Here it is here
This should be read by everyone
6 Likes
No disrespect @Shu but I would argue that while you may not be running your nodes from a datacentre, you have a fair whack of datacentre-class kit at homeā¦
Colour me jealous 
7 Likes
Sounds good.
So will you be assigning a value at which it will be decided if a node is to be shunned or not?
I hope though it will not lead to preferred nodes or not preferred nodes. This would result in a class system for nodes with elite nodes compared to non-elite and eventually weird segmentation of the network. To prevent this weird network connectivity, it needs to be shunned or not shunned and no preference of routing which results in nodes being bypassed based on group think.
For instance a country with slightly less performant internet, not enough to be a major issue in the network but enough to always have nodes in that region with a lesser score. That countryās nodes will end up being bypassed for data access, bypassed for churning, and eventually not be considered a close node for chunks because the preferred routes never see them.
I hope this is not ignored as I suspect once this is implemented then then this path will be gone down and VERY HARD to ever see happening in the network because it will take time to occur and many times longer to ever notice. Though I suspect it will be since its a juicy idea to implement preferred routes for performance sake. Remember the troubles with sections, well this will cause many troubles but difficult to isolate.
Please do not ignore the above warning as some rant, keep it in mind or we will be debugging it in a years time
[EDIT] The one exception to āpreferredā node is when the client is choosing which node to download from. In this case it is like the client-sever situation that preferred routing assists. Choosing which node to upload to is based more on cost. This also is a mesh style of network and not client-server style, all nodes are essential to keep on even level for data security.
And handled with kids gloves with pleas by Bux herself to not stress or do anything that could cause issues. Now this is fair enough to keep the network working well, but not fair to say its doing great.
Extremely important is the need to relieve ISP routers of buffer issues which would allow them to handle the 4MB chunks that will be flying across the network when actually being used. That window size, known about since before TCP was invented, need to be set to a real size (128KB would be good). Thats one reason the network has to be treated with kids gloves and great care taken on uploaders and their effect.
Mind you on the other side of the coin it is good to see the network is holding together better than the initial introduction to 4MB max chunk size. This will be due to many tweaks, and kids gloves being used. The window size in QUIC will have a greater effect stability wise than many of the other tweaks combined. Maybe even enough to remove the kids gloves since massive churn will not cascade the error rate due to all ISP routers crapping out.
Actually a lot of old time communications principles need to be revisited as some historical lessons learnt by old timers are completely unknown it seems. The network is just a part of it. Simple to fix though which will be good. The network side though is working reasonably well in my opinion, just needs a couple of communication parameters set right (*cough* *cough* QUIC window size)
Not when 3 weeks in and my nodes are not even 6% full (<< 1000 responsible records per node)
You mean not testing at home? One chunk every so often through my router and usually not max size will not test ISP routers.
tl;dr
Its great to see the improvements made and the network working so much better than when the 4MB/32GB network was introduced. But without the kids gloves being taken off it cannot be claimed to be āWorking quite nicelyā without qualification. For instance ISP routerās buffers are not being overloaded with the very low upload rate and virtually no downloads and kids gloves applied.
I guess the QUIC window size issue and some others have been sidelined due to home nodes being reduced greatly due to CPU usage being enforced. Some at 1/10th and others like me about the same. But data centre nodes (including homes like DC conditions eg shu, me, etc) have not been reduced to the same extent and probably make up a much greater percentage of nodes now compared to when 4MB/32GB nodes were introduced.
16 Likes
I donāt have the technical knowledge to dive into or counter some of the (sometimes painful) criticism being shared. But this doesnāt feel like the right way or place to address it like this.
If the goal is better communication, attacking the team isnāt going to help. It doesnāt give confidence for people reading along who are just getting to know the network. And isnāt that exactly what we need? More people trusting this network and wanting to get involved.
I get that this forum is here to share ideas, concerns, and even frustrations, but itās also about how we do that. Sure, you can be outspoken, but there are limits (calling something a betrayal, for example?). Please be more respectful. By the way, I recall you saying something similar to Weebert onceā¦
6 Likes
Wohooow - I didnāt recognise the āsummariseā button yet but it does a decent job summarising a topic it looks to me!
3 Likes
I think the lack of information and hence a lack of trust in the network itself makes many of us wonder if itās right to pull people in.
Iām certainly not willing to invite anyone at the moment, not until I can witness data flying up and down with significantly higher volume.
7 Likes
Believe me it was painful to write that - and as I said - I hope to be proved wrong and have to apologise in due course.
But it would be more painful to say nothing and trust to luck - cos that has not worked. We have no API, there are still big questions re registers and at the risk of sounding like a stuck record no meaningful stress testing has been carried out.
As for respectful, I have until now been very circumspect in my criticism - in which I can assure you I am not alone in. Its not just me who is thinking these things. Now Im saying it out loud while others may be more reticent but would still broadly agree with me. Perhaps it may carry a little more weight coming from someone who has previously been supportive through thick and thin.
