some feedback

[koo5]

hi, some feedback:

what is number of backup copies? how does it relate to the number of suppliers, redundancy and requested size?
the mystery of this setting makes me realize i dont really know how the backup works, even after reading your whole site.
this setting needs explanation in the program or documentation.
does the backup work like ubuntu one? is it automatically updated when the data change? when/how does the backup happen?


tried to register on the forum, but it is a pain. kaptcha was too hard:)
(maybe needs better software, like quora, preferably something with openid)


and a feature request: not all data i want to back up needs encryption. could we have selectively unencrypted backup between friends, too?

[Veselin Penev]

On Sunday, April 14, 2013 1:24:17 PM UTC+4, koo5 wrote:

hi, some feedback:

what is number of backup copies? how does it relate to the number of suppliers, redundancy and requested size?
the mystery of this setting makes me realize i dont really know how the backup works, even after reading your whole site.
this setting needs explanation in the program or documentation.
does the backup work like ubuntu one? is it automatically updated when the data change? when/how does the backup happen?

This is a number of copies of single folder you wish to keep on remote machines. When you click to do a backup on a folder - you made a one backup copy. Typically you want to keep only one copy, but for some tasks you can keep more copies - oldest will be deleted. You can set schedule to backup the folder at given time, so all this will be automatically. If content of this folder is changing periodically - you can keep some history for your data this way.

Number of suppliers is just a number of remote machines that keeps your data. Less suppliers - may be faster but less secure. More machines gives you more strong distribution of your data and so more stable storage. For 64 suppliers you can lose up to 10 remote machines at once and still be able to restore your data, this is because of redundancy and ecc codes. However we did not tested the soft on such many suppliers yet. I think 7 suppliers at the moment should give you the best protection.

The requested size is how many megabytes you need for your data. This volume will be taken from your suppliers.

I definitely should provide more info about every option in the software. Will do it soon.

This works differently from Ubuntu One or any other online backup service.

Your data is stored not on the single server (or some another centralized solution) but on the machines of other users - just like you. Absolutely distributed.

You can do a single backup copy of a given folder or set a schedule to start this at given time automatically. 

We did not monitor your hard disk content to check when the folder content were changed. 

So No - it did not automatically updated when the data change. May be I will add this feature one day.

 

tried to register on the forum, but it is a pain. kaptcha was too hard:)
(maybe needs better software, like quora, preferably something with openid)

ok. I changed the captcha settings to make less noise.

you need to enter the secret answer: "Vincent Cate" :-)

 

and a feature request: not all data i want to back up needs encryption. could we have selectively unencrypted backup between friends, too?

No, we dont have this at the moment, and I dont think we will. 

Why do you want unencrypted backups?

In our understanding, storing the data on friends machines is not so secure as distributed in the whole world.

 

[koo5]

thanks for the reply!

i think i just finally realized the green arrow starts the backup! Oh and the other one is for sheduling, and thats gonna be delete and some sort of update? Those four icons need a tooltip. Now it's starting to make sense.

now its saying its preparing blocks and delivering them, interface has slow responsivity, lots of cpu use, but i am using jnettop to watch my traffic and there is not any corresponding heavy net activity on my computer, just some peer to peer /xxx.xml chatter.

i will send a message from the program later so you can check the logs.
Now the interface besides the debug icons is just black and dead.
I had to kil it and now it refuses to start again, where is the lockfile?

This is a number of copies of single folder you wish to keep on remote machines.

So number of copies is actually number of historic versions + 1. Ok.

 

tested the soft on such many suppliers yet. I think 7 suppliers at the moment should give you the best protection.

so number of supplier is the number of chunks my data is split into. Redundancy stays constant.

7 suppliers = can loose 3 at once? Loosing 10 people from 64 at once sounds very likely to happen sooner or later. Will there be more redundancy options?

I definitely should provide more info about every option in the software. Will do it soon.

All the other options were clear enough or not interesting for me.
I just pointed out what were the showstoppers for me, as a user. Problems with texts in interface, lack of obvious function/tooltips, backup settings explanations missing. The rest of the user interface is ok and everything seems clear.

| ok. I changed the captcha settings to make less noise.

okay, ill try again sometime..maybe.. Still would be nice if you found an openid addon to phpbb.  Remember you are dealing mostly with privacy freaks here

Why do you want unencrypted backups?

In our understanding, storing the data on friends machines is not so secure as distributed in the whole world.

who says my friends are not distributed in the whole world?:) And if i would make them use this app, the idea that we wouldnt encrypt our data for each other, and share them instead, comes naturally. I would love to see my friends music collections.

As a programmer, i admire your code, the use of automata, the transparency the debug tools and the directory structure provide, and i hope to dig deeper and learn from it. have a nice day !

[Veselin Penev]

понедельник, 15 апреля 2013 г., 20:07:49 UTC+4 пользователь koo5 написал:

thanks for the reply!

i think i just finally realized the green arrow starts the backup! Oh and the other one is for sheduling, and thats gonna be delete and some sort of update? Those four icons need a tooltip. Now it's starting to make sense.

