Jump to content











Photo
- - - - -

AstralBoot, using IPFS (InterPlanetary File System) as boot server


  • Please log in to reply
9 replies to this topic

#1 Nuno Brito

Nuno Brito

    Platinum Member

  • Team Reboot
  • 10492 posts
  • Location:boot.wim
  • Interests:I'm just a quiet simple person with a very quiet simple life living one day at a time..
  •  
    European Union

Posted 14 January 2018 - 06:46 PM

Astralboot is a golang server that provides network services to boot virtual and metal machines from pxe boot. The following services are provided

  1. DHCP , for ip allocation and boot information
  2. TFTP , simple file transfer
  3. HTTP , for serving images and configs

It can pull its data files out of ipfs, which means that they are downloaded on request and then stored locally.

 

I'm still thinking on what could be the big advantage of this system compared to the common DHCP + TFTP combo. Then I see the potential, the files are referred by hash and doesn't matter where they might be located.

 

This means that we would just need to the specify the hash signature for the ISO and end-users would be able to download this image with more changes of success than hosted at a single location, which is useful when URLs or full sites disappear after some years.

 

https://github.com/ipfs/astralboot

 

:cheers:


  • ZEE likes this

#2 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 13988 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 15 January 2018 - 08:43 AM

Hmmm.

 

Right now. more than anything else it looks like a solution in search of a problem. :dubbio:

 

IPFS in itself may represent a great thing, but this stuff? :unsure:

 

:duff:

Wonko



#3 Nuno Brito

Nuno Brito

    Platinum Member

  • Team Reboot
  • 10492 posts
  • Location:boot.wim
  • Interests:I'm just a quiet simple person with a very quiet simple life living one day at a time..
  •  
    European Union

Posted 15 January 2018 - 10:18 PM

Looks like a very useful thing for multibooting without worries on getting the correct ISO images.

 

Not sure how much problems this could cause with Windows ISO images since they would be easily uploaded/downloaded/shared by any IPFS client.



#4 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 13988 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 16 January 2018 - 10:27 AM

Looks like a very useful thing for multibooting without worries on getting the correct ISO images.
 
Not sure how much problems this could cause with Windows ISO images since they would be easily uploaded/downloaded/shared by any IPFS client.


Let me understand the difference.
Till now, if you want to boot from - say - this or that version of Ubuntu you:
1) use a web browser and google to find the Ubuntu site
2) you find a download link for a suitable .iso
3) you (slowly) download it, saving it locally
4) you (hopefully) boot (quickly) to it, do the whatever prompted you to boot from it, and then go to bed
5) the day after you decide to boot again the Ubuntu, you select the locally stored .iso and boot (quickly) to it

In a hypothetical IPFS scenario:
1) use a web browser and google to find a site with the IPFS id/token/whatever
2) you find a IPFS id/token/whatever for a suitable .iso
3) you (slowly) download it, saving it locally
4) you (hopefully) boot (quickly) to it, do the whatever prompted you to boot from it, and then go to bed
5) the day after you decide to boot again the Ubuntu, you select the locally stored .iso and boot (quickly) to it


In the new revolutionary paradigm:
1) use a web browser and google to find a site where the IPFS files are listed
2) you find a download link for a suitable .iso
3) you download it (slowly) while at the same time booting to it, do the whatever prompted you to boot from it, and then go to bed
4) the day after you decide to boot again the Ubuntu, provided that you manage to find it locally, you quickly boot to it, more likely you loop to #1

As said I can understand the potentiality of a semi-de-centralized web, i.e. IPFS, it is only the actual IPFSpxe booting that seems to me like - besides the limited and experimental status of the thingy - like not particularly useful

:duff:
Wonko

#5 Nuno Brito

Nuno Brito

    Platinum Member

  • Team Reboot
  • 10492 posts
  • Location:boot.wim
  • Interests:I'm just a quiet simple person with a very quiet simple life living one day at a time..
  •  
    European Union

Posted 16 January 2018 - 04:07 PM

With IPFS, the ISO is automatically downloaded from wherever on the Internet it might be.

 

For the end-user, there is no URL or manual process. The script will automatically look for the hash of the ISO and download it locally.