And I finally got fed up with legitimate concerns from well-respected community members being brushed off. Unlike them I cannot claim any truly deep technical understanding but I have been around this project for a very long time (too long, some may say) and I want to make my concerns known. Any test/QA work I have done was mainly in mechanical engineering but the principles of testing and QA cross the boundaries.
Which is why I totally welcome the suggestions from @rreive as he is a software guy with the appropriate test/QA experience. I am not the guy to write the test schedules but like many others here I can likely tell the difference between a test schedule that is adequate and one that is not.
What worries me most is that there has been no real mention beforehand of a proper end to end test schedule for the network. Yes of course there are the test routines Ci, Jenkins etc etc - they only test the code is correct, not whether this works under real life conditions. And real life conditions will have a much smaller % of datacentre nodes and their associated fat pipes and routers with massive session table capacity, multiply-redundant PSUs etc etc.
13 Likes
You donāt suggest how this should be addressed in your opinion.
My take is that it hasnāt been addressed by Autonomi, after months of forebearance and self censorship by those who saw the increasing dissonance between words and reality, until there was no alternative but to start raising the issue because it became blatantly deceitful.
Thatās not something that āshouldā be ignored IMO, no matter how bad it looks to anyone arriving or who prefers to bathe in bullshit.
And what has been the response from Autonomi? Meagre, warm words when there was anything at all, and no change. Questions about this are mostly ignored. The reality is ignored: launch was achieved on time, code freeze happens in a few weeks, the WP, API, advice for investors are allā¦ er um, and pigs are now flying over us daily. They donāt care. Obviously they donāt see what is happening here as an issue or they would be rectifying at least the communication.
We donāt know why, what their priorities are, and why the community doesnāt figure among them becauseā¦ Autonomi donāt share any of this. Outside the detailed tasks of the dev team, we arenāt told what they are up to.
13 Likes
Man, I wish Elon got this much shit for consistently underdelivering. 
12 Likes
Would it help if we were calling this early beta of the data storage part of the network, and Jan is launching the later beta with TGE (testing?) and later on will be the launching of the apis for application building?
4 Likes
Elon has/had the resources to keep on working after multiple vehicles exploded in flight/ on impact/nearly worked but fell over etc etc. And Elon can do more or less what he wants. Autonomi has all sorts of constraints on its actions.
Autonmi does not have Elons resources. I think we have ONE chance to make this work, so it might be a good idea to be as totally prepared and confident as possible.
Can you say you are even 90% confident that it will be alright on the night?
8 Likes
Maybe there could be another way ā like creating a private group where OGs like you and southside can share their alarming concerns directly (and constructively) with someone from the team @JimCollinson.
That way, we keep the trust of the wider community while ensuring important feedback is heard.
1 Like
āWeā can know exactly what to call this. Iāve been saying for months that we werenāt in beta for example.
What is missing IMO is honesty, realism, clarity and respect from Autonomi.
We used to have this which is why it is so blatantly absent this year.
12 Likes
I get your point, and others as well. Just attempting to lighten the mood honestly. I think if we all try to keep our chins up and help where we can, then yes we will.
Elon. Thatās a whole topic on its own, haha.
7 Likes
What we are looking for is the chance to help where we can and taking part in structured testing directed by Autonomi is exactly what we should be doing.
Now that we know for certain we can spend our attos on uploads, we should get organised and arrange a planned series of uploads by one team and have another team download and verify them. Tens of thousands of times. And thats just for starters.
As for Elon, he got his own thread in Off-topic. I tend to stay out of itā¦
āThe least despicable of the identified oligarchsā is my current take on him. I reserve the right to change my mind.
4 Likes
That was my position for many years but he has gone completely troppo recently . .
3 Likes
Taking discussions private is not a good approach and I politely refused to do that when someone from Autonomi suggested it. That amounts to censorship, and IMO hiding the reality.
I had kept my concerns quiet for months, began raising them occasionally without them being addressed, and eventually decided to call out what was to me blatant deceit.
Iām still surprised that people who I think must understand this, are criticising the ānegativityā people who wonāt swallow it and donāt want others taken in by it.
I donāt want people arriving here to be told things that arenāt true without pointing this out and am glad Iām not entirely alone in this.
14 Likes
imo the Emerging Autonomi Community needs its own Test work to do development on, you cant do this on the production network and that is where the release candidates should be parked by Maidsafe, so the Community can go at it with a broad array of Blackbox testing from every angle, before any Release Candidate gets released into the wild, with permissioned instrumentation and logging, using the existing tool sets upgraded to handle such work. Maidsafe needs to think this through in my opinion, and redirect rewards to get this funded and built fast, now that the tech has been validated to work under light load. The whole security set of tests should bne Community centric aimed at a clone of the currently operating production network first.