Using tool tips is nice thing, why not?  

now its saying its preparing blocks and delivering them, interface has slow responsivity, lots of cpu use, but i am using jnettop to watch my traffic and there is not any corresponding heavy net activity on my computer, just some peer to peer /xxx.xml chatter.

Interface is not so great that is true.

I was thinking to use web browser to do it, but finally give up on this idea and made it using wx.HtmlWindow.

It become slower because lots CPU spends to encrypt the data when you start the backup, so gui just waits for main process a lot. 

If you did not see any traffic it may be possible that p2p transports did not work properly - so data did not transferred at all. 

However if you are connected to the central serve it should work.

Do you see green icon in the tray or it is gray? 

Also, it seems you are running from sources, why you did not install the .deb package from the downloads? 

The .deb release should update your software automatically every day using cron.daily script.

I give sources just to provide the code, you need to install dependencies by hand and be sure the main process works.

Try to run from command line in this way:

cd datahaven-revXXXX

python dhn.py

 

i will send a message from the program later so you can check the logs.
Now the interface besides the debug icons is just black and dead.
I had to kil it and now it refuses to start again, where is the lockfile?

Logs are in the ~/.datahaven/logs

Seems like some exception appears in the main process so it become unresponsive and gui fails.

Did you have installed wx-python ?

To be fair I did not tested yet running from sources on clear machine.

I think I have to do this and make some manual like: "how to install DataHaven.NET from sources step by step".

 

This is a number of copies of single folder you wish to keep on remote machines.

So number of copies is actually number of historic versions + 1. Ok.

 

tested the soft on such many suppliers yet. I think 7 suppliers at the moment should give you the best protection.

so number of supplier is the number of chunks my data is split into. Redundancy stays constant.

7 suppliers = can loose 3 at once? Loosing 10 people from 64 at once sounds very likely to happen sooner or later. Will there be more redundancy options?

Yes. Here is the ecc maps and correctable errors we use:

64: 10,

26: 6,

18: 5,

13: 4,

7: 3,

4: 2,

2: 1,

 

I definitely should provide more info about every option in the software. Will do it soon.

All the other options were clear enough or not interesting for me.
I just pointed out what were the showstoppers for me, as a user. Problems with texts in interface, lack of obvious function/tooltips, backup settings explanations missing. The rest of the user interface is ok and everything seems clear.

Thanks. I think I am working in right direction when using "http server <-> web client <-> gui" solution for making GUI.

It is very easy to make a new pages, we really do not need some fantastic design. But refreshing the gui is tricky in this way.

So I am working on this in last days. 

 

| ok. I changed the captcha settings to make less noise.
okay, ill try again sometime..maybe.. Still would be nice if you found an openid addon to phpbb.  Remember you are dealing mostly with privacy freaks here

Will check this.

 

Why do you want unencrypted backups?

In our understanding, storing the data on friends machines is not so secure as distributed in the whole world.

who says my friends are not distributed in the whole world?:) And if i would make them use this app, the idea that we wouldnt encrypt our data for each other, and share them instead, comes naturally. I would love to see my friends music collections.

No. I dont think DataHaven.NET will have any kind of data sharing.

 

As a programmer, i admire your code, the use of automata, the transparency the debug tools and the directory structure provide, and i hope to dig deeper and learn from it. have a nice day !

Cool. Thanks. I definitely should provide more info about DataHaven.NET on the web site, just dont have extra time to publish my work.

Take a look at another small project related to state machines: http://code.google.com/p/visio2python/

 

[nejaka osoba]

Those four icons need a tooltip. Now it's starting to make sense.

Using tool tips is nice thing, why not? О©╫

just in case we dont understand each other: i dont see any tooltips

Do you see green icon in the tray or it is gray?О©╫

i dont know, i didnt pay attention to the tray icon. And it seems to work ok now.

Also, it seems you are running from sources, why you did not install the .deb package from the downloads?

because i like to have and see the source for security and tinkering reasons and this app really interests me

The .deb release should update your software automatically every day using cron.daily script.

i would still be happier if i could keep up to date with the sources, manually. It would be nice if it was just a git (svn) pull away like with other projects.

I give sources just to provide the code, you need to install dependencies by hand and be sure the main process works.

iirc only one dependency was missing, it was easy

Try to run from command line in this way:

cd datahaven-revXXXX

python dhn.py

О©╫

exactly what i do.
Also figured out why it wouldnt re-start: there was an editor with one of the source files open. guess it greps for the filename:)

Did you have installed wx-python ?

seems so. everything worked.

Thanks. I think I am working in right direction when using "http server <-> web client <-> gui" solution for making GUI.

It is very easy to make a new pages, we really do not need some fantastic design. But refreshing the gui is tricky in this way.

So I am working on this in last days.О©╫

О©╫

thanks for all the explanation of the backup settings.

im sure you will be able to polish the interface over time, responsivity and clarity-wise. Maybe i will be able to help with the GUI texts in the future.

