#ipfs

/

      • eibriel has quit
      • bastianilso____ has quit
      • eibriel joined the channel
      • dignifiedquire has quit
      • fleeky_
        ok i am doing a bit of sillyness right now and am wondering if anyone has any insight, so i am using ssh as a socks proxy to connect to my internal network in sf from berlin, ive installed ipfs on a freebsd box there and i can get a list of swarms and such from the console, wanted to check out the web interface so i set the api and gateway addresses to the internal ip instead of the loopback 127.0.0.1 address, i can connect to the webui but i c
      • annot actually get any info about peers.. any idea why?
      • also seems like adding files doesnt work .. so maybe i misconfigured the config file ?
      • does the gateway need to be writeable ?
      • fleeky__ joined the channel
      • jbenet
        fleeky_ it's because the API defaults to bind on 127.0.0.1:5001
      • fleeky_ for security. you can make it bind to the ip addr in the interface of your vpn -- but note that it opens it up to anything in that network.
      • fleeky_ what i usually do is use an ssh proxy and rebind my web traffic to go over that proxy, so i can use https://127.0.0.1:5001/webui and have 127.0.0.1 mean the machine i'm remotely connected to.
      • fleeky_ has quit
      • fleeky__
        but i cant do multiple socks proxies can i ?
      • the network i am on is trusted also you cant access it from the outside at all
      • although it would be amusing to do an ssh socks proxy inside of an ssh socks proxy
      • so i changed the gateway ip back to 127.0.0.1 , so i guess changing the api address is whats important in this regard, but its still not displaying anything useful
      • btw , would be a cool use of ipfs if you could ssh to an ipns hash for instance to ssh proxy into a box that is behind a nat
      • anewuser joined the channel
      • f4 joined the channel
      • jbenet
        fleeky__ check the js console-- i still thnk it's a permissions issue.
      • may want to use 0.0.0.0:5001 if you're sure the box is totally isolated from the outside.
      • for those macos people -- https://github.com/jbenet/macos-ssh-proxy makes ssh http proxying nicer
      • galois_dmz joined the channel
      • Kane` joined the channel
      • chris613 joined the channel
      • lkcl has quit
      • fleeky__
        js console ?
      • could be a permission issue
      • how to check the js console
      • galois_dmz has quit
      • i am getting a whole lot of connection refused in the daemon's output in bash console
      • jedahan joined the channel
      • oh btw , does it make sense to add internal ip addresses to the bootstrap list
      • was going to run ipfs on a few puters and was thinking it would allow them to connect to eachother faster that way
      • jedahan has quit
      • Mateon1
        fleeky__: IPFS uses MDNS, which means that a message is broadcast to the local network every once in a while so nodes find themselves automagically.
      • This might not work if machines are on separate routers, though.
      • fleeky__
        yeah i was just thinking for speeding things up , but i guess it happens fast enough then ?
      • Mateon1
        Yeah, MDNS is pretty fast
      • kaotisk has quit
      • And even if MDNS can't work (machines on separate routers), IPFS can still find the nodes relatively quickly, since it tries local addresses for each peer as well.
      • fleeky__
        ahk
      • Mateon1
        You can see what addresses your nodes will try by typing `ipfs id`, it will return a JSON object, including a "Addresses" list
      • More generally, if you get the peer ID of a node, you can query the addresses with `ipfs dht findpeer <peer id>`, this works from any node
      • galois_dmz joined the channel
      • jedahan joined the channel
      • IlanGodik has quit
      • fleeky__
        jbenet: i set ~/.ipfs and ~/go-ipfs to 0755 still same behaviour , also tried setting "Access-Control-Allow-Origin" to "http://internal-ip" no dice
      • kaotisk joined the channel
      • apiarian has quit
      • herzmeister has quit
      • herzmeister joined the channel
      • jbenet
        fleeky__ :( -- maybe dignifiedquire or daviddias or kubuxu can help when they're around
      • SuperPhly joined the channel
      • uzyn joined the channel
      • cemerick joined the channel
      • kaotisk has quit
      • cubistbuddhist[m joined the channel
      • SuperPhly has quit
      • Mateon1 has quit
      • Guest68719 has quit
      • pp7y7 joined the channel
      • pp99s has quit
      • ivo_ joined the channel
      • cemerick has quit
      • HostFat_ joined the channel
      • HostFat__ has quit
      • dmr joined the channel
      • fleeky__
        jbenet: thank you for the help anyways , led me to some interesting searches
      • jbenet: i guess the feature / bug result of this is that i will end up using the cli more ;)
      • UnderSampled joined the channel
      • UnderSampled
        hello
      • I'm not quite sure, what does "ipfs files" run on?
      • deltab
        hi
      • what do you mean by "run on"?
      • soloojos has quit
      • Mateon2 joined the channel
      • UnderSampled: are you referring to the command line subcommand 'files'?
      • UnderSampled
        deltab: yes
      • deltab
        what do you mean by 'run'?
      • soloojos joined the channel
      • briarrose has quit
      • briarrose joined the channel
      • briarrose has quit
      • briarrose joined the channel
      • uzyn has quit
      • reit joined the channel
      • herzmeister has quit
      • herzmeister joined the channel
      • anewuser has quit
      • UnderSampled has quit
      • mguentner has quit
      • mguentner joined the channel
      • mguentner2 joined the channel
      • flyingzumwalt has quit
      • mguentner has quit
      • whyrusleeping
        anyone remember the term for when people ask for a solution to a problem that isnt their original problem?
      • like, someone trying to install X thinks they need to do Y, so they go and ask how to do Y without mentioning X?
      • Oatmeal joined the channel
      • slothbag has quit
      • whyrusleeping has quit
      • whyrusleeping joined the channel
      • lidel has quit
      • lidel joined the channel
      • robattila256 has quit
      • galois_dmz has quit
      • jn__
        whyrusleeping: i think that's called the "X-Y problem"
      • whyrusleeping
        jn__: its so obvious i missed it, thanks
      • wallacoloo___ joined the channel
      • chris613 has quit
      • mib_kd743naq joined the channel
      • mib_kd743naq
        hi
      • I am trying to find some examples of the MetaData sub-structure: https://github.com/ipfs/go-ipfs/blob/master/uni...
      • the reason I am looking into this is that I would like to somehow add mtime's to my trees
      • sourcediving indicates this kind of metadata has been entirely omitted :/
      • ( both in unixfs and in the ipld sketch of it... )
      • deltab
        I guess because it's not inherent to the content: two people could add the same file contents and get different hashes
      • mib_kd743naq
        it is not inherent to the content I absolutely agree
      • deltab
        git likewise doesn't track any metadata other than name and executable/regular/directory/symlink
      • mib_kd743naq
        but it is inherent to a listing ( i.e. a directory/link node )
      • deltab: precisely - git *does* track the creation-time in a higher-level structure ( the commit node )
      • https://www.kernel.org/pub/linux/kernel/v4.x/ <---- this is the experience I am after essentially
      • deltab: re "add the same file and get different hashes": this is already the case by design - chunking/linking choices influence the outcome
      • already today we have:
      • deltab
        creation of the commit, not the files; but you could add a file of timestamps
      • mib_kd743naq
        --chunker rabin + --raw-leaves + --trickle-dag
      • deltab
        sure, but it's something to be avoided where possible
      • mib_kd743naq
        I understand your point, but I disagree. For example the project I am embarking on will be placing stuff on ipfs that go-ipfs itself is unlikely to ever be able to produce
      • pfrazee has quit
      • i.e. "domain-knowledge-based chunking"
      • anyway - back to... hm... I guess I should just open a note section
      • yale_ is now known as yale
      • deltab
        agreed there can be uses where it'd make sense to add more metadata — blockchains, for instance
      • kvda_ has quit
      • jchevalay has quit
      • btw I've used filesystems that didn't always have timestamps: DFS and ADFS
      • timestamps and file types were eventually added, by repurposing the load and exec addresses
      • so you could have a mix of files within a directory: some untyped with load and exec addresses, others typed and timestamped
      • meems joined the channel
      • meems
        greetings
      • meme machine
      • mmeme maachine
      • HenryNguyen5 joined the channel
      • aaaaaaaaa____
        any advice for how to overcome "blockservice is closed" on ipfs add (v0.4.4)? https://www.irccloud.com/pastebin/UCjuZUgt/
      • Oatmeal has quit
      • SuperPhly joined the channel
      • lkcl joined the channel
      • mib_kd743naq
      • galois_dmz joined the channel