have you noticed nostr got a lot quieter in the last weeks? if you did, its probably because your notes are being rejected by naieve throttling on ALL the big relays.
you wont notice it when you post because clients dont surface this info to you, for all you know just nobody liked your post, but the reality is it did not post.
🍄
Really?
yes really, i tested it and got a 30% success rate
Well that seems shitty
this is why i've been an avid subscriber of paid relays... i now also rent one off you and i get confirmations off all three of them almost every time
though the other day i had amethyst totally eat two notes i wrote on my phone, so idk, there could be some factor of clients in this problem
most def amethyst is the most hurt rn, cause it spams drafts
amethyst isn't keeping a local cache?
if you use a local relay it can but i dont think it sends just to that.. from what ive seen they go everywhere. i turned it off by disallowing it from amber signer.
i meant, in-app drafts cache
cos i had two notes disappear the other day, again, was the last straw for me, and an excuse to use my mobile even less
While technically a protocol that can decentralise we seem to be centralising... weird.
People do realise we actually could always decentralise right? Remember forums & irc?
i member
I was literally thinking about exactly this today.
Yes it has felt different. What’s being throttled? Or to ask another way, what are they doing when they’re throttling?
nostr's growing too big for free relays without funds to expand infrastructure
people like me and nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyv9kh2uewd9hj7qghwaehxw309aex2mrp0yhxummnw3ezucnpdejz7qgwwaehxw309ahx7uewd3hkctcpz9mhxue69uhkummnw3ezuamfdejj7qg6waehxw309a5xjan9w3skc6ewdehhxarjxyhxxmmd9uq35amnwvaz7tmwdaehgu3ww35x2umpd4jkxct59e5k7tcprdmhxue69uhhg6r9vehhyetnwshxummnw3erztnrdakj7qpq0npj3gydmv40m70ehemmal6vsdyfl7tewgvz043g54p0x23y0s8qm9unzk have been banging on about this for ages because we are both experienced with systems administration and structural limitatations and budget problems
well for example, damus relay now only allows a couple messages per minute, that incudes ALL kinds, zaps, likes, follows, profiles, replies.. drafts, for sure would burn this before you even post.
How would a user be able to know if/when a relay is throttling?
clients usually cover up the failure.. but most relays will respond back and say why its not currently accepting a particular note so that you would know to retry.
This is why I post to the moonboi relay
#🌝
Which paid relay do you suggest?
i suggest you bug your fav client dev about using the outbox model, because changing relays is only part of the solution
what's this outbox model?
its like autopilot for relay settings
Isn't that still "changing the relays"?
yes but if your client doesnt know where to look for your follows itll be just as quiet when they change or vice versa
Exhibit A: Distributed systems made by 5 year olds.
I just made a post about this a couple days ago: nostr:note1ka3pp0gc83cflnnhd58j3pehnk5z8g6f4evd4hjjtgcpvwkmwj2qjqp7w2
Using noStrudel, it will show you the status of every event generated from your account. I've started noticing that with any event whether it be a reaction, zap, note, etc. just as you mentioned, there's always at least a few relays that don't accept it for one vague reason or another.
The thing that really bothers me about it is that with most of them you can't go and see what rule you broke or what the specific cause was if they don't include it in the 5-6 word error.
nostrudel did a great job showing how simple it can be to show the status, now it needs outbox model for reading notes from follows and similar award winning UI for that also.
relays can easily have as many words or landing pages as possible, thats just a matter of giving them a reason to.. (eg, that their users can see the messages)
Agreed, it's UI/UX is still definitely rough and awkward in a few places. I like to actually swap between it and Gossip. Gossip feels even more weird, kind of like an application from the 90s or something, but I appreciate the amount of lower-level info it provides.
Thanks for the info... I guess in these specific cases it's probably like you said, they haven't really had a reason.
Damus appears to be the only one rejecting my events right now. That might be because noStrudel doing something fancy like pushing the latest version of replaceable events if it detects damus serving older versions of my follows profiles for example.
ya, im sure you could also get perpetually silenced if someone reposts your old notes too..
What? If I post your old events to relay, relay will reject your new events? I don't think so. Rate limiting should apply to the connection and not to the author.
its both
in fact its mostly by pubkeys because most relays cannot tell what youre doing on your connection, its all websocket messages.
It’s based on connection/ip not author
the death of nostr lol
No such thing as a free relay. 🤙🏼
I write to multiple private relays and the big relays pull from there.
This is why an extremely configurable relay implementation that's really easy to use, was important to me.
It is time for the ecash nut 🥜 per note relay implementation
I wonder if I'd even notice tbh, I'm here to talk to myself lol
Clients should be retrying notes anyways
There are broadcast buttons.
If your event makes it to one of the medium or large-sized relays, it gets pulled by aggregators and quickly shows up all over. Then you get around rate limits.
The problems arise for people only using the very biggest relays and/or with no outbox model.
This last note was on at least 98 relays within 15 seconds, according to https://nadar.tigerville.no/
Damus broadcasts automatically with geometric backoff if it fails to write to the relay, you don’t need to do it manually. If clients don’t do this then notes are easily lost during connection issues
Yes, I have that problem on Freerse.
I think Nostrudel retries. Never seem to lose anything, there. Same with Coracle.
yeah nostrudel and coracle have both had this since i first used the beginning of this year... the client should keep a queue and a checklist of which relays it has got to and some sort of scheme for backing off
one of the differences with web clients though is that queue is nixed when you refresh the page, and sometimes the page scripts are broken and make this refresh happen on clicks unintentionally
Unthrottling the Damus relay would be a nice premium feature. That's one of the best relays.
Good idea
Unthrottle + long term backup by mirroring to a purple relay for members 🤔
i already subscribe to wine and land i'd be up for damus as well
Same.
Probably would go for Damus Premium for the GitCitadel npub, to maximize reach.
Would also be good if the Premium offered a good spam filter or some advanced search, or something.
On the relay, I mean.
Some of us are relay customers, more than client customers. We're sort of client nomads, but we use the same relays everywhere.
And we'd be more-inclined to use a client, if it offered us controls for a relay, like relay.tools does.
this is what all proper network clients do... you can see examples of it in the bitcoin core and btcd p2p code as well
front end devs are not often so well versed in the problems of socket networking, being used to HTTP APIs... websocket libraries don't give you any tools for this either
🥲📠
WHAT?
AND I THOUGHT IT WAS JUST BECAUSE I TOOK A BREAK FROM SENDING.
I WANT TO SEE THE MANAGER!
nostr:note1tt2p2f7k4f4nu7c3vmg8qfwkrtd6e42dkwwntjn7e9t4cf365mwsg7grme
I noticed that there are errors on all relays except my own public relay ..
