DNS seed "seed.bitcoinstats.com" doesn't support filtering while the comments says it does #29911

Is there an existing issue for this?

• I have searched the existing issues

Current behaviour

It's stated in chainparams.cpp that "seed.bitcoinstats.com" supports filtering. However, this isn't true.

https://github.com/bitcoin/bitcoin/blob/c05c214f2e9cfd6070a3c7680bfa09358fd9d97a/src/kernel/chainparams.cpp#L137

Expected behaviour

Since "seed.bitcoinstats.com" supports filtering for NODE_NETWORK, when asked for NODE_NETWORK, it should return archival nodes. It doesn't.

Steps to reproduce

dig x1.seed.bitcoinstats.com.

[I] calvin@bitcoin ~> dig x1.seed.bitcoinstats.com

; <<>> DiG 9.18.21 <<>> x1.seed.bitcoinstats.com
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 60119
;; flags: qr rd ra; QUERY: 1, ANSWER: 25, AUTHORITY: 0, ADDITIONAL: 1

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 1232
;; QUESTION SECTION:
;x1.seed.bitcoinstats.com.      IN      A

;; ANSWER SECTION:
x1.seed.bitcoinstats.com. 60    IN      A       13.112.109.34
x1.seed.bitcoinstats.com. 60    IN      A       3.211.51.136
x1.seed.bitcoinstats.com. 60    IN      A       3.8.38.104
x1.seed.bitcoinstats.com. 60    IN      A       64.227.26.97
x1.seed.bitcoinstats.com. 60    IN      A       69.55.55.188
x1.seed.bitcoinstats.com. 60    IN      A       34.241.94.237
x1.seed.bitcoinstats.com. 60    IN      A       3.124.65.19
x1.seed.bitcoinstats.com. 60    IN      A       3.81.243.143
x1.seed.bitcoinstats.com. 60    IN      A       51.195.28.51
x1.seed.bitcoinstats.com. 60    IN      A       46.137.123.109
x1.seed.bitcoinstats.com. 60    IN      A       138.68.173.152
x1.seed.bitcoinstats.com. 60    IN      A       69.4.94.226
x1.seed.bitcoinstats.com. 60    IN      A       142.93.200.9
x1.seed.bitcoinstats.com. 60    IN      A       54.68.82.186
x1.seed.bitcoinstats.com. 60    IN      A       18.202.91.9
x1.seed.bitcoinstats.com. 60    IN      A       52.30.154.84
x1.seed.bitcoinstats.com. 60    IN      A       116.203.99.217
x1.seed.bitcoinstats.com. 60    IN      A       13.112.109.34
x1.seed.bitcoinstats.com. 60    IN      A       3.211.51.136
x1.seed.bitcoinstats.com. 60    IN      A       3.8.38.104
x1.seed.bitcoinstats.com. 60    IN      A       64.227.26.97
x1.seed.bitcoinstats.com. 60    IN      A       69.55.55.188
x1.seed.bitcoinstats.com. 60    IN      A       34.241.94.237
x1.seed.bitcoinstats.com. 60    IN      A       3.124.65.19
x1.seed.bitcoinstats.com. 60    IN      A       3.81.243.143

;; Query time: 312 msec
;; SERVER: 1.1.1.1#53(1.1.1.1) (UDP)
;; WHEN: Fri Apr 19 19:35:56 KST 2024
;; MSG SIZE  rcvd: 453

Of the returned addresses, these address do not support NODE_NETWORK

13.112.109.34
3.8.38.104
64.227.26.97
69.55.55.188
34.241.94.237
142.93.200.9
18.202.91.9
52.30.154.84
116.203.99.217
13.112.109.34
64.227.26.97
69.55.55.188
34.241.94.237

One can easily check bitnodes data: https://bitnodes.io/nodes/3.8.38.104-8333/

A handful of the nodes here were also unreachable at the time of this writing.

Relevant log output

No response

How did you obtain Bitcoin Core

Pre-built binaries

What version of Bitcoin Core are you using?

v23.0.0 but this doesn't matter here

Operating system and version

Also doesn't matter

Machine specifications

I don't think this matters for this issue

oh interesting! there's some discussion in #29809 about how to update the comments for DNS seed filtering. also cc @naiyoma

I guess I should add that I don't think the nodes returned by "seed.bitcoinstats.com" are good.

