MaidSafe Dev Update 23th August 2016 - TEST 8

This is my vault after 5 days 24/7 :

        2016-08-29 20:53:20	stdout	INFO 20:53:20.677716085 [routing::stats stats.rs:172] Stats - User (Request/Success/Failure) - Get: 2785/2067/718, Put: 751/747/2, Post: 245/244/1, Delete: 0/0/0, GetAccountInfo: 123/116/7, Refresh: 7122
        2016-08-29 20:53:20	stdout	INFO 20:53:20.677701575 [routing::stats stats.rs:162] Stats - Hops (Request/Response) - GetNodeName: 106/152, ExpectCloseNode: 838, GetCloseGroup: 40490/73159, ConnectionInfo: 15643, Ack: 198450, GroupMessageHash: 115700
        2016-08-29 20:53:20	stdout	INFO 20:53:20.677687966 [routing::stats stats.rs:158] Stats - Direct - NodeIdentify: 376, NewNode: 1043, ConnectionUnneeded: 135
        2016-08-29 20:53:20	stdout	INFO 20:53:20.677646786 [routing::stats stats.rs:151] Stats - Sent 463000 messages in total, comprising 640646955 bytes, 1980 uncategorised, routes/failed: [176321, 7750, 3390, 3334, 3362, 3211, 3012, 2849]/1346
        2016-08-29 20:51:21	stdout	INFO 20:51:21.443737077 [safe_vault::personas::maid_manager maid_manager.rs:192] Stats - 9 client accounts.
        2016-08-29 20:51:19	stdout	INFO 20:51:19.853871488 [safe_vault::personas::data_manager data_manager.rs:657] Stats : Client Get requests received 1928 ; Data stored - ID 193 - SD 59 - total 196427226 bytes
3 Likes

Like every sites listed in this thread my site disappeared. It didn’t even last one day. And now TEST8 network has only 131 nodes (yesterday it had 138 node). All this is very sad.

Maidsafe should do something about it. I repeat something I said several times but I propose to reduce drastically the PUT balance until safecoins are introduced. For example, from 500 to 50. That’s enough to create great sites. Videos and mp3 can be used on alpha network.

Note: I will try again next Friday under a new account and a new name. So it’s not too late for sending me your vault ids along a label and possibly a URL so that I add them to my site. I will relaunch it under a new account and a new name, normally next Friday.

2 Likes

@tfa OTOH think about all the good data they must be getting on failure modes for the current implementation!

In addition to the plain mystery of accounts just disappearing, but not really, since they can’t be re-registred.
I’ve had one account end up with 0/0 PUTs. That makes it impossible to do anything interesting (like authorize an app).
I got one account where it consistently fail on authorization during “creating application directory” step for the demo app.

I’m sure it’s a treasure trove of debug logs for @MaidSafe

6 Likes

I’ve given up on this Testnet now. The vault chugs along nicely doing whatever it’s doing but nothing else works at all. The launcher doesn’t connect to the demo app any more and hardly any sites load. Time for 8a?

1 Like

I don’t know how vault reputation will be implemented but would be interested to know more about that. I’d expect time up will be an important factor, if the network prefers nodes that have heritage that might prevent new nodes causing too much disruption. Even perhaps a signed note from the group in case it goes down for a short while would allow backbone nodes to return quickly. The liability atm seems to be one weak bandwidth can cause havoc. An interesting problem, which I expect will be one that benefits from a large network. It’s good to see the tests fail as that must help devs understand where the limits lie but I’m keen to see some vault management introduced, as that will be key.

1 Like

This one is new to me. Those devs these days! They didn’t tell us probably for testing purposes :yum:. Curious what we’ll learn in tonight’s update.

3 Likes

Would love to have known that. Would that be the kind of thing @frabrunelle could ask the dev team when they’re putting together the release notes?

3 Likes

Wasn’t it meant to be like this?

6 Likes

Sorry @polpolrene. It is my mistake there. I was intending to refer to the test net at that point of time when the issue was raised. Before alpha for a couple of testnets we had hosted vaults in droplets with throttled bandwidth. Not the ones right now. I will edit the comment in that issue

3 Likes

Ohw no worries. I knew about some testnet some time ago that was created to test stability to the max.

2 Likes

I think OP says Aug 23th

I am seeing a pattern lately…lol. :joy:

Dev Update: “We will be operating a network that will be wiped periodically. We will be trying to throw everything we can at it to see what happens and analyze the network. It may be slow, have connection issues or lose data.”

Users: “My data is gone!” “My Demo app wont connect!” “It seems to be super slow.”

6 Likes

Yep… agree. Seems some people don’t pay enough attention and clog the forum. Maybe a big bold notice of the caveats stickied at every new update and every download page @frabrunelle

1 Like

Ehh…I think a lot of it is enthusiasm as well, better the devs know redundant info than assuming a glitch is baked into the testnets and not reporting it…just cracks me up a bit. :slight_smile:

1 Like

Why can a vault not simply do a speed test after installation and decide whether or not to run based on preset criteria?

The ideas behind the Safe network I’m in love with and have been following this forum since I became aware of the Alpha. I have no programing skills what so ever. Even html gets me scratching my head.:confused: So JUST incase this helps anyone attached is a screen shot of a vault running on a slow connection of 1mb up and 10 down.

I thought it was interesting that when you first run the vault you get lots of network activity but then it drops off over time. I would think when the network launches you are going to get loads of users with this type of setup. Wish I had the skill to contribute more!

8 Likes

I just PM’ed you. At the moment I don’t have much time, but one of the first things I’d like to do is make a full documentation on the CSP headers, their specific meaning and reasons and what that means for you as a web-developer. This includes a trouble shooting guide of common problems, how to find and address them.

My biggest issue was that I had a dependency (date-picker), which insisted on injecting CSS-Code. As this was happening at start-up time, the error message wasn’t very clear on the source of that problem. As inlining CSS has kinda become a thing in the Reactjs-Community, this might be a big problem for adoption in that sphere. Though I completely see the reasoning behind disallowing inline-CSS (you can do a lot of evil. But then, the JS can still inject style-attributes on tags, so… you aren’t really limiting stuff here). I am torn on this one.

I think we need to take a deep look with some of the feedback you’ve already given and I have experienced and decide which rules in there do actually make things more secure and which are just making it harder to develop for it (and thus limiting adoption).

1 Like

Inline CSS is what I’m hitting too. RemoteStorage.js use it so they can include an html widget, button icons etc along with everything else all in a single js file. I haven’t looked into the security pros & cons yet, but I’m not really qualified to judge anyway. I look forward to your input, meanwhile I’ll look for workarounds.

Edit: @lightyear please sign up to the safedev forum and see the CORS thread I started there.

I have added again my sites under new names. I have implemented a program that periodically reloads their files. This is an experiment to see if that helps keeping them available on a longer period.

New names are (depending on your browser):

You can also make your own vaults visible to the world in safe-galaxy3. If you PM me your vault id(s) and optionally a label and a URL, then I will add a button displaying them prominently.

For information TEST8 network has currently 130 nodes (and 5 are mine).

5 Likes

Finally, the experiment lasted 2 days and a half. The client part gets a time out error now.

The sites are still available but are frozen at around 10:43 UTC today. At that time the network had 129 nodes. Another experiment is to see how long they will last in that state.

I will stop my nodes (progressively) and destroy my droplets. I will recreate them when a more reliable network is available or if I have another idea for tests.

Edit: http://mystats.safenet/ had the same fate: it is blocked at 2016-08-28 23:30, but it is still available today.

2 Likes