SAFE Network Dev Update - March 26, 2020

Yeah, this is a wording tweak, rather than a flow change I think. We need to instill in people the permanence of the operation they are instigating, but still offer a safety net—the 5 second undo window—where we can, just in case of a button fumble etc.

It will cost a very small amount yes. But we are going to need to tune when and where these indicators/dialogs are shown, as it might get a little fatiguing if they are always interrupting the flow.

We have the componentry designed which is the important thing, and they can be deployed when we learn more about the cost, and user tolerance to levels of expense, closer to launch.

If you check out the feature status board, you’ll also see we have spending limits and thresholds (both global and on individual PUTs) planned to help in this regard, but they are not considered part of the Minimum Viable Experience (MVE).

For the most part, ‘moving’ data won’t really be moving anything, just putting an entry in an index, so the cost would be trivially small.


Like a lot of things we use. Cars, devices running off disposable batteries and so on.

We fill the car up with Petrol (gas for americans) and rarely worry about starting the car thinking I am using petrol at so many mL per second/minute. We simply keep an eye on the gauge every so often (or at least most do) and not worry so much about the fractions being used along the way.

We need to somehow create that atmosphere where we know usage means using up a resource that needs to be topped up or replaced when used up. Our society understands that from kids using batteries in our toys to adults using petrol in our cars.

The more its used the faster we need to replenish the tank.


Thanks for the results @Southside. I’ll take a look at it.

We might not need this right now. We’ve been working on something else for you guys to try out.
Details coming soon :slight_smile:


By the way, this has a perfect explanation :slight_smile:
What must’ve happened here is that the Client would’ve failed to bootstrap to the section.
We wait for 30 seconds before trying to bootstrap again and it would’ve succeeded the second time and uploaded the file in .76 seconds.


Yeah, I was thinking of it akin to using mobile data. I really don’t want my phone to ask me to confirm every Kb action, maybe I’d like to check my balance every once in a while, or probably want to know when I’m spending more than an average month.

But I probably would want to take a beat before making a long distance premium call, or beginning a pricy data roaming session.


Unfortunately I have to say no. I bought the machine 2017 to be safe machine to do all my crypto stuff. There is no significant amount of money in it anymore, but I want to continue to have it as private as I can. If there would be some kind of problem in the future I would hate to be suspicious about anyone else than me.

@bochaco is there any benefit for Maidsafe to find out exactly what’s wrong with my system, or would it be as good if I just update my system to 18.04 for example?


If you are using this as a safe machine, I would upgrade it. There are a lot of security fixes in kernel upgrades. So I would recommend upgrade to latest LTS version for your own safety. However if this is s safe machine I would not run alpha/beta test code on it either.

If you want to run the tests etc. on that machine I would seriously consider a docker container or similar.


@toivo, I totally understand. I used to have a couple of well-off housebound personal clients that I set up low-spec linux laptops for the sole purpose of doing telephone banking. No way were bored grandkids getting to play with these. The machines were totally locked down and stripped
of everything I could manage to remove bar a browser set to the banking home page. Only themselves myself and the highly trusted accountant had any access at all. Sadly they both passed away last year within days of each other and the accountant retired. It was fairly good fees for simple work.

As @dirvine has said, it would be best to upgrade this to 18.04 at least. If you would still like to help test, then install VirtualBox

Make a linux virtual machine of any flavour you like and get tore in , knowing that in the incredibly unlikely event of anything going seriously wrong, the only thing that can possibly be affected is a single file, totally isolated from all your valuable crypto stuff.


Try my script

plug, plug

1 Like

I have been looking with interest at certain directories in the latest safe-api on the github page :slight_smile:

1 Like

I just noticed the other topic :slight_smile: :slight_smile: :slight_smile: :slight_smile: :slight_smile:

Please DO NOT run any of my test-scripts in this new resource. It would be far too easy to clog it up with random data.
I’m sure we will find a way of responsibly testing the performance here

1 Like

Perhaps instead of rewording the permanence bit, rather use a different word than undo? Perhaps Cancel or Abort with a 5 second timer counting down?


