<lgierth>
don't have any better ideas at the moment
<shangteus>
lgierth: trying....
<lgierth>
but the fact that it can't even find addresses that it could try connecting to, is very weird, and hints at issues which are not NAT
<jkh>
shangteus: I guess we could always change our local port to pull from ipfs master, like @lgierth says
<jkh>
shangteus: sure couldn’t *hurt* at this point. :D
<shangteus>
jkh: that is some idea.. latest version doesn't look *stable* for us anyway :)
<jkh>
lgierth: I really do think the OS X ipfs is horkle at this point - it’s acting very strange overall. Takes the daemon about 5 minutes to start up, for example. “Initializing daemon…<5 minute wait> Swarm listening on /ip4/10.250.1.18/tcp/4001 … (more swarm things here) Daemon is ready”
<jkh>
lgierth: and then doing an ipfs add on the same OS X host takes another 5 minutes
<jkh>
lgierth: clearly something is banging its head, hanging, and then eventually succeeding after some log-ass timeout or some such
<jkh>
lgierth: (well, OK, success optional)
<jkh>
lgierth: I suppose I could try to build ipfs for OS X using the same github repo...
<lgierth>
oh wow that's fucked up
<lgierth>
we're going to ship v0.4.3-rc1 this week
<jkh>
lgierth: Let me try building from master. Like I said to @shangteus - can’t hurt at this point!
<lgierth>
could you file an issue about these 5 minute hickups, and that issue about a lack of addresses? to make sure it doesn't get lost
<jkh>
lgierth: but 0.4.2 is pain
<lgierth>
(if master doesn't help)
<jkh>
lgierth: will do!
<jkh>
lgierth: thanks for all the help regardless - it does give @shangteus and I a bit more hope. :)
<lgierth>
youre very welcome -- i'm happy to have you explore ipfs
<lgierth>
i
<lgierth>
#m going to bed now, tomorrow's another day full of meetings :) good night!
<shangteus>
lgierth: thanks for help
<jkh>
lgierth: good night!
<shangteus>
lgierth: good night
<dyce_>
can you do a 1 way git clone from ipfs address?
<jkh>
17:16:47.393 ERROR core/serve: Path Resolve error: Failed to get block for QmU3o9bvfenhTKhxUakbYrLDnZU7HezAVxPM6Ehjw9Xjqy: context canceled gateway_handler.go:510
<jkh>
fascinating...
<kpcyrd>
dyce_: do the regular steps to publish a repo over http, then `ipfs add -r` and then just `git clone http://localhost:8080/ipfs/Qmasdf`
ashark has joined #ipfs
ashark has quit [Ping timeout: 240 seconds]
cemerick has joined #ipfs
M-matthew is now known as Matthew[m]
stoogenmeyer has joined #ipfs
r04r is now known as zz_r04r
alhof_ has quit [Quit: Be back later ...]
alhof_ has joined #ipfs
rgrinberg has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
stoogenmeyer has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
<dyce_>
kpcyrd: very cool!
stoogenmeyer has joined #ipfs
alhof_ has quit [Ping timeout: 250 seconds]
palkeo has quit [Quit: Konversation terminated!]
JesseW has joined #ipfs
herzmeister has quit [Quit: Leaving]
herzmeister has joined #ipfs
skoocda has quit [Ping timeout: 244 seconds]
gavin has quit [Ping timeout: 264 seconds]
cemerick has quit [Ping timeout: 244 seconds]
<ipfsbot>
[js-ipfs-api] taravancil opened pull request #321: Add example of using name in the browser (master...name-example) https://git.io/vKwZN
computerfreak has quit [Remote host closed the connection]
skoocda has joined #ipfs
wallacoloo has joined #ipfs
ashark has joined #ipfs
ashark has quit [Ping timeout: 250 seconds]
disgusting_wall has joined #ipfs
joeyh has joined #ipfs
jaboja has quit [Remote host closed the connection]
matoro has joined #ipfs
AtnNn has quit [Ping timeout: 240 seconds]
AtnNn has joined #ipfs
<jkh>
- fetching Qmaau1d1WjnQdTYfRYfFVsCS97cgD8ATyrKuNoEfexL7JZ via ipfs api
<jkh>
[1 / 2] fetched dep: go-crypto
<jkh>
fatal error: all goroutines are asleep - deadlock!
<jkh>
… *sigh* ...