[Veselin Penev]

вторник, 16 апреля 2013 г., 23:29:58 UTC+4 пользователь koo5 написал:

Those four icons need a tooltip. Now it's starting to make sense.

Using tool tips is nice thing, why not? О©╫

just in case we dont understand each other: i dont see any tooltips

sorry for my english, I mean I will do it in future. Thanks for advice.

 

Do you see green icon in the tray or it is gray?О©╫

i dont know, i didnt pay attention to the tray icon. And it seems to work ok now.

Also, it seems you are running from sources, why you did not install the .deb package from the downloads? 

because i like to have and see the source for security and tinkering reasons and this app really interests me

The .deb release should update your software automatically every day using cron.daily script.

i would still be happier if i could keep up to date with the sources, manually. It would be nice if it was just a git (svn) pull away like with other projects.

Yeah. I thought about that.

I have registered the project on the google.: https://code.google.com/p/datahaven/

So I think I can open a svn repository here. 

 

I give sources just to provide the code, you need to install dependencies by hand and be sure the main process works.

iirc only one dependency was missing, it was easy

Try to run from command line in this way:

cd datahaven-revXXXX

python dhn.py

О©╫

exactly what i do. 
Also figured out why it wouldnt re-start: there was an editor with one of the source files open. guess it greps for the filename:)

Did you have installed wx-python ?

seems so. everything worked.

Thanks. I think I am working in right direction when using "http server <-> web client <-> gui" solution for making GUI.

It is very easy to make a new pages, we really do not need some fantastic design. But refreshing the gui is tricky in this way.

So I am working on this in last days.О©╫

О©╫

thanks for all the explanation of the backup settings.

im sure you will be able to polish the interface over time, responsivity and clarity-wise. Maybe i will be able to help with the GUI texts in the future.

I am waiting for next release of wxWidgets, it have a new widget called wxWebView. Currently I am using wxWidgets 2.8 and DataHaven.NET GUI is done using wxHtmlWindow. So I hope wxWebView will be better and so I can improve the GUI easy.

I will be happy for any attention to the DataHaven.NET project, any help appreciated. Thanks!

[koo5]

hi.

do you know about the high CPU usage when sending blocks?
i tried doing some profiling but there were various things i had to solve first and the report didnt make much sense anyway, there is time spent in wx app mainloop() but i dont know if it is cpu time or real time.



Dne neděle, 14. dubna 2013 11:24:17 UTC+2 koo5 napsal(a):

[Veselin Penev]

Yes, I know about it.

The time is spent to encrypt the packets.

I was using cProfile python module to optimize twisted reactor usage and decrease the number of delayed calls.

Something like that: 

    import cProfile

    import p2p.dhnmain

    ret = cProfile.run('p2p.dhnmain.main()', 'profile.report')

Than using pstats I can watch the results from generated profile.report file.

There are also some graphical tools to do that also.

суббота, 20 апреля 2013 г., 2:30:32 UTC+4 пользователь koo5 написал:

[koo5]

the deb package that is currently available, in the preinst script:
cp /usr/share/datahaven/datahaven.desktop /home/$SUDO_USER/.config/autostart/
fails, was meant to be in postinst?

[Veselin Penev]

Hm...

If $SUDO_USER is set - it should work, because we have create the
~/.config/autostart in previous command:

mkdir -p /home/$SUDO_USER/.config/autostart/


If $SUDO_USER is not set it should not rich this line.
Strange, I do not understand why it is get failed.

But you right - it will be better to move that into postinst.
Will do and reply.

Thanks for you reports!


10.05.2013 19:45 PM, koo5 пишет:

> --
> You received this message because you are subscribed to the Google
> Groups "datahaven" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to datahaven+...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

[nejaka osoba]

sorry, i should have said the problem was in the non-existence of the source file

all ok now, thanks

On Sun, May 12, 2013 at 11:18 AM, Veselin Penev <vese...@gmail.com> wrote:

Hm...

If $SUDO_USER is set - it should work, because we have create the ~/.config/autostart in previous command:

mkdir -p /home/$SUDO_USER/.config/autostart/


If $SUDO_USER is not set it should not rich this line.
Strange, I do not understand why it is get failed.

But you right - it will be better to move that into postinst.
Will do and reply.

Thanks for you reports!


10.05.2013 19:45 PM, koo5 пишет:

the deb package that is currently available, in the preinst script:
cp /usr/share/datahaven/datahaven.desktop
/home/$SUDO_USER/.config/autostart/
fails, was meant to be in postinst?

--

You received this message because you are subscribed to the Google
Groups "datahaven" group.
To unsubscribe from this group and stop receiving emails from it, send

an email to datahaven+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

--
You received this message because you are subscribed to a topic in the Google Groups "datahaven" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/datahaven/8ybPDmbZWFo/unsubscribe?hl=en.
To unsubscribe from this group and all its topics, send an email to datahaven+unsubscribe@googlegroups.com.