Fleming Testnet v1 Release - *NOW OFFLINE IN PREPARATION FOR TESTNET V2*

[Vort]

@peca

And when lazydog.jpg already exists, it should overwrite it, right?

Yes. And exactly this happens on Windows.
Maybe just first try was successful and second one failed.

upd. Redirection is done not by Safe utils. It is performed by operating system.

[Sascha]

You can see the time below. I tried the same dog command a couple of minutes later, and now it’s just sitting there.

sascha@sascha-Librem-15-v3:~$ time safe dog safe://hygoygykdytfkxqmdazdgsrubwi5cozmuysaajupww9yoa7kh6zg4uejrbe

== URL resolution step 1 ==
Resolved from: safe://hygoygykdytfkxqmdazdgsrubwi5cozmuysaajupww9yoa7kh6zg4uejrbe
= File =
XOR-URL: safe://hygoygykdytfkxqmdazdgsrubwi5cozmuysaajupww9yoa7kh6zg4uejrbe
XOR name: 0x43044aa7b963c5c66b1261a576c85d7305b184cdb4a7c10c755cf5cda9a1240a
Native data type: PublicBlob
Media type: image/jpeg

real 0m8,160s
user 0m0,791s
sys 0m0,175s

sascha@sascha-Librem-15-v3:~$ time safe dog safe://hygoygykdytfkxqmdazdgsrubwi5cozmuysaajupww9yoa7kh6zg4uejrbe

[dirvine]

@Antifragile

Security of blockhcian is achieved with very few miners, basically even with 1 single miner.

Probably better to say cpu load or hash power. It does not relate to number of nodes (we are both saying similar)

I think like this, blockchain is >50% to attack (in some cases in partition it’s 33%).
The 50% is total hashpower / difficulty etc.
Safe is >66% to attack (>33% to stall) of total Elder node count (which by extension is total node count).

[dirvine]

@happybeing

But both are just sitting there for me. Is anyone else seeing this?

That could be a hang, if you ctrl-c and retry it may work. There is another issue though and more to come very soon

[happybeing]

@dirvine

That could be a hang, if you ctrl-c and retry it may work. There is another issue though and more to come very soon

I’ve been doing this, will try more vigorously!

[StephenC]

Update from MaidSafe HQ

First of all thank you to everyone who has helped contribute to the success of Fleming Testnet v1. Your testing and feedback is absolutely invaluable in highlighting areas for improvement, all of which we are reading through and taking into consideration, and many of which we are already in the process of resolving.

To prove that we are able to achieve this testnet’s goal of having a secure and stable network running with all the milestone Fleming features in place is a huge shot in the arm for all of the staff at MaidSafe, and I hope for our community as well.

We have community nodes as Adults and Elders, splits occurring and rewards being paid.

What’s Next?

At the time of writing, testnet v1 is still up & running, approaching 4 days since it was opened to the public.

With your help, we have identified several improvements and so our intention is to quickly release a testnet v2 with some of the early fixes we are working on.

We anticipate having a testnet v2 ready to deploy within the next day or so.

At the moment, the scope of the fixes for v2 are:

1. To make the node joining process a lot smoother, we are implementing a queue of infants so that you would no longer need to keep trying to add your node to the network. So in v2 when you try to add your node you join the infant queue, and when a space becomes available in a section, whichever infant in the queue is closest to that section’s XorName space will be selected to be promoted to join that section as an adult. Using the XorName here helps maintain balance in the testnet.

2. We are also removing some limits we had on new nodes joining the testnet - in v1 we would wait until a certain percentage of space was filled before allowing a new node to join, but now we will also allow a new node to join as soon as any current node becomes full. This should allow more nodes onto the testnet

3. In testnet v1 we have observed a particular section (prefix 10) become stuck in a DKG loop - it seems this was due to 2 community nodes in that section being promoted to be Elders, but it turned out that other nodes could not establish a direct connection to them to perform Elder duties. This led to a (successful) DKG vote to demote those Elders to Adults, then a further DKG vote to promote the 2 oldest Adults to be Elders…which, you guessed it, were the same 2 nodes that couldn’t be connected to previously. This loop continues endlessly and the section does not perform any other actions in the meantime, causing some of the issues reported in this thread and others. The simple fix for this is to only allow a node to be promoted to Elder if we can establish a connection to it, otherwise it will remain as an Adult regardless of its node age.

   Note that remaining as an Adult is not a disadvantage to this node, it can continue earning rewards in that role. This way we are not excluding anybody from the network just because they are behind problematic NATs, etc.

Can I keep Using Testnet v1?

@Toivo

Is there still point to try to join as a node? Do you Maidsafe folks expect to gather some valuable information still? Or is there something specific you would like us to try?

Please feel free to continue testing with testnet v1, if you wish to do so. We have lots of data and logs so don’t anticipate needing any more on the issues that have already been flagged, but we’ll be keeping an eye out for anything else coming in, and requesting logs wherever we need them.

You can also of course hold off for v2 to be released, at which point we will be asking for you to try it out for us and see whether the fixes we make have had the desired impact. As stated above, we anticipate that a testnet v2 will be released within the next day or so.

What Will Happen to Testnet v1 and its Data?

When testnet v2 is ready to be released we will announce on the forum that we are taking testnet v1 offline. All of the bootstrap nodes that we have on Digital Ocean will then be destroyed and the testnet will go offline.

We will then deploy new, clean D.O. nodes for testnet v2, and once they are confirmed up and running, we will share the node_connection_info.config with you. We will be asking that everyone who attempts to use testnet v2 then removes their .safe folder before trying to interact with testnet v2, i.e. $ rm -rf $HOME/.safe/. This will help prevent any confusion and false issues due to pollution from testnet v1.

Therefore, all testnet v1 data will be deleted.

Are We Working on Other Fixes?

Yes. Multiple other fixes and improvements are already in progress, some even from before we released testnet v1. One of the main priorities we have at the moment is completing the implementation of Anti Entropy across the board - we believe that this will resolve many of the errors that people see in their logs, and make the testnets even more reliable and scalable. This, along with others, are in progress for future testnet iterations.

[Sascha]

You were right. I stopped it, and upon restart the dog got here quickly.

sascha@sascha-Librem-15-v3:~$ time safe dog safe://hygoygykdytfkxqmdazdgsrubwi5cozmuysaajupww9yoa7kh6zg4uejrbe

== URL resolution step 1 ==
Resolved from: safe://hygoygykdytfkxqmdazdgsrubwi5cozmuysaajupww9yoa7kh6zg4uejrbe
= File =
XOR-URL: safe://hygoygykdytfkxqmdazdgsrubwi5cozmuysaajupww9yoa7kh6zg4uejrbe
XOR name: 0x43044aa7b963c5c66b1261a576c85d7305b184cdb4a7c10c755cf5cda9a1240a
Native data type: PublicBlob
Media type: image/jpeg

real 0m12,016s
user 0m0,849s
sys 0m0,230s

[dirvine]

Superb, the amazing thing to me is

1. Any node can now earn SNT, regardless of their connection (Acess For Everyone!!!)
2. The speed of Iterations here shows what a focussed team in agreement can do! Humbling to watch these guys do what they do. They all deserve what’s coming their way soon.

[happybeing]

@Sascha

You were right. I stopped it, and upon restart the dog got here quickly.

Not for me

[Vort]

@StephenC

Therefore, all testnet v1 data will be deleted.

Will it be possible to reupload the same test files at the same addresses as before?

[peca]

@Vort

Maybe just first try was successful and second one failed.

That was my first thought too. I tried it 3 times and the result was always the same.

Now I tried it with different file which throws error, same thing. When the file desnt exist it throws error in less than 5s and finishes, if file exists no errors and never finishes. Again tried 3 times with same results.

[Michael_Hills]

@StephenC

Yes. Multiple other fixes and improvements are already in progress, some even from before we released testnet v1.

Progress, light-speed progress, these are days I’ve been waiting for! Great job everyone. I feel Fleming and beyond is soon upon us!

[Vort]

@StephenC

This loop continues endlessly and the section does not perform any other actions in the meantime

Not much important at this point, but maybe it makes sense to ask operators of that nodes to shut them down (restart)?

[BambooGarden]

@StephenC

v2 ready to deploy within the next day or so

jesus. can you not at least wait until thursday?
i can hardly keep up with this

[Zoki]

@BambooGarden

jesus. can you not at least wait until thursday?
i can hardly keep up with this

Yea… you wait years for a testnet then 2 or 3 arrive at the same time!!!

Typical.

[davidpbrown]

@bochaco

just because you are missing the --self-auth flag.

That’s made a big difference!.. now it just works.

@happybeing

the result is the same. The following was reported as working quickly 35 mins ago:

Had no problem with either dog or the .jpg downloading.

I’m midway through uploading a 1GB and can see the CPU and upload network and then the balance with safe keys balance is going down. Unclear how long it’ll take or how much 1GB costs in total; so, progress hard to judge but looking good so far.

[Vort]

@davidpbrown

Unclear how long it’ll take

I was trying to guess progress by looking at consumed traffic with Process Explorer.
Used traffic should be 6 or 7 times higher (do not remember exactly) than file size (or even more).

upd. Did you started at 19:29 UTC+3? I see lots of chunks.

[Guybrows]

The fact that you are pumping out new test nets so quickly is great to hear. Well done guys . Exciting times and making the wait to get here long forgotten.

[JayBird]

The team is on fire, it’s lovely to watch. Exciting times!

Thought I’d have more time to get myself sorted with wifi but things are moving so fast I thought I’d ask - is there anything useful that can be done with only intermittent internet access? Not online very often at that. I could upload things to testnets when online, but have no chance of becoming a node and chunking myself up to the eyeballs, is that the situation?

I think I saw people renting an always on server, or something like that. You then control it and play with it when you’re online through some sort of ssh thingy or I don’t know what, and it is then on all the time? Then chunks galore? Or what thanks!

[bochaco]

@wydileie

A feature request for the team after using the test net. Can you implement bash completion for the CLI? It would make things much easier to work with without having to use the help command all the time.

We already have an implementation of this: https://github.com/maidsafe/sn_api/blob/master/sn_cli/README.md#shell-completions, I’ve never used it so it would be good to know if it needs some fix/improvement if that’s something you’d be interested in.