That means imo Maidsafe would very wise in funding and deploying a lot of system node fleets in the Cloud an enlisting the Community to run the black box tests on those core Test Network fleets armed with test dummy tokens, which can also be used to develop Apps.
The āEnlisted BlackBox Testersā and āAutonomi App Buildersā would then work with the Maidsafe assigned Maidsafe Community Test Network engineers to drum up on the Cloud Fleets Colo and the MetalaaS fleet imo.
as well as those Enlisted Blackbox Testers (for each test round) that want to run single system fleet and node tests on equipment they have can choose to optionally join this community test network and create their own black box testing and reporting using Maidsafe Core team instrumented versions of the current production code,
which allow the Enlisted Blackbox Testers EBTs to āpermissionā turn on the refashioned instrumented logging sent to a Community and Maidsafe read accesible central logging volumes in the cloud, of flat files for each system node fleet and in select community determined test and builder node only sets, logging results (say using OpenTelemetry) that can then be rolled up into a proper postgres db system selectively to generate Test reports (Jasper or the like, why not) versus test pass/fail metrics of the Test Acceptance Plans per round of testing, Reporting graphed and profiled by type of safenode host gear and size of fleet yadda, yaddaā¦
So we start to get a clear picture of what is going in the primary, secondary and exception use case sense from multiple test perspectives run by the EBTs
Itās possible to round up all these logs with a pub/sub Event Driven Architecture which is more efficient and scales way better than rest api,
so that would be my vote, transform the existing tooling for testing purposes to pub/sub event drive design and run a number of hosted FOSS Solace pub/sub brokers in the cloud to act as traffic cops, diltering reporting and graphic Traffic in close groups as well as between close and far groups and pumpbing that at say Kafka log repositories selectively for each test wave , so that traffic can be properly analysed with and between close groups.
Heck I can likely get my friend at Solace to help out, in his spare timeā¦,
My current Solace contact was my Principal Lead Architect at Platespin back in the day , previously he had worked for TransArc before IBM bought them (and on early Java when it was called āoakā)
n.b- Solace are working on totally distributed Agentic AI to build a cognitive mesh so they would probably be an excellent partner for Autonomi, I know for a fact they are looking to deploy their early prototype of Agentic AI Cognitive Mesh at scale
Start the video at 18:26from CTO Shawn McAlister of Solace as he intros the lead AI guy Ed Funnekotter who gives a demo of how Agentic AI works in the event driven Cognitive Mesh
Keep in mind Solace provide the infrastructure for event driven architectures at some of the biggest companies in the world handling Millions of events per day
I can make intros to Shawn and Ed via my contact there for Maidsafe and key member of the Builder Community to get the ball rolling , if you like ,
if Maidsafe and the Community is interested,
but please first watch the video first to understand Agentic AI and Cognitive Mesh, and do a bit of homework on the pubsub broker space, they dovetail into a lot of other projects like Apache Kafkaā¦
I think Solace would be a very good partner fit with where Maidsafe is going with personal private AI , and how to monetize that app , outside looking inā¦ 
The good news its early days but happening really FAST., and Solace is the clear agile leader in the space , so easy to work with
@dirvine Solace are the ex NewBridge Guys backed by Terry Mathews Canadaās biggest Billionaire and founder of NewBridge and cofounder of Mitel ,
for those not familiar (or much younger ) Newbridge sold out to Alcatel for $10B+ USā¦ so at the time smartest of the smart in the Internet WAN spaceā¦
and then theex Newbridge members founded Solace two years later, with many of the core team still in their 30s at the timeā¦ they are all now all in their 50s , a few of them maybe older
Solace also know and understand OpenTelemetry very well and how to scale itā¦so that expertise would in theory would be really handy for the proposed Community BlackBox Test effortā¦
All doable Glad to help hook up Maidsafe to Solace, if there is interest
8 Likes
I want to know what everyone in the community has to say. Negative feedback is normal in any project. I can judge for myself what the truth is based on all available information. If we have a lot of secret/private info, then how could I ever be an investor, knowing Iām at an informational disadvantage? If I need to be protected from the truth there are lots of other projects where I wonāt be allowed to know what is really happening. I want to be at this project where people get to the truth quickly. Getting everyoneās input is important to having the project succeed.
6 Likes
I do understand your point, and I can see why taking discussions private might feel like censorship or hiding the reality. Perhaps it wasnāt the best suggestion on my part.
That said, for me, itās more about how itās discussed. The way things are said can either move us closer to solutions or further away from them. If the goal is to ensure that concerns are addressed and the truth is upheld, I donāt think attacking or escalating will achieve that.
I truly hope we can get to a place where thereās more clarity, the atmosphere improves, and weāre all pulling in the same direction again. Now I rest my case.
8 Likes