<jkh>
Current status with building ipfs master on OS X: ipfs 1, jkh 0
AtnNn has quit [Ping timeout: 240 seconds]
ashark has joined #ipfs
ashark has quit [Ping timeout: 272 seconds]
AtnNn has joined #ipfs
Stebalien has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
ylp has quit [Ping timeout: 264 seconds]
ylp has joined #ipfs
apiarian has quit [Ping timeout: 250 seconds]
apiarian has joined #ipfs
stoogenmeyer has joined #ipfs
mgue has quit [Quit: WeeChat 1.5]
stoogenmeyer has quit [Ping timeout: 264 seconds]
joeyh has quit [Ping timeout: 260 seconds]
mgue has joined #ipfs
rgrinberg has quit [Ping timeout: 258 seconds]
Stebalien has quit [Ping timeout: 250 seconds]
sol56 has quit [Ping timeout: 244 seconds]
sol56 has joined #ipfs
espadrine_ has joined #ipfs
stoogenmeyer has joined #ipfs
ashark has joined #ipfs
ygrek has quit [Ping timeout: 272 seconds]
ashark has quit [Ping timeout: 258 seconds]
sol56 has quit [Ping timeout: 276 seconds]
sol56 has joined #ipfs
treaki__ has joined #ipfs
treaki_ has quit [Ping timeout: 272 seconds]
skoocda has quit [Ping timeout: 272 seconds]
stoogenmeyer has quit [Remote host closed the connection]
<JesseW>
It now has nearly 100% test coverage, and seems generally complete...
<Guest74378>
question about the ipfs dht command - my daemon is running, and I have peers
<Guest74378>
but my name resolve with no id does not resolve
<Guest74378>
nor does the ipfs dht findpeer <myid> return anything
<JesseW>
Guest74378: what do you mean by "my name resolve with no id"?
<Guest74378>
as i understand it, if I call ipfs name resolve with no arg, it should resolve the latest pub under my node id
<Guest74378>
I am getting mdns lookup errors in the daemon's output
<JesseW>
Hm. So when you run `ipfs name resolve` you get no output?
<Guest74378>
The error is... Error: Could not resolve name.
<JesseW>
hm. has it worked in the past?
<JesseW>
what does `ipfs id` say?
<Guest74378>
i get the standard output with the ID, pubkey, addresses, agent/protocol versions
<Guest74378>
and yep - it was working last week; been developing with the go daemon and the js-ipfs-api
<JesseW>
What about `ipfs name resolve ipfs.io`
<JesseW>
which gives me /ipfs/QmTzQ1JRkWErjk39mryYw2WVaphAZNAREyMchXzYQ7c15n
<Guest74378>
that resolves for me
<JesseW>
and --debug doesn't seem to provide much useful info
sol56 has quit [Ping timeout: 252 seconds]
<Guest74378>
this is the last debug output before the error - nothing jumps out to me: 21:18:05.907 DEBUG cmd/ipfs: Executing command via API main.go:328
<JesseW>
yeah, it didn't seem to be very helpful when I ran it :-(
<Guest74378>
hmm.....
<JesseW>
We've about reached the current limits of my knowledge, sadly.
<JesseW>
hopefully someone more knowledgable will chime in...
<Guest74378>
that's ok! Thanks for the help
<JesseW>
sure. sorry it didn't actually work. :-/
<Guest74378>
been dealing with this for a few hours
<Guest74378>
wondering if I might need to reinstall - that's a crappy thought!
<Guest74378>
but running low on options
devbug has joined #ipfs
<JesseW>
I doubt reinstalling is needed...
ylp has quit [Ping timeout: 264 seconds]
ylp has joined #ipfs
stoogenmeyer has joined #ipfs
stoogenmeyer has quit [Ping timeout: 250 seconds]
<ipfsbot>
[js-ipfs-api] taravancil opened pull request #322: Add tests for add (master...add-tests) https://git.io/vKwga
<JesseW>
Guest74378: I presume you've poked around with `ipfs log`
devbug has quit [Quit: ZZZzzz…]
sol56 has joined #ipfs
devbug has joined #ipfs
Guest74378 has quit [Ping timeout: 250 seconds]
sol56 has quit [Ping timeout: 252 seconds]
* JesseW
is grumpy that apparently the list of loggers *DOESN'T* fully match the names displayed by the log messages...
<JesseW>
for example, "swarm2" and "nat"
<JesseW>
aren't loggers, but are displayed as though they were...
<ipfsbot>
[go-ipfs] RichardLitt pushed 1 new commit to feat/config-docs: https://git.io/vKr3L
<ipfsbot>
go-ipfs/feat/config-docs 7866ede Richard Littauer: Capitalized acronyms, add ToC...
ruby32 has joined #ipfs
apiarian has quit [Ping timeout: 264 seconds]
j12t has joined #ipfs
apiarian has joined #ipfs
<ipfsbot>
[go-ipfs] Kubuxu force-pushed feat/config-docs from 7866ede to ca0c689: https://git.io/vK2ew
<ipfsbot>
go-ipfs/feat/config-docs 4307c10 Jeromy: document the config options...
<ipfsbot>
go-ipfs/feat/config-docs d7a1bda Jeromy: add newer config options and reformat...
<ipfsbot>
go-ipfs/feat/config-docs ca0c689 Richard Littauer: Capitalized acronyms, add ToC...
j12t has quit [Ping timeout: 250 seconds]
ruby32d has joined #ipfs
espadrine has quit [Ping timeout: 260 seconds]
ruby32 has quit [Ping timeout: 272 seconds]
ylp has quit [Ping timeout: 264 seconds]
ylp has joined #ipfs
mildred has quit [Ping timeout: 240 seconds]
rgrinberg has joined #ipfs
ashark has joined #ipfs
apiarian has quit [Ping timeout: 260 seconds]
apiarian has joined #ipfs
ashark has quit [Ping timeout: 276 seconds]
computerfreak has quit [Quit: Leaving.]
rgrinberg has quit [Quit: WeeChat 1.5]
rgrinberg has joined #ipfs
cow_2001 has quit [Quit: Praise Kük!]
stoogenmeyer has joined #ipfs
stoogenm_ has joined #ipfs
stoogenmeyer has quit [Ping timeout: 260 seconds]
<achin>
richardlitt: ping
<richardlitt>
hey man
<richardlitt>
My message make sense?
<achin>
nope :)
<richardlitt>
Basically, I've totally failed to make the newsletter recently
<richardlitt>
in github.com/ipfs/newsletter/issues/44
<richardlitt>
You have some code in the python file, some commits about updating it
<richardlitt>
I am going to close that PR -can you pull out your code and put it into a new PR?
<achin>
does "close this newletter" mean publish it? or toss it out?
<richardlitt>
Toss it. It's now so late it needs to be significantly updated
<richardlitt>
100% my bad.
<achin>
because if you're going to toss it out and start a new one, then just close the PR, right?
<achin>
i can for sure reapply my commits to the new PR, but i don't see the point in removing my commits from the current PR if its just going to be discarded
<richardlitt>
hmm
<richardlitt>
So, I don't know when I will start a new PR
<richardlitt>
And I was hoping to offload having to think about your commits to the next one
<richardlitt>
I don't mean removing them from the current PR, just... make sure that they are in the main code base
<richardlitt>
So... putting them in a new PR, I guess?
<achin>
ahh, i see. ok, so my commits that updated the newsletter, i'll just ignore those
<achin>
but my commits that update the tools, i'll make sure that gets into the master branch
<richardlitt>
That was not the most coherent response. Let me try again: I am super grateful for your work on the code, and I think that that should be captured. I want to make sure it is. Unfortunately, it is tied into a PR that is not going to be merged. It actually shouldn't be in that PR, anyway, because a PR should be about ONE thing. So, can you copy over the
<richardlitt>
commits to a new PR against master, so that I can merge just your code? I think that would be the best move. That way, I can close this inactive PR, and when i start a new one, your commits are already there.
<achin>
yep, makes sense
<richardlitt>
Exactly.
<richardlitt>
Awesome!
<richardlitt>
Thanks.
espadrine has joined #ipfs
ashark has joined #ipfs
PrinceOfPeeves has joined #ipfs
espadrine has quit [Ping timeout: 244 seconds]
espadrine_ has joined #ipfs
ashark has quit [Ping timeout: 260 seconds]
stoogenm_ has quit [Remote host closed the connection]
alhof_ has joined #ipfs
espadrine has joined #ipfs
espadrine_ has quit [Ping timeout: 240 seconds]
<victorbjelkholm>
daviddias, what is "require('lx')" all about? Empty package
<daviddias>
ahah
<daviddias>
it is our JavaScript and Node.js meetup
stoogenmeyer has quit [Remote host closed the connection]
stevenroose has joined #ipfs
<ipfsbot>
[go-ipfs] whyrusleeping created fix/files-cp (+1 new commit): https://git.io/vKr6a
<ipfsbot>
go-ipfs/fix/files-cp 660a4a9 Jeromy: mfs: fix copying into directory with no given filename...
zorglub27 has joined #ipfs
<ipfsbot>
[go-ipfs] whyrusleeping opened pull request #2977: mfs: fix copying into directory with no given filename (master...fix/files-cp) https://git.io/vKriI
j12t has joined #ipfs
Guest67544 has joined #ipfs
espadrine has joined #ipfs
espadrine_ has quit [Remote host closed the connection]
taw00 has quit [Ping timeout: 240 seconds]
cryptix has joined #ipfs
cryptix has quit [Client Quit]
j12t has quit [Ping timeout: 240 seconds]
Guest67544 is now known as weex_
stoogenmeyer has joined #ipfs
espadrine_ has joined #ipfs
espadrine has quit [Ping timeout: 240 seconds]
zorglub27 has quit [Ping timeout: 252 seconds]
espadrine_ has quit [Ping timeout: 260 seconds]
<ipfsbot>
[go-ipfs] whyrusleeping created deps/update-gx (+1 new commit): https://git.io/vKrMD
<ipfsbot>
[go-ipfs] kpcyrd opened pull request #2980: Increase 512kbytes object put limit to 2mbytes (master...feature/2mb_object_put) https://git.io/vKoer
gmcquillan has joined #ipfs
gmcquillan is now known as gavin
gavin has quit [Read error: Connection reset by peer]
<whyrusleeping>
kpcyrd: sorry, i just *had* to find something wrong with that PR
<whyrusleeping>
couldnt make it that easy on you
<whyrusleeping>
otherwise i would have to give away the "worlds easiest contribution" award
<Kubuxu>
"there should be test case for that" :P
Encrypt has quit [Quit: Quitte]
Encrypt has joined #ipfs
<lgierth>
always leave the campsite better than you found it
s_kunk has quit [Ping timeout: 260 seconds]
j12t has joined #ipfs
cemerick has joined #ipfs
gmcquillan has joined #ipfs
palkeo has joined #ipfs
palkeo has quit [Changing host]
palkeo has joined #ipfs
gmcquillan is now known as gavin
cemerick has quit [Remote host closed the connection]
stoogenmeyer has joined #ipfs
skoocda has quit [Ping timeout: 276 seconds]
<jkh>
whyrusleeping: oh hey, there you are!
