Why are 30% of nomadnet site inaccessible
Started by welo ·
If you click random sites in the announcement stream you'll only be able to connect to about 60-70% of them, with the rest getting stuck at establishing connection, I always assumed this was just network instability but some nomadnet sites like Rmap.world or the official rngit for reticulum are generally always connectable which makes me think it's just badly configured instances. Does anyone have any idea which is the case
You could maybe distinguish between the two possibilities with rnprobe or something similar. Would be interesting to see the statistic of which sites respond to probes and which you can connect to properly
aetherlab wrote:
Or they are just not online all the time.
true in general but for nomadnet sites I see through announcements that must either mean that the nomadnet application has been started very recently or its resending an announcement after X amount of hours, either way that implies that the computer its running on is online.
It could be from people who have a nomadnet site but don't have a background daemon so if they open nomadnet for like 10 seconds to check a message the site goes down afterwards but I couldn't imagine that accounts for all the unconnectable sites.
I can't prove this but nomadnet sites that are smaller than around 300~ bytes and thus fit into one packet (I read this somewhere but can't remember the source) are significantly snapper and don't seem to suffer this issue ever of randomly going offline.
I feel like I could configure my interfaces to be more performant too. I read on somebody's nomadnet site that they were getting timed out when trying to load pages, and changing their interfaces helped some.
welo wrote:
aetherlab wrote:
> Or they are just not online all the time.true in general but for nomadnet sites I see through announcements that must either mean that the nomadnet application has been started very recently or its resending an announcement after X amount of hours, either way that implies that the computer its running on is online.
It could be from people who have a nomadnet site but don't have a background daemon so if they open nomadnet for like 10 seconds to check a message the site goes down afterwards but I couldn't imagine that accounts for all the unconnectable sites.
I have a bit of a hunch that it's more due to announce rate limiting at various points across the hops. You're actually seeing an announce from a while ago, even though it shows to you and your client as coming up just now. So it's not a perfect presence indicator the way you're using it, I'm afraid
To anecdotaly add to what AKAFrosty said above, I have wondered if some sort of limit was being reached. I notice that links stuck on establishing occur more at certain times of day than others. There could be a combination of factors here, but if I am on late / early morning European time, and there are far more announces appearing in the list (assuming N/S Americas, Canada etc.) then the rate at which I can successfully establish connection with node index pages drops significantly.
I suspect that some backbone nodes have additional barriers in place that block or limit access from other backbone nodes. Like fail2ban or other automated throttling/blocking systems.