Launch of a community safe network

#362

Question, I’ve noticed that the all nodes contain IPv4.
Does the network support IPv6 only nodes or is IPv4 mandatory?

Also, will I have more space to upload than on the alpha 2 network?
Around a hundred times will do. I think that’s 100k credits from what I recall?
I like getting my site online and then tinkering with it.

Although ten times the amount of what I got on the alpha 2 network should be enough to redo it in one swoop.

#363

The network supports both IPv4 and IPv6. In fact the nodes from Hetzner have simultaneously both kinds of addresses.

The crust config file I have provided contains IPv4 addresses but it would have worked if I had provided IPv6 addresses (syntax is slightly different with IP address inside square brackets, e.g. "[2a01:4f9:c010:2781::1]:5483").

Sorry, you won’t. The limit is strictly the same (1000 PUT/POST operations).

#364

How would IPv6 only nodes work for people that have IPv4 only?

I use my IPv4 for my personal cloud and my phone has IPv4 only.
That means that my Peertube and regular website raspberry Pi server use IPv6 only.
That means few people can see it, and it’s a really niche subject, so I can’t advertise it.

The most useless of the two is the Peertube site, so I’m willing make the switch to the safe network for that one.
But like I said it’s IPv6 only.

1 Like
#365

Good question, but it is more than that. right now if you bootstrap an ip6 net then all nodes are ip6 etc. same with ipv4. Or you can have the network almost split. I propose what we will do is have vaults advertise their ip6 and ip4 connection. Therefore nodes connecting will chose what they want, if we favour ip6 then it is good news for the industry.

tl;dr We support both, but individually right now, it is not a huge task to support both concurrently, but likely not before launch. Like hole punch this is one for the after launch TODO list. ofc like hole punch if there is a significant (realistic and measurable) impact on adoption then we move it up the list.

9 Likes
#366

What will you implement at launch: IPv4 or IPv6?

IMO IPv6 is better because it is the future and it will be needed if we target hundreds of millions of users.

I have tested that clients (Safe Brower and Safe Auth CLI) connect successfully to the network with IPv6 only addresses in their crust config file.

In both cases the client was dual stack (IPv4 + IPv6):

  • Safe Browser on my home PC with my ISP that supports both
  • Safe Auth CLI on a Hetzner VPS

I suppose that many vaults are also dual stack (at least those from Hetzner are).

Maybe that helped to make it work.

I intend to go further and test if a vault can connect with IPv6 addresses in it crust config file. If it connects with the right number of vaults in its routing table, would that indicate that the network doesn’t split?

Edit: Connecting a vault with IPv6 addresses didn’t work, so my second question is irrelevant. Error was: “Bootstrapper has no active children left - bootstrap has failed”

1 Like
#368

Was this question directed at me? IPv6.

Anyway, I haven’t been able to add the node yet.
The binaries seemed to be for windows?

I’m using Arch Linux Arm and so after binaries didn’t seem to work, I’m trying to install from source.
But anyone who knows Arch Linux, this means installing packages that normally any other distro would have, so in case anyone else gets this problem, here’s what’s needed to be done:

sudo pacman -S rustup libsodium pkgconf gcc make automake
rustup install stable
rustup default stable
cargo update
cargo build --release
1 Like
#369

Sorry @folaht, I was unclear. This question was directed at @maidsafe. So, a gentle reminder for them.

I come back to what I said earlier: when I connect a IPv6 client, I can see in the connected node logs that IP address of client is the IP address of the gateway of my docker docker_gwbridge network. This means that docker somehow does the translation from the host public IPv6 address to the private internal IPv4 address, which means that the vault still manages IPv4 traffic.

A confirmation of this is that the connection is unsuccessful when the IPv6 client tries to connect to a non docker contact node (error is: [crust::main::bootstrap] Bootstrapper has no active children left - bootstrap has failed).

If anyone is interested I can publish a crust config file for IPv6 clients (with only 3 contact nodes, because only 3 of them have joined my docker swarm)

1 Like
#370

Sorry, at launch i think we might have both. @dirvine is also of that idea AFAIK - put both stuff in the cache file and give it around and can connect via either (or whichever is supported, gets connection first etc).

5 Likes
#371

I seem to have the same problems as bzee.

[admin@Yrofoni app]$ RUST_LOG=info ./safe_vault
INFO 15:58:51.646537778 [<unknown> <unknown>:96] 

Running safe_vault v0.18.0
==========================
INFO 15:58:52.687343041 [<unknown> <unknown>:116] Created chunk store at /tmp/safe_vault_chunk_store.l9TyS05qshgf with capacity of 34359738368 bytes.
ERROR 15:58:56.777440326 [crust::main::bootstrap mod.rs:192] Failed to Bootstrap: (FailedExternalReachability) Bootstrappee node could not establish connection to us.
INFO 15:58:56.779628149 [<unknown> <unknown>:269] Bootstrapping(b611ef..) Failed to bootstrap. Terminating.

This should be correct no?

#372

I interpret your screen shot as an indication that your equipment is IPv6. The problem is that the community network is IPv4 and so a vault cannot connect to it with IPv6 (though a client can as explained above).

#373

Yes.