My assumptions is that it's just caching all the nodes it's previously seen and not checking if they're still providing all the services it previously has.

cc @cdecker

I guess I should add that I don't think the nodes returned by "seed.bitcoinstats.com" are good.

My assumptions is that it's just caching all the nodes it's previously seen and not checking if they're still providing all the services it previously has.

Indeed, I am checking the infrastructure now. And it.look like the crawler died. I'll see if I can get it back with reasonable effort.

I guess I should add that I don't think the nodes returned by "seed.bitcoinstats.com" are good.

My assumptions is that it's just caching all the nodes it's previously seen and not checking if they're still providing all the services it previously has.

I have stopped using these DNS seeds. I found only 2 useful when I last tested:

seed.bitcoin.sipa.be seed.bitcoin.wiz.biz

yea seeds resolve to nothing, almost 80% nodes are down

almost 80% nodes are down

Seeds are up but the results are down? Can't do anything they are hardcoded forever.

Do not use them. dnsseed=0 and fixedseeds=0

I have stopped using these DNS seeds. I found only 2 useful when I last tested:

seed.bitcoin.sipa.be seed.bitcoin.wiz.biz

I'm not sure the overall situation is that bad. It looks like at all times, at least two thirds of DNS seeds are returning ~70-99% reachable nodes.

Individual seeds/seeders have issues on occasion, but that's not really a problem if they're fixed eventually.

Since I'm monitoring the seeds anyway, I can set up some sort of alerting if that helps.

I have stopped using these DNS seeds. I found only 2 useful when I last tested:

seed.bitcoin.sipa.be seed.bitcoin.wiz.biz

These have a very low TTL set, which is fine.

The DNS seeds using the DNSSEC setup https://github.com/sipa/bitcoin-seeder/pull/85 will renew the DNS zone every hour and have a TTL of 1h set.

The DNS seeder https://github.com/sipa/bitcoin-seeder is only providing long running Bitcoin nodes. So caching for one hour is not an issue.

Just checking in on this. Seemingly nothing's been changed since posting about this back in April

So just to add a bit of context here, I was using my own implementation of the seeder in order to add a bit of diversity. However that scanner has been flaky for quite some time and I am having trouble to find the time to fix up the scanner. I am planning to get back to this, however I cannot give a timeline, so if this is a real issue, please feel free to remove the seed from the code, and I'll re-add it once it has been fixed up :+1:

Not sure if it's helpful to do anything in between releases, but if it's still broken in August, I'd suggest to remove it for 28.0 (a new seeder will be added in #30007, so the total number wouldn't change).

I spoke with @emzy and I decided to revive this seed, and migrate to a more standard deployment as well. The overhead of the custom solution did not add enough quality, and caused the seeder to rot and break. Using sipa's seeder will solve that, and reduce the burden of maintenance.

No timeline yet, but I'll dedicate a couple of weekends

Only bumping this because there's a good chance this seed will serve zero addresses by release time.

Seeder been stuck since January and has deteriorated from 37 down to currently 5 addresses:

» for addr in seed.bitcoinstats.com x9.seed.bitcoinstats.com; do echo -n "unique addresses from $addr: " && host $addr | grep address | awk '{print $NF}' | sort -u | wc -l ; done

unique addresses from seed.bitcoinstats.com: 5
unique addresses from x9.seed.bitcoinstats.com: 5

Only bumping this because there's a good chance this seed will serve zero addresses by release time.

Pristine node count is already zero.

I was a bit too late getting this one fixed, apologies. But it is working again now. I'll let it run for a while, let it settle in and prove it's stable before aiming for reinclusion.

What would be the best way for that? Should I open a new PR, or do we want to revert the commit removing the seed? And more importantly, do we even need / want an additional seeder? If not I'm happy to shut it down permanently and concentrate on other ways to assist the network 🤗

What would be the best way for that?

I think a new PR for re-adding is fine.

And more importantly, do we even need / want an additional seeder?

Opening an issue to gather opinions is probably the best way to have this discussion.