<jkh>
whyrusleeping: you going to be around a little later?
<whyrusleeping>
jkh: here i am!
<whyrusleeping>
yeah, i should be
<whyrusleeping>
i'm in UTC+1 at the moment
<jkh>
whyrusleeping: we (over at iXsystems) are wrestling with various build issues on master with OS X and I thought I’d at least document and then send them to.. you? The issues are different between 10.11.6 and 10.11.7, just to make things more fun
<whyrusleeping>
oh joy
<whyrusleeping>
please do send any and all info you have to me
<jkh>
whyrusleeping: Not sure if some of them are Go related either - still trying to binary search my way through all of the build failure output
<jkh>
whyrusleeping: I did notice that bin/check_go_path would always fail for *any* value of $GOPATH - reading that script, it’s not clear how it could work at all
<jkh>
whyrusleeping: but making it always exit 0 enabled builds on OS X of any flavor to proceed. :)
<whyrusleeping>
jkh: hrm... thats not good
<whyrusleeping>
we did change that recently
_marvin_ has joined #ipfs
* whyrusleeping
sighs
<jkh>
whyrusleeping: I was going to ask if you had a reference platform / configuration you used to test the various OS combinations, and if that was generating success for you with the current master
<whyrusleeping>
we didnt test our build system with freeBSD
<whyrusleeping>
we just cross compiled for testing
<jkh>
whyrusleeping: well, with respect to the check_go_path script, given what I saw of the calling conventions for it on OS X, it couldn’t work unless:
<jkh>
-while [ ${#} -gt 1 ]; do
<jkh>
- if [ "$PWD" = "$2" ]; then
<jkh>
+while [ ${#} -gt 0 ]; do
<jkh>
+ if [ "$PWD" = "$1" ]; then
<jkh>
exit 0
<jkh>
which I did on 10.11.6 and it was happy again
stoogenmeyer has quit [Remote host closed the connection]
<whyrusleeping>
>.>
<whyrusleeping>
it works on osx for our CI systems
<jkh>
whyrusleeping: we have an incredibly diverse set of OSes here at work running in dozens of VMs - I could easily set up builds on each, just to see if and how they fall consistently. :)
<whyrusleeping>
let me check what they run
<whyrusleeping>
that would be realllly cool
<whyrusleeping>
we don't have that infrastructure
stoogenmeyer has joined #ipfs
<jkh>
whyrusleeping: I’ll set that up today..
<jkh>
whyrusleeping: We’re really trying to incorporate ipfs into FreeNAS so that we can arbitrarily publish files / VMs / containers directly from the platform, but it (ipfs) is really fighting us. @shangteus was going over some of the woes he’s having here yesterday.
Anselmo has quit [Quit: WeeChat 1.0.1]
<jkh>
whyrusleeping: We just don’t understand why it seems to hang / misbehave from machine to machine, even on the same network. @lgierth saw some of the hashes that @shangteus posted yesterday and thought “weird!"
* whyrusleeping
scrolls back through irc history
<Kubuxu>
jkh: which version of IPFS are you using?
<jkh>
Kubuxu: 0.4.2
<Kubuxu>
we are finalizing RC on days, 0.4.2 has quite some mileage on it and is being problematic (and has few not so nice bugs).
cemerick has joined #ipfs
<Kubuxu>
yeah, it might be that. There was bug in stream multiplexing dependency in that build, we resolved it with the author.
<jkh>
Kubuxu: I thought I’d build from master yesterday and see if I could reproduce the 30 second hangs on *any* operation in OS X, but like I was saying to @whyrusleeping, even that has been hard
<jkh>
It probably doesn’t help that because of my history, I also run early OS X betas. :-/
<jkh>
I can’t even get macOS Sierra to load all of the deps
<whyrusleeping>
jkh: judging by a quick reading over of the issue lgierth you and shangteus talked about yesterday, It looks like NAT traversal issues
<Kubuxu>
that is interesting, some of our devs are using OS X too.
<whyrusleeping>
jkh: we're deciding tomorrow what we're going to focus on for the next few months
<lgierth>
whyrusleeping: even with `ipfs swarm addrs | grep <peerid-from-findprovs>` being empty?
<whyrusleeping>
we're *definitely* going to prioritize NAT traversal higher
stoogenmeyer has quit [Remote host closed the connection]
<jkh>
Kubuxu: It would be useful to see what they set GOPATH to
<jkh>
Kubuxu: I am following the Google recommendations and setting it to $HOME/work
j12t has quit [Ping timeout: 260 seconds]
<Kubuxu>
whyrusleeping: are you sure it is NAT, we used to think the same and 0.4.2 it was smux but it is swarm0 then it isn't smux.