Hi, Sorry for the delay, been distracted with the shiny new shared section :slight_smile:

Baby Fleming is like many babies, shows a lot of promise, has much maturing to do but is basically sound. I’m sure it will get good parenting from the devs and the community god-parents will always be around to keep it right.

Well thats enough sentimental sh*te… I’m basically impressed, there are some rough edges to be ironed out, we will doubtless uncover some things that will need sorted but we have a very responsive dev team.
Hopefully we will get a lot more eyes on it soon to look for issues that so far have been missed.
Self-professed non-techies should start believing in themselves and give this a try. You will get lots of help and support here.

One of the simplest and easiest steps you can take to get into it right away is to download VirtualBox and make a VM (Virtual Machine) on your computer in which you can play to your hearts content in the secure knowledge that any mistakes you may make can go no further and all your precious files and programs are securely isolated. You can put Windows or Linux on your VM. I say Linux cos most of us who can help right now are mainly Linux users but we need to expand that to include Windows-capable helpers too.

Do not be scared to ask for help - There is no such thing as a stupid question. We all once knew nothing - or felt we did.


Thanks a bunch, I will do this!


Thanks heaps @Southside. I’m afraid I will have to wait for the big shiny button David said would be there so that idiots like me can use the system. I’m having fun reading all the posts written in a different language than English. I can feel progress occurring but I’m still in a very dark room


Nonsense, you are simply in a dimly lit corner of a big room cos you have not asked or been told where the light switch is.
I dare you , install VirtualBox, then shout for more help.
It can’t be that difficult, even I managed to do it with a lot of help from others. Now its my turn to share that knowledge and pay back what I was given.
But the first thing you must do is believe in yourself and remember there are no stupid questions.
We all (mostly) have a bit more time on our hands these days so lets make the best use of it.


My experience so far:

By running all 8 nodes with safe vault run-baby-fleming my machine runs out of memory by nodes working on 100% processor load after couple minutes (not measured). Interestingly, nodes 7 & 8 do not take part in this – they are tiny processes with 0% load all the time.

When I figured out how to run less nodes, things got better. With 1, 2 and 3 nodes all was fast and snappy. No constant load on processor and CLI commands responded quickly.

With 4th node joining the party, things started to get messy. It took couple minutes for processor load to settle down (not measured, by about 2-5 min.), during this time CLI commands had some lag. Last node to stop eating my processor was newly joined node, the 4th one.

5th one spoiled everything. Until now (about 1h after joining) processor load is 100%, although nodes 1 & 2 are not eating my memory, unlike 3, 4 & 5, which currently eat 20% memory each and counting. 1 & 2 oscillate around 1% mem each, but are higher on CPU (~60% each, while 3, 4 & 5 ~23%). I’m not sure if it’s like that from the beginning, or was it different just after #5 joined. CLI commands lag heavily, even simple, like safe keys balance, and sometimes I get errors. Just like with default 8 nodes.

After running more safe keys balance commands, vaults 1 & 2 started to grow.

Next step was to try to kill vaults from the youngest ones. After killing 5, 4 & 3 in ~5min intervals, nodes 1 & 2 still use all the processor and grow in memory usage, so I’m going to kill them as well. safe keys balance stopped to get errors (“Connection timed out when bootstrapping to the network”) and return balance after ~10s.

After killing vault #2, genesis vault memory usage dropped from 15% to 3% and CPU also from 100% to ~70%. Commands run fast (~1s), although not as fast as with fresh 3 nodes. Time to kill it too!

BTW my system is little outdated: 1.8 Core 2 Duo, 4G mem, SSD drive, Arch Linux.


TBH that does not surprise me at all given you have only 4Gb memory.
Thank you for trying so diligently to get this running.


Tried to put big file. Succeeded up to 700MB movie file, that lasted ~7 min. with 100% CPU, safe-cli process used 1.4G memory during that, and each vault’s directory has grown by ~700MB. After upload everything returned to normal, vault processes using almost no mem/cpu. This was with 3 vaults running.


Thank you for the heavy work team MaidSafe! I add the translation into Bulgarian in the first post :dragon: