I’m finally getting a chance to try the Launcher and create an account. If failed 3 times trying to create the account. I finally said maybe it’s really going through and just kicking an error for some reason. Sure enough the credentials I used to create the account worked, even though it said failed to create the account each time.
Maybe I’ve misinterpreted things, wasn’t the idea to multi partition large HDD’s. Not sure what benefit partitioning has if were not coupling vaults to those partitions.
Are we now saying that 1 vault per machine is best, regardless of that machines resources.
1 Like
Eventually. But at present they’re mainly testing networking, so I suppose they would like to avoid the complication of network connections becoming bottlenecks, so they can test one thing at a time.
I thought that would be the case and would hope this to be clarified in OP for each test. Clarifiying in OP may have prevented the massive vault take down…which is maybe what we wanted in previous tests.[quote=“Ross, post:1, topic:9829”]
We are testing that accounts are stable and that community members participating are able to log in and out as they like without experiencing issues.
[/quote]
So we know what your testing, we should then be informed in how to best help you achieve it. So in this case, something like “please only run 1 vault per machine, but create many accounts and many login attempts”
1 Like
Chris, I am not sure whether that would be the preferred method. I would expect that humans are going to behave like humans, so let them do their thing and see what happens. Although what you are saying is reasonable, I would rather stuff just happens like in the real world.
There is always going to be someone who chooses not to follow such advise 
5 Likes
WOW! I’m super duper impressed! It’s not even taking 1 percent of CPU! Sometimes, it fluctuates to 10 percent. I’m running i7-2600k.
It’s not even memory intensive! Averaging .5 percent of 16gb ram.
Now this is what I’m talking about!
7 Likes
Right now I’m not following the advice
but then I have 20Mb/s upload so I don’t feel bad about running 5 vaults from work
5 vaults from home where I can get 1.1Mbs upload at best would be daft cos Id be throttling each vault to 200Mb/s - which would give misleading results if I understand @Viv correctly above.
I had missed the point about registering many accounts and testing logins, uploads etc so I’ll make a few now.
FWIW here is my latest log from the single vault on a home ADSL connection
.
INFO 02:21:53.914582325 [routing::stats stats.rs:151] Stats - Sent 49000 messages in total, comprising 55355318 bytes, 22 uncategorised, routes/failed: [6780, 705, 556, 509, 418, 380, 317, 301]/187
INFO 02:21:53.914771747 [routing::stats stats.rs:158] Stats - Direct - NodeIdentify: 69, NewNode: 746, ConnectionUnneeded: 3
INFO 02:21:53.914785876 [routing::stats stats.rs:162] Stats - Hops (Request/Response) - GetNodeName: 77/22, ExpectCloseNode: 264, GetCloseGroup: 3879/6800, ConnectionInfo: 5617, Ack: 20148, GroupMessageHash: 10883
INFO 02:21:53.914797853 [routing::stats stats.rs:172] Stats - User (Request/Success/Failure) - Get: 41/40/1, Put: 19/19/0, Post: 12/12/0, Delete: 0/0/0, GetAccountInfo: 0/0/0, Refresh: 326
3 Likes
For sure and specific advice can empower those who wish to help or hinder I guess.
What prompted me to raise this, was the start/ stop of 1200 nodes…I took this as being an entity outside of Maidsafe, dropping the nodes with ‘no malice or attack’ but maybe that was someone internally firing off a script [quote=“dirvine, post:112, topic:9829”]
There was an error that caused possibly 1200 nodes to start up and stop again. It was done with best intentions and will be a great help moving forward. No malice or attack, but it would have meant it was over 80% possibly and the network then struggles.
[/quote]
Ugh, I can’t register into the network with safe_launcher. It fails.
Yeah, the previous weekly update where Ross said they were running 50 nodes on one machine got people thinking that that was the thing to do on the next testnet.
Now, 50 vaults per machine at ethernet speed (i.e., on a LAN) would be something to experiment with.
1 Like
Going Off Topic… but I feel this is important.