<jkh>
Kubuxu: But like I was saying to @whyrusleeping earlier, when I do that, the check_go_path script always vomits, regardless of where I try to put the ipfs-go build directory
<jkh>
Kubuxu: and if I override that or change it to check its arguments a little differently, the “install” rule in the Makefile blows up with:
<jkh>
go install -ldflags="-X "github.com/ipfs/go-ipfs/repo/config".CurrentCommit=0e4c195"
<jkh>
init.go:11:2: cannot find package "github.com/ipfs/go-ipfs/assets" in any of:
<jkh>
Kubuxu: so, I’m really not sure how this is supposed to be built. There’s not really a guide that covers how ipfs interacts with Go’s “interesting” build dependency system
<Kubuxu>
can you echo your GOPATH? and pwd?
<whyrusleeping>
(pwd of go-ipfs git repo)
<jkh>
jkh@katana-> echo $GOPATH
<jkh>
jkh@katana-> pwd
<jkh>
whoops, thank you IRC
<jkh>
sorry, it stripped the paths because they start with / :) One sec.
<jkh>
I will readily admit that I write everything in C and Ruby so this Go beast is a new experience
<jkh>
whyrusleeping: DOH!
<jkh>
Now I feel sheepish
<whyrusleeping>
lol, dont worry
<whyrusleeping>
that just means we need clearer docs
<whyrusleeping>
people expect 'make' to just work
<jkh>
whyrusleeping: are those “go get” instructions still relevant?
<whyrusleeping>
yeah, 'go get -d' just downloads things to the right place
<whyrusleeping>
it might error, but thats expected
<whyrusleeping>
(temporary hack until we get gx more integrated into go)
<jkh>
whyrusleeping: So, if you just do the git clone straight from github, is this functionally indistinguishable or does “go get” do more things in addition to just fetching the bits?
<whyrusleeping>
functionally the same
<jkh>
whyrusleeping: I mean, git clone into src/github.com/ipfs
<jkh>
OK, coo
ygrek has quit [Ping timeout: 252 seconds]
<jkh>
whyrusleeping: OK, MUCH happier. Thanks, and sorry again for derpiness
<jkh>
So that at least puts OS X 10.11.6 on the “happy build” list
<jkh>
let’s see if it’s also on the happy runs list
reit has quit [Ping timeout: 260 seconds]
<jkh>
jkh@katana-> ipfs version
<jkh>
ipfs version 0.4.3-dev
<jkh>
so far so good..
<Kubuxu>
:D
<whyrusleeping>
its gonna want you to run a migration if you start it over an existing older repo
<lgierth>
whyrusleeping: good opportunity to merge that auto-mgiration code
<whyrusleeping>
yeah... true
<whyrusleeping>
wanna give that one final look over lgierth and Kubuxu ?
<whyrusleeping>
i hardcoded the dists path
<Kubuxu>
will do that right now.
<jkh>
whyrusleeping: I saw that - I didn’t have anything interesting in the repo so I just got out my axe and deleted it, then re-initted. Sorry if you wanted to use that as a test case for migration. :-/
<whyrusleeping>
lgierth: i definitely did that at one point...
<whyrusleeping>
wtf
<lgierth>
the --migrate option is taken care of <3, but the prompt should probably still be in the command, not deep down in fsrepo
<jkh>
whyrusleeping: OK, that’s a LOT better!
<jkh>
whyrusleeping: The 30 second hangs have gone away, and I am able to access files published from behind our NAT
<whyrusleeping>
whoa
<Kubuxu>
whyrusleeping: told you!!
<lgierth>
haha nice
<whyrusleeping>
well thats nice to hear :D
<shangteus>
hello guys
<whyrusleeping>
shangteus: heyo!
<lgierth>
hey shangteus
<Kubuxu>
\o
<whyrusleeping>
sorry for not being around yesterday
<lgierth>
he was playing werewolf
<shangteus>
jkh: then I'm going to try to use 0.4.3-dev in FreeNAS build
<shangteus>
I was trying to build master from forked repo but that actually was a bit more tricky than I expected :)
<shangteus>
personally I haven't tried IPFS on Mac, only on FreeNAS on version 0.4.2
<Kubuxu>
do it, switch to 0.4.3-rcN and 0.4.3 when they are available.
