was that relay VM lost its IPv6 address.

Are there any errors reportet in the Tor logs in your machine?