That was what my question was all about. I just read these parts:

The network supports both IPv4 and IPv6.

tl;dr We support both,

I now see you two meant, the safe network itself supports it,
but can only be configured to one or the other.
And of course that means IPv4 for the community edition. :slightly_frowning_face:


Another question:

I want to be able to keep updating the website I’m working on on the same URL.
If I publish a website and I’m out of account, does that mean the URL can no longer be edited and a new URL has to be assigned?

Because that’s the case with alpha 2.

#374

Yes, you cannot update your site anymore when its put balance is empty. In the final network you will be able to refill the put balance with safecoins, but until then this is not possible

You cannot even do that because only one account can be associated to a forum user. We might lift this restriction if there are more nodes allowing for more disk space (but the network doesn’t take that direction for now).

#375

If I’m understanding the dashboard correctly, it looks like the hosts have plenty of available disk space. The host with the highest used space is neo–00, with a utilization of 0.5% of it’s space (166 MB out of 32 GB).

#376

Currently, the network is stable around 13/14 nodes but in the past, it had fluctuations with the number of nodes falling several times to 8:

egrep "Exact network size: 8" /var/log/safe_vault.log
I 19-02-28 23:54:37.842574 | Exact network size: 8                      |
I 19-03-18 12:08:28.530986 | Exact network size: 8                      |
I 19-03-18 12:21:46.576438 | Exact network size: 8                      |
I 19-03-18 12:27:35.509238 | Exact network size: 8                      |
I 19-04-10 21:39:13.595242 | Exact network size: 8                      |
I 19-04-11 21:20:16.879570 | Exact network size: 8                      |

Each account can create 8 * 1000 MB = 8 GB of data, which is 1/32 of the total space provided by 8 vaults.

And “Max space” column just displays the max_capacity parameter in safe_vault.vault.config file, which is the disk capacity that each vault should dedicate to the network. The OP has announced that this capacity is 32 GiB, but it is not enforced by network code, and we know that some vaults don’t have it.

So, current reserve is not enough to allow several accounts per forum user.

Besides, if it was, a question would be: do we allow it or do we open data upload to level 1 forum user?

2 Likes
#377

yes of course, it’s just a test network. Let people play with it and get a feeling with it. I don’t expect a huge increase in data usage, not everybody is gonna use it.
My vault is running for months now and it only has 200mb stored, it gets boring :wink:

#378

Hmm, interesting. Why are vaults misrepresenting their capacity? Is 32 GB a barrier to participation as a vault? I have a VPS I hardly use (Ubuntu 16.04 LTS), but its drive is 25 GB.

I’m not quite understanding the math…

  1. I thought the maximum number of PUT operations is 1000, and that 1MB is the limit per PUT (1000 MB max per account). Maybe that’s just for the official testnet though.
  2. I’m not sure what the “8 *” represents (“8 * 1000 MB”).
  3. Assuming 8 GB of data max per account, wouldn’t that be more than 1/32 of the total space provided by 8 vaults? (Or is there no redundancy?)

On the official testnet, I’ve already used 174 out of 1000 PUT operations just screwing around with tiny amounts of data. It surprised me how fast those were spent. As far as I can tell, the PUT quota never renews (it’s been over a month since I racked this up), and there’s no way to buy more (or earn more by running a vault). I haven’t yet tried mock infrastructure (running SAFE Browser or Web Hosting Manager in test mode). As it is, I have enough difficulty getting the software to work whenever there’s a new release :confused: (usually I run into one problem or another even when following the instructions carefully). But I probably will someday try a local mock network be able to try more of the API out and maybe try to develop a simple app.

I’ll recuse myself from taking a public position on this specifically due to potential conflict of interest. :sweat_smile:

I’m curious as to the worst-case scenario of the network getting full. Is it a lot of work to reinitialize the network (i.e. delete everything and start again fresh)? I get that a goal of the SAFE network is to be able to save everything forever, but I’m not sure how many people actually expect it (there is no stable version 1.0 of the software released yet).

#379

If it happens, than it is interesting to see how the network will behave and I see it then as a big success :slight_smile: . But to be realistic, I don’t even think my vault would even store 1 GB. The most dedicated people have access now, and only 200MB has been stored on my vault. With level 1 access, only a few will try some things out and that’s it I’m afraid.
I really hope the opposite to be honest so the network is getting alive and people are getting enthusiastic about it.
So for me it’s fine if more people get access to it.

1 Like
#380

Why not open it up, unrestricted (well at first to L2 or L1 but without put limits) and see what happens, take it down and start over? Stressing and crashing the network seems much more interesting than running a copy of alpha 2 the same way.
Sorry if the answer is obvious. I hope to join when I find the time, and find out myself.

#381

I just published a small website on the alpha2 testnet since then and it cost me 27 PUTs (5 files totaling 779 KB).

#382

Because some users want to participate but have a lower disk space or didn’t even pay attention to this requirement.

Each data chunk is duplicated 8 times, 1000 PUT operations, each PUT may add up to 1 MB, so a possible use of 8 * 1000 * 1 MB = 8000 MB = 8 GB per account.

Total space provided by 8 nodes is 8 * 32 GB. An account may add 8 GB which represent 8 / 8 * 32 = 1 / 32 of the total space.

L2 or L1 is possible with Alpha 2 which is much more robust than community network. No put limits is too easy to crash.

The big difference is users connecting their own vaults. Stressing the network wasn’t my objective.

4 Likes