<shangteus>
and I was able to use hashes pointing directly to files
<shangteus>
NAT did not matter
<whyrusleeping>
rcs should be ready within 24 hours
<jkh>
well, I can certainly say that I am behind a NAT here at home, and our FreeNAS test machine is also behind a NAT
<jkh>
and I can publish in both directions successfully now using the newer bits
<shangteus>
but sharing directories led to disasters
<lgierth>
whyrusleeping: yeeeah... i wanna be asocial this evening and get through this changelog
<shangteus>
I was able to `ipfs add` them
<shangteus>
but `ipfs get` only within the same network
<shangteus>
NAT pretty much is killing the thing on FreeNAS on IPFS v0.4.2
<jkh>
shangteus: Let’s try to build the newer thing
<shangteus>
yep
<jkh>
shangteus: i’m now happily using it in both directions for both files and dirs, and I’m even double NAT’d at home
<shangteus>
I'm going to try that now and let you guys know soon
<shangteus>
thanks for your support :) I hope this 0.4.3 is going to solve our problems
<jkh>
shangteus: NAT to wifi going to a point-to-point microwave radio dish going to another NAT going to a UHF radio going 15 miles to another UHF radio hooked to a comcast modem.. Hmmm. TRIPLE NAT actually!
<shangteus>
:)
<jkh>
Wow how does this even work
j12t has quit [Remote host closed the connection]
<lgierth>
jkh: oh hey was it using /udp/utp or /tcp in 0.4.2?
<shangteus>
they've put some magic inside the bits
<shangteus>
trust me
<lgierth>
because /udp/utp was really buggy and we disabled it for now
<shangteus>
it's the only way
<jkh>
lgierth: I’m pretty sure TCP but I blew that binary away when I reinstalled the -dev image on top of it; I could check
<jkh>
lgierth: my macOS Sierra box at work has gone to sleep and I’m trying to wake it up
<lgierth>
don't worry about it
<jkh>
I need to drive in anyway; I’ll be on the work network in an hour
<jkh>
and because our IT department is certifiably insane, that should be an even better test network
<jkh>
makes my double radio to comcast network look normal
* jkh
lives way up in the mountains
<whyrusleeping>
just having utp enabled would break things
<whyrusleeping>
:/
<jkh>
shangteus: What’s your luck been in making your own ipfs-go port that uses the master bits?
<jkh>
shangteus: theoretically not different than our current port, just a change in where it gets the bits from, no?
<lgierth>
whyrusleeping: i'm still wondering why `ipfs swarm addrs | grep <peerid-from-findprovs>` turned out empty
<jkh>
shangteus: I mean, the later version you’ve been using has the gx refactoring and such
<whyrusleeping>
no connection?
<lgierth>
whyrusleeping: the object they published *did* show up
<whyrusleeping>
lgierth: then 'ipfs swarm peers | grep peerID'
<lgierth>
and findprovs came up with exactly their two peerids
<jkh>
shangteus: BTW, you should show these guys how the ipfs publishing of VMs in FreeNAS 10 is meant to work
<shangteus>
sorry, I got into porting to 0.4.3-dev for a moment :)
<jkh>
whyrusleeping: That is interesting! Does the FUSE filesystem for ipfs also consume the same API, so you get the same “cheap copies” effect for file copies?
<shangteus>
jkh: we'd have to rework FreeBSD port a bit
<jkh>
whyrusleeping: I mean, for transparent file copies, using cp(1) and such “natively"
<whyrusleeping>
jkh: the fuse mountpoint uses the same code underneath, but doesnt get the same benefit
<lgierth>
jlstack: you mean whether they exist in the core api?
mildred has joined #ipfs
<shangteus>
jkh: right now it's using releases in tar.gz to build itself
<whyrusleeping>
since the fuse interface doesnt pass us the reflink stuff AFAIK
<jlstack>
Yes
<lgierth>
jlstack: i'm not sure we'll add these to the core api
<lgierth>
jlstack: do you have a use case?
<jlstack>
No, just trying to fulfill the requirements for the python client
<lgierth>
jlstack: don't worry about it if it's for the sake of completeness
<shangteus>
jkh: we'd have to do manual checkout, but I don't know if it's worth it if they are doing RC soon anyway
<jkh>
whyrusleeping: ah. :(
<whyrusleeping>
jkh: but! the on disk storage will be the same as the files api stuff
<shangteus>
jkh: and we are going to be able to use release again
<whyrusleeping>
the only downside is that we have to rehash everything :/
<lgierth>
jlstack: we probably wanna move away from exposing the node config over the api, since it's very implementation specific
<jlstack>
There also exists node_start and node_stop that do not appear to be implemented in the other clients. Should these be implemented? Does an endpoint exist for the rest API?
<lgierth>
jlstack: and it's also pretty sketchy from a security standpoint
<Kubuxu>
jkh: FUSE doesn't have those properties, FUSE doesn't expose API for that, I would love if it did.
<lgierth>
jlstack: node_start and node_stop don't exist yet at all - don't worry about it for now
<shangteus>
okay... about VM sharing... right now we have a single JSON file, so called 'template' storing VM configuration and links for fetching VM images
<jlstack>
lgierth: Could the specification be revised to better reflect what actually is required of an implementation?
<lgierth>
jlstack: wanna make a PR or an issue for it? :D
<lgierth>
i'll be back in 5
<shangteus>
our filesystems are ZFS and to share a VM we're basically snapshotting ZFS datasets storing VM bits
<shangteus>
cloning them
<jlstack>
lgierth: I will
<shangteus>
storing them in separate cache directory
<shangteus>
and then there's a template created from current VM configuration with references to IPFS hashes referencing cache directory
<shangteus>
storing freezed VM images
<shangteus>
when we have ready json from which all other links/configuration could be read
<shangteus>
it's also being published via IPFS
<shangteus>
from other user's perspective it's just pasting IPFS hash into our CLI when creating a new VM
<shangteus>
FreeNAS recognizes input to be IPFS hash, fetches it
<shangteus>
and then using data from JSON it's starting to create a new VM
<shangteus>
when all other files/images
<shangteus>
are referenced as IPFS hashes also
espadrine_ has joined #ipfs
<shangteus>
that's pretty much it..
espadrine has quit [Ping timeout: 272 seconds]
jager has joined #ipfs
<jkh>
Also, @jakub wants copy-on-write in IPFS. That is all. :D
j12t has joined #ipfs
jlstack has quit [Ping timeout: 250 seconds]
jaboja has joined #ipfs
<Kubuxu>
jkh: depends what you define with CoW in IPFS
<Kubuxu>
if you add one file and and second one that is similar, blocks will be reused
bsm117532 has quit [Remote host closed the connection]
j12t_ has joined #ipfs
bsm117532 has joined #ipfs
j12t has quit [Ping timeout: 240 seconds]
espadrine_ has quit [Ping timeout: 252 seconds]
Stebalien has joined #ipfs
rendar has joined #ipfs
<jkh>
Kubuxu: I am thinking @jakub probably meant file references which became copies only once things changed
<jkh>
Kubuxu: you know, the holy grail. :)
<Kubuxu>
We are thinking about that too.
zorglub27 has quit [Ping timeout: 264 seconds]
<shangteus>
Kubuxu: that would be cool
<Kubuxu>
There is WIP branch for that but we have to re-implement it.
<Kubuxu>
if someone of you could link me a reflink copy (0cost copy) in C for ZFS it would be great (I know how to do in in btrfs but no idea about ZFS).
<Kubuxu>
I will bookmark it for later.
ylp has quit [Ping timeout: 264 seconds]
Encrypt has quit [Quit: Quitte]
<jkh>
Kubuxu: Hmm? ZFS already works that way natively due to the way the block allocator works
<jkh>
Kubuxu: But maybe you’re referring to something very specific that I’m not getting
<Kubuxu>
oh, so will just `cp fileA fileB` be 0 cost?
<jkh>
Kubuxu: Ah, no, it’s not that granular. It works on the dataset or volume (which is a specialized kind of block device) level. If you snapshot or clone, it’s zero cost
<jkh>
Kubuxu: Though if you enable dedup on the dataset in question, then file copies become references
<jkh>
Kubuxu: so, I suppose it’s possible to say that *with* dedup on, yes, per-file copies are zero cost
<jkh>
Kubuxu: But the DDT (dedup table) has its own cost, so turning on dedup isn’t something you want to do by default
<Kubuxu>
yeah, that is what I am asking for, with btrfs you open two files and run ioctl over those two fd`s which does 0 cost clone of a file.
mildred has quit [Quit: Leaving.]
mildred1 has joined #ipfs
<jkh>
Kubuxu: Yeah, no, ZFS doesn’t work like that - it’s entirely an implementation detail how the copy is stored.
<jkh>
Kubuxu: Not visible at the higher layers of the system
<shangteus>
Kubuxu: native support in ZFS goes like this: create zfs dataset; snapshot it; clone snapshot
<jkh>
right
<shangteus>
Kubuxu: resulting dataset is going to be 0 cost copy of snapshot
<Kubuxu>
that is what I am asking for: https://github.com/zfsonlinux/zfs/issues/405 the ZFS way seems quite more complicated, can't guarantee that I will do it but whyru is using ZFS.
<shangteus>
Kubuxu: we use this mechanism in VM publishing. I have to create actual dataset from dataset's snapshot (clone) to fetch data from it
<shangteus>
Kubuxu: whoa.. that looks far more complicated than inbuilt ZFS mechanisms
<shangteus>
Kubuxu: I did not mean that usage would be hard :)
stoogenmeyer has quit [Remote host closed the connection]
Shibe has left #ipfs ["Leaving"]
stoogenmeyer has joined #ipfs
zorglub27 has joined #ipfs
<ckwaldon>
I'm trying to run the ipfs daemon on OSX, but I keep getting mdns lookup errors because of a failed binding to a udp port. Is this a known issue? I'm running 0.4.2
<Kubuxu>
It isn't critical subsystem.
<Kubuxu>
if the daemon working or not?
<Kubuxu>
lgierth: how is the CHANGELOG coming?
<Kubuxu>
we need to release that 0.4.3-rc1
<lgierth>
yeeeah.... this evening
<lgierth>
promise
<lgierth>
i've already covered like 15 PRs
<lgierth>
it's ~115 in total
<ckwaldon>
I guess the daemon is working... I did just succeed at pulling data from the network in one context. I thought the UDP port might be stopping that from going through.
<Kubuxu>
no it is mdns, it is used for local peer discovery
<lgierth>
might be related to fd limit?