 

The command would go in the lines of:

ipfs get myHashToBeDownloaded

https://ipfs.io/docs...mands/#ipfs-get

 

 

The question is if whether exists such ISO hash in the network. You need have at least one computer somewhere to be hosting the file with that same hash. For Ubuntu it will work great. I imagine for example that if/when Ubuntu gets bought by Oracle that they will discontinue the domain http://ubuntu.com and that would be the case when using only hashes instead of a fixed URL path would shine.



#6 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 13988 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 16 January 2018 - 04:32 PM

Still, you have to either know the hash or do some sort of lookup for it, at least once, or accept a pre-compiled (by whom?) lists of such hashes[1].

 

But as said, I do understand the potentialities of IPFS :) , I am failing to see the advantage of pxe booting from IPFS (as a matter of fact I am failing to see any advantage in non-local pxe booting and only see very small ones in local pxe booting, outside a given server-client specific setup or particular setups, such as a repair or forensics laboratory). 

 

:duff:

Wonko

 

[1] I hope you understand the potential risk of blindly booting from a "hash" provided by a third party, don't you?



#7 Nuno Brito

Nuno Brito

    Platinum Member

  • Team Reboot
  • 10492 posts
  • Location:boot.wim
  • Interests:I'm just a quiet simple person with a very quiet simple life living one day at a time..
  •  
    European Union

Posted 16 January 2018 - 04:39 PM

The hash is always the same, doesn't change and the user gets the same file that the author intends (this is verified).

 

It is the same thing as looking for SHA1 fingerprint and getting the file associated. Outside of the IPFS portion that puts the files on your folder, it is the same functioning as other boot tools.



#8 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 13988 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 16 January 2018 - 05:31 PM

The hash is always the same, doesn't change and the user gets the same file that the author intends (this is verified).

 

It is the same thing as looking for SHA1 fingerprint and getting the file associated. Outside of the IPFS portion that puts the files on your folder, it is the same functioning as other boot tools.

 I know, noone ever said that the hash would change, I am perfectly familiar with the concept of hashes.

 

Right now, if you go to - say - the Ubuntu site, you have to look for a suitable .iso, then you get a link to download it:

https://www.ubuntu.c...ownload/desktop

with no hash/no immediate way to check what you downloaded, but if you go on the actual good site:

http://releases.ubuntu.com/

http://releases.ubuntu.com/artful/

 you get besides the link to the .iso also hashes to verify that the download was correct:

 

The IPFS has the advantage that the actual .iso link is the hash and viceversa and that a link is actually a link to all possible mirrors of that particular hash.

 

Still you have to look for the hash at least once, or have someone that provides you the hash (again at least the first time), just like you have to look for the name and URL of the .iso in our current, traditional, web.

 

The hash in itself is not "human readable" and it is obviously not possible to "decode" it to a given filename, and, even if this was possible, nothing would guarantee about the contents of a file with a human readable name if not the source that provides the file and the hash.

 

It would be appreciated if you could read my replies and possibly answer to what is said in them (as opposed to whatever else that it isn't in them).

 

 

:duff:

Wonko


  • Nuno Brito likes this

#9 Nuno Brito

Nuno Brito

    Platinum Member

  • Team Reboot
  • 10492 posts
  • Location:boot.wim
  • Interests:I'm just a quiet simple person with a very quiet simple life living one day at a time..
  •  
    European Union

Posted 16 January 2018 - 09:54 PM

I was under the impression this tool would be all setup for just grabbing files from IPFS.

 

:mellow:



#10 Wonko the Sane

Wonko the Sane

    The Finder

  • Advanced user
  • 13988 posts
  • Location:The Outside of the Asylum (gate is closed)
  •  
    Italy

Posted 17 January 2018 - 02:15 PM

I was under the impression this tool would be all setup for just grabbing files from IPFS.

 

:mellow:

Sure it is, still the point remains WHAT will be got?

WHO guarantees that the provided hashes correspond to WHAT?

 

https://github.com/ipfs/astralboot

 

 

 

The default hashes for booting are included in the git repository , put them into place by running.
cp refs.toml.dist refs.toml

 

https://github.com/i.../refs.toml.dist

boot = "QmXRvmqk9DjhGsLPeWF1nZ8EryGYnD8j4iQoFLSh9hUJbH"
#rocket = "QmdLDMoJBMwtFT13PLPP4SF3oxFnrNLF1kUdDqnxwYxDaZ"
#spawn = ""

:duff:

Wonko






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users