Wiki: filesharing / free4all

Edit this page: filesharing / free4all

But do use more bandwidth. And more IPv6! Many devices choose to not use Legacy IP / will automatically disable it.

Search & Indexing

Shares

Server software

• Copyparty is the server software that is used on many of these servers
• RandomFtpGrabber - randomly downloads stuff, Python script
• lftp - nice cli, supports HTTP dir listings, can do recursive transfers (mirror), supports queuing
• rclone - file transfer swiss army knife, can do FTP, SFTP and HTTP dir listings as well, can do mounts and ncdu-like view

Discussion

FileShare Software

What else do you use, and what is your opinion of software such as CopyParty, Retroshare, and *arr Phsare? I find it questionable that many servers here currently provide unencrypted traffic via FTP or HTTP protocol; that's why I welcome CopyParty, despite its overhead. And what actually happened to RetroShare and protocols such as DCC+?

I don't like CopyParty since it seems it does not play nicely with wget. I understand that active FTP is a questionable protocol entirely, but I don't see a threat model why encryption matters. In a filesharing context I'd understand the need for file-integrity, which would be included in encrypted protocols. Encryption would ensure confidentiality if traffic would be intercepted or eavesdropped - I don't assume that this takes place on any wired connection here at the CCC. Encrypted protocols such as HTTPS usually imply certificates and certificates often imply DNS. DNS implies having an IP first. If you prepare your server at home you cannot do that. Also, especially with not-server-hardware such as old PCs or Raspberries sometimes crash and getting new IPs afterwards. Since most fileservers are accessed from the internet - you see in your longs that a lot of traffic directed to hosting providers - with offering/preferring http when serving - I have hopes that hosting provides transparently proxy and cache the traffic, reducing their bandwidth requirements. Therefore, offering plain http with caching headers allows a one-to-many transfer, while https would only be one-to-one - adding a sustainability aspect to our filesharing hobby.

Why does everybody still uses shitty protocols for sharing (big) files? Ever heard of peer to peer technologies like bittorrent? Decentralisation ftw!

» I don't like CopyParty since it seems it does not play nicely with wget

Please call the copyparty helpdesk (extension 92029) -- wget shouldn't be having any trouble with cpp, so curious to see what you're encountering.

» wget shouldn't be having any trouble with cpp

I use wget --tries=10000 --continue --progress=bar --show-progress --timeout=300 --retry-connrefused --recursive --level=0 --timestamping --no-remove-listing --no-parent "$url" for downloading, this works fine to mirror whole directories via ftp or http-index. If I want to replicate that "whole dir" behaviour I need to use ?zip and add -O in wget to safe as a file with a name. So to be exactly, the --recursive seems not to work with cpp.

-.-.-.- You may be looking for --reject-regex="?zip=crc"

» Ever heard of peer to peer technologies like bittorrent

Sure, but finding appropriate trackers is a hazzle. Also, torrent would surpass the CCC - that is not the filesharing spirit I am looking for at the congress - I like the idea of bringing "servers" here to share files. Going from my access logs, a lot of people download from their home or hosted servers - I'd therefore assume that a torrent may live way longer than congress and could get out of hands. I think torrent is a great thing for legal obtained file - but for the things I host - I don't think that they are free of any watermarks - with the wiki beeing public and all IPs per default beeing public I see the risk of data potentially beeing attributable to me and data staying online for way longer than the congress - exposing me to a potential legal risk while putting me in a position that I cannot control the flow of data anyhow. Torrent is to my knowledge a download-only protocol, reducing the possibility to receive any data. Personally, I'd rather prefer IPFS over torrent because it can be integrated in a webpage and integrate seamless. Maybe the CPP people can implement sharing files via IPFS and we can see the adoption the next congress (ipüfs add/pin + x-ipfs-path header).

IPv6 Setup on modern Linux

I think the following used to work with netplan, however I'm out of luck this year (using Ubuntu 24.04 on ARM64). Why is it still so damn hard to get Linux to accept SLAAC?

network:
  version: 2
  renderer: networkd
  ethernets:
    eth2:
      accept-ra: yes
      ipv6-privacy: no

just use systemd-networkd instead of netplan.

[Match]
Name=eth0

[Link]
ActivationPolicy=up
Multicast=true
RequiredForOnline=yes

[Network]
DHCP=yes
DNSDefaultRoute=yes
EmitLLDP=nearest-bridge
IPv6AcceptRA=yes
IPv6DuplicateAddressDetection=1
IPv6PrivacyExtensions=kernel
LLDP=yes
MulticastDNS=yes

[DHCPv4]
Anonymize=yes
UseDNS=no
UseDomains=no
UseHostname=no
UseNTP=no
UseRoutes=no

[DHCPv6]
UseDNS=no
UseDomains=no
UseHostname=no
UseNTP=no

(probably without privacy extensions)