<ckwaldon>
fd limit?
<Kubuxu>
lgierth: nope
<Kubuxu>
just 4343 might not be on REUSEPORT
<ckwaldon>
What do you mean?
<Kubuxu>
ckwaldon: just a impl detail, I don't use OSX so we will work something out.
<ckwaldon>
Okay. I'm getting some kinda inconsistent behavior. I was able to run an `ipfs block stat` on a hash, but `ipfs cat` and `ipfs get` both hang when I use them with the same hash...
<lgierth>
ckwaldon: what did stat say? it might have child blocks which can't be fetched (that would be the reason for cat and get hanging)
<lgierth>
or eeeh.... ipfs object stat <key>
<lgierth>
whyrusleeping: can blocks have links, or is that on the layer above? (i.e. on the object layer)
<whyrusleeping>
blocks cant have links
<whyrusleeping>
thats the object laye
<whyrusleeping>
r
<ckwaldon>
Object stat seems to hang as well
<ckwaldon>
The one time that block stat worked, I got back: 15:33:39.616 ERROR flatfs: too many open files, retrying in %dms0 flatfs.go:127 15:33:39.616 ERROR flatfs: too many open files, retrying in %dms100 flatfs.go:127 15:33:39.716 ERROR
Encrypt has joined #ipfs
<ckwaldon>
Sorry about that formatting... ugh. The hash I'm working with is QmYA2fn8cMbVWo4v95RwcwJVyQsNtnEwHerfWR8UNtEwoE
<lgierth>
oh yeah that's ulimit
<whyrusleeping>
ckwaldon: did you only see one of those messages?
<lgierth>
try latest master which bumps the ulimit automatically, or set ulimit -n 1024 in the same shell before running ipfs
<ckwaldon>
I saw all of those messages, not just one.
skoocda has joined #ipfs
<ckwaldon>
I would try the latest master, but I'm actually trying to test the py-ipfs-api library against my local node and it seems better to test against the released version than not.
<lgierth>
ckwaldon: we're about to release -rc1 hopefully tomorrow
<lgierth>
eeh 0.4.3-rc1
<ckwaldon>
I see. For the purposes of working on a client, would you recommend working against the release candidate or against the older, stable version?
mgue has quit [Ping timeout: 250 seconds]
<lgierth>
take the release candidate
<ckwaldon>
Okay, will do.
<lgierth>
and if it's blocking you, go with master :)
ckwaldon has quit [Ping timeout: 250 seconds]
ralphtheninja has quit [Quit: leaving]
zorglub27 has quit [Ping timeout: 260 seconds]
mgue has joined #ipfs
disgusting[m]1 has joined #ipfs
alhof_ has quit [Quit: Be back later ...]
alhof_ has joined #ipfs
ylp has quit [Ping timeout: 264 seconds]
ylp has joined #ipfs
bsm117532 has quit [Read error: Connection reset by peer]
bsm117532 has joined #ipfs
bsm117532 has quit [Client Quit]
alhof_ has quit [Ping timeout: 276 seconds]
bsm117532 has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
<alu>
If anyone is gunna be at HOPE/DEFCON I'd love to show ya how it all works. I have great visual documentation on my blog as well.
devbug has joined #ipfs
skoocda has joined #ipfs
mildred1 has quit [Ping timeout: 276 seconds]
dvim has quit [Quit: Connection closed for inactivity]
TheWhisper has joined #ipfs
ashark has quit [Ping timeout: 276 seconds]
rendar has quit [Quit: std::lower_bound + std::less_equal *works* with a vector without duplicates!]
<TheWhisper>
back
Oatmeal has quit [Ping timeout: 258 seconds]
<TheWhisper>
Wrong channel
<kholdstayre>
what do we have to do to start getting involved with ipfs
<jkh>
whyrusleeping: you still awake?
<jkh>
whyrusleeping: I reproduced that weird build problem with macOS Sierra BETA
<jkh>
whyrusleeping: Doing everything by the book this time and everything...
<jkh>
lgierth: OK, how about you? :)
mbauermeister has joined #ipfs
<mbauermeister>
https://www.youtube.com/watch?v=tARrqlPFpV4 it is a freeware it will unlock all the hidden features all DVD and blue players it will rip any disc and make an exact copy of it when you insert a blank disc in it
mbauermeister has left #ipfs [#ipfs]
stoogenmeyer has joined #ipfs
stoogenmeyer has quit [Remote host closed the connection]
stoogenmeyer has joined #ipfs
ruby32d has quit [Remote host closed the connection]
deltab_ has joined #ipfs
ygrek_ has joined #ipfs
kholdstayre has quit [Ping timeout: 252 seconds]
slothbag has joined #ipfs
__uguu__ has quit [Read error: Connection reset by peer]
ygrek has quit [Ping timeout: 260 seconds]
Encrypt has quit [Quit: Quitte]
deltab has quit [Quit: leaving]
deltab_ is now known as deltab
ruby32 has joined #ipfs
herzmeister has joined #ipfs
Akaibu has quit [Quit: Connection closed for inactivity]