Are there any explanations about why the full node count (at current block height) has dropped so low?
We’re at around 130-135 nodes, whereas that count had tracked closer to 170 or more for the past year or two. Can any of the infrastructure engineers weigh in about this Blockchair site? (Are they accurately tracking all of the nodes on the network?)
Is it possible that people are deciding to pull their full node off the network, or has something technical happened that might be causing the decrease?
I think more folks are running nodes. If I had to guess, Id say blockchair is not accurate. The good news is ZF is making it much easier to others to spin one up!
The only way that I know to do it for normal users is to run the Ziggurat crawler and I’m betting it will likely find roundabout the same number, but it does vary depending on the time and the seed ip’s etc.
Also, you’ll have to link these 170 figures. Not saying that’s exactly wrong but I don’t quite remember that post NU5. Or maybe that’s what you’re referring to like what dude says with the Spam? Yeah that def
this is very very big step to get more nodes. also zebra node runs so lightly on windows also now, im just syncing up one extra. but i think node runners should be incentivised also somehow a bit more also.
Thanks for running that crawl. Among the magicbeans does it count if they’re within a certain proximity of the highest height?
~170 was just off the top of my head, I check the network stats once or twice a month, and for a long time we were sitting around 150-170 nodes up near the current height. (Like you mention, I’m reflecting on just the past couple years)
It specifies them against the protocol version which is 170100, I think with just handshakes. I don’t think it actually requests any state information like if it’s currently syncing or fully synced, I’ll have to check.
My other question is related to what tool can I use to scan the Zcash network for the number of nodes?
And another question has to do with what are some ways to hide yourself nowadays. Especially interested in tools built into the node to hide their metrics. This could be a very relevant topic in the future, if in some jurisdictions node holders will expose themselves to the risk of penalties from the state.
Thanks in advance to all who will join this discussion.
This website publishing those IPs should consider not doing that, but the IPs of nodes are necessarily public so with a bit of effort anyone can find them.
Last time I did that on another network, I had setup a node on each continent, ran ss on a regular basis while logging the output. After running this for a month or so, I have merged the logs and extracted the relevant IPs. There’s certainly a better way, particularly as this method does not guarantee to map the entire network, but that was good enough for my needs at the time.
It really depends of what you are trying to achieve. If you want to run a node to help the network, then just run one in a datacenter in a jurisdiction where it is allowed. If you want to run a node to get better performance when you need to query the network, while in a place where it is illegal to run one, you could just run one behind a VPN. Alternatively, you could run it behind Tor.
On the Tor point, I’m pretty sure most people are using it in a way that can easily be detected, which may actually increase their risk. If you go that route inform yourself on how to do this well. Maybe a starting point here.
As another datapoint here, the DNS Seeders track as many nodes as they can find, and identify nodes that they consider “good” for advertising to others (via several metrics, such as consistent long-term uptime and reachability). The actual number of full nodes in the network will be higher due to e.g. nodes that are not externally reachable, but have themselves connected into the network.
The seeder I operate currently reports around 100 “good” nodes. It’s been stable around this level for the past year (I don’t keep historic records, but checked Slack for times I pasted its output into a channel).
Those nodes are on the same magic but use a different protocol version from 170100 and 170110, so ignore them. Crawl from this past weekend showed about half of zcashd’s were upgraded to 5.10.0 from 5.9.1 and 11 zebrad 1.9.0’s.
The bigger concern is still 58% of boocks come from the same pool minerstats If that pool doesn’t agree with a protocol change, the change doesn’t happen.