I agree with @Viv
Farming Incentive Problem
Right now, farmers run as many vaults as possible. Why? Because they collect more chunks, increasing opportunity for more GET requests which earn more Safecoin.
The SAFE Network ends up with unnecessary “hops” which adds more traffic. Even though farmers know multiple vaults throttle their bandwith, they are still motivated because the reward for Safecoin is higher with multiple vaults compared to a single vault. Unfortunately, this hinders Networks performance.
Farming Incentive Solution?
I’ve been wrestling with this problem since last year. And the only possible solution I came up with is to “limit” 1 vault per machine. People will find ways around it and run multiple vaults. But for the average farmer, they should only be able to run 1 vault per machine… allocate the storage space and they’re done.
EDIT UPDATE: If the Network prioritizes chunk storage to high ranking vaults (high Bandwith availability) then farmers are incentivized to make one strong vault over many weak ones. This means we don’t need to “limit” the average farmer to 1 vault.
6 Likes
That is seriously impressive organic tester/adoption, but can’t say I’m surprised - given all the positive enthusiasm on this forum!
1 Like
Might be technically naive here, but would it be possible to enforce that the first node after #1 was a resource intensive node ie if you want to run multiple vaults you take on a disproportionate share of the network overhead.
IP address, problem solved. Run many vaults as you want but if it goes through same IP address, then the farming incentive decreases drastically.
1 Like
Right now there are no farmers. People added vaults to the testnet, probably because they wanted to help the testing, as they understood it.
Your proposal disregards all kinds of diverse combinations of bandwidth, disk storage space, and uptime, and the tendency of the economics of any of those factors to change in a few years. It’s too complex to settle by a fixed rule or fixed set of rules, on a network that does not exist yet. A pricing mechanism is the best real-world system yet devised for optimizing the use of resources in a network of millions of actors.
Automatically “grading” vaults in terms of their available, average bandwidth is a good start. That appears to be what I saw on the current testnet, as reported in my vaults’ logs, with my local vault “changing state” to “client” while the vault on my cloud server made no such change. Finer classifications would allow a vault to be priced, as with grades of commodities in the marketplace. If unused capacity is available then it drops its price until enough buyers fill it up. For comparison, Ethereum has the concept of “gas”, which i’ve seen defined as “the internal pricing for running a transaction or contract.”
5 Likes
Having issues launching vaults on linux. I’m not sure if this is because you’re restarting the network/fiddling with versions or what. I’m hesitant to start up the launcher as I’m not sure which version I need given there’s a new laucher that needs to be released. So I’m not sure which launcher I need/can be used.
An example of sharing above/ below network average ?
What does this mean?
WARN 02:04:11.861380100 [routing::core core.rs:1903] Node(fd5f7f..) Prepared con
nection info for PeerId(35bc..) as f0e5de.., but already tried as f0e5de...
WARN 02:04:12.575856700 [routing::core core.rs:1903] Node(fd5f7f..) Prepared con
nection info for PeerId(df89..) as f16bdf.., but already tried as f16bdf...
WARN 02:05:04.322378700 [routing::core core.rs:683] Node(fd5f7f..) Prepared conn
ection info for PeerId(222e..) as e5f587.., but already tried as e5f587...
I’m getting quite a lot of those
2 Likes
One small file upload failed
es.udhr.safenet upload failed not allowing retry, where that is a simple 14kB index.html similar to others that were successful eitherside of that time. I put that same file then to se.udhr.safenet instead without a problem.
One large file upload failed
eye-eye.gif upload “failed” message despite that being successful - I checked that by downloading the file and sha256sum. I’d uploaded the index.html first, as service creation, then just that 4.9MB image. So, that’s an identical experience to last time. If you download eye.eye.safenet and try upload you might get the same result.
DNS lookup seems slow at times and the launcher froze in a way I’d not seen before when allowing the demo app.
http://explore.yvette.safenet
=> partial scan only
http://ar.udhr.safenet
http://el.udhr.safenet
http://en.udhr.safenet
http://eye.eye.safenet
http://fr.udhr.safenet
http://hello.safenet
http://ko.udhr.safenet
http://se.safenet