Normally my Tor relays get guard status after the first ~8 days. Since that wasn’t happening this time around I was trying to find the reason for that. I suspect the reason is the weighted fractional uptime. After only a few days of running the relays, the data center contacted me that they gave me the wrong IP block. The block already was used elsewhere (yes, I know…) so they had to stop routing it to my VLAN. It then took 3 days (weekend) to switch to the new IP block, hence Tor registered that the relays now have 3 days of downtime.
But what uptime is required for becoming a Guard? I have read it’s 98%, but if that’s the case, would that mean that it would take 150 days in total to be able to get to Guard on these relays? I made these relays to be guard relays (exit relays will hopefully follow soon-ish) and it would be bad if they:
Take this long to get to Guard.
Won’t have much wiggle room/slack for downtime (future security updates and enhancements, maintenance etc.) for the next ~2 years or so.
Assuming uptime (but correct me if I am wrong! ) is the cause, my question is: is there anything I can do to speed it up or to improve this situation? If it really takes 150 days and it will make maintenance tight for at least 2 years then it might be better to start over. But starting over would be a huge waste of bandwidth so I hope there is some other option.
In my case, I have got the guard flag after 15 days of uptime for my relay.
The first flag you get is the stable one, then after a while you get guard. If you have a significant downtime that occurs (a few days for example), you lose those flags and start from the beginning. If it’s light (a few hours), they automatically give them back to you after a few hours.
After you get the guard flag, there is indeed their verification process which is ongoing and will take more or less time.
This is a normal process since stability must be ensured on the network, it is a security if a relay is offline for too long, the network does not give it back the same status because it could be logically unstable.
Thanks for responding. So do I understand correctly that when you have significant downtime, afterwards you start from the beginning with the stable → guard process? That’s actually perfect because then my new relays should become guards fairly soon. I was afraid that 3 days of downtime at the start of the relay’s existence, would forever make the uptime stats mediocre and hence make it extremely time consuming process to get them to guard status.
Your relays will be fine. You have to make sure that they are online 24/7 and stable to not lose your flags. You should get them soon.
It is possible to restart them to do updates but it should not last too long. It is important that everything is up to date on your OS and packages.
Like you said, after a significant downtime, you start from the beginning with stable → guard. What you should note is that your relays are already registered and seen on the network so they won’t need to be known again (it doesn’t say that your relays are new in the metrics page). This will save time and make the process of assigning flags faster.
This is what I have experienced and seen since I have been an operator.
) is the cause, my question is: is there anything I can do to speed it up or to improve this situation? If it really takes 150 days and it will make maintenance tight for at least 2 years then it might be better to start over. But starting over would be a huge waste of bandwidth so I hope there is some other option.