#docker

/

      • myusuf3
        I am attaching to docker instance
      • locks up the machine
      • ideas?
      • backjlack
        myusuf3: This is a problem with docker doing something to the terminal, I've seen it happen both with the docker daemon and with the client.
      • reset usually fixes it.
      • myusuf3
        reset?
      • just shutdown the instance?
      • backjlack
        myusuf3: It looks like we're talking about different problems.
      • myusuf3: Does your whole box lock up or just the docker instance?
      • kencochrane joined the channel
      • myusuf3
        docker instance
      • when I attach I can't close it
      • backjlack
        Ok, I see.
      • I'm not sure what's supposed to happen in that case.
      • geku has quit
      • bfulton joined the channel
      • bfulton has quit
      • btubbs joined the channel
      • creack
        myusuf3: ctrl-c does not work?
      • myusuf3
        creack no it does not; I needed to open up new terminal and kill docker
      • creack
        docker deamon, client or standalne?
      • are you using the binaries or build from the sources?
      • zz_shykes is now known as shykes
      • kencochrane has quit
      • shykes
        sorry, ran out of battery :-/
      • sandfox has quit
      • joffrey has quit
      • myusuf3
        creack I am using the binaries
      • creack actually not sure which; kind of flying blind here due to lack of documentation (hint hint) I have an idea of what I am doing
      • joffrey joined the channel
      • Xeago has quit
      • shykes
        myusuf3: any specific requests for the docs? Is it just that the "getting started" part is not complete enough?
      • ToQoz joined the channel
      • ringi
        i'd love a doc about how to create a new image without going into interactive mode and typing in commands manually then going into another terminal and create the image
      • someth
      • something* like a simple and easy way for beginners to eg. setup a new image derived from base / busybox whatever with nginx, php and so on
      • that is easy to pack inside of a script
      • myusuf3
        yeah
      • i raised an issue ehre
      • destructuring
        always freaks me out: read tcp 127.0.0.1:4242: connection reset by peer
      • btubbs_ joined the channel
      • ringi
        are you on windows?
      • nvm, gotta go afk
      • destructuring
        on ubuntu
      • btubbs has quit
      • lucaspiller___ has quit
      • owen1 has quit
      • reeze joined the channel
      • belak joined the channel
      • sub joined the channel
      • sub has quit
      • sub joined the channel
      • ixx has quit
      • sub_ has quit
      • belak51 has quit
      • geetarista has quit
      • geetarista joined the channel
      • owen1 joined the channel
      • ixx joined the channel
      • ixx is now known as Guest62276
      • lucaspiller___ joined the channel
      • raspi2 joined the channel
      • lwf joined the channel
      • myusuf3- joined the channel
      • myusuf3 has quit
      • FUF has quit
      • FUF_ joined the channel
      • ferai joined the channel
      • joffrey_ joined the channel
      • joffrey has quit
      • tg has quit
      • lwf_ has quit
      • raspi1 has quit
      • jefferai has quit
      • titanous has quit
      • tg joined the channel
      • tg has quit
      • tg joined the channel
      • fss joined the channel
      • myusuf3- is now known as myusuf3
      • fss has quit
      • dominikh
        shykes: just pushed a squashed version of the lazy allocator changes
      • shykes
        dominikh: ok - will double-check but should be good to go
      • I like the ascii art in your tests :) Makes them very easy to parse. Good job.
      • dominikh
        shykes: sure thing. I changed the commit message to reflect the new design, and I renamed populate() to run(). rest is identical to before :)
      • hehe. said ascii art actually made me find a bug in the implementation :)
      • shykes
        even better
      • One thing I noticed - for each new IP you go over the entire address space from 0
      • dominikh
        I do not
      • I start one after the last IP, and try at most one entire round through the address space
      • shykes
        I guess it's this loop that confuses me: https://github.com/dotcloud/docker/pull/272/fil...
      • dominikh
        the k loop is "loop at most once through the entire address space". k itself isn't used though. the offset into the address space is i, declared on line 215
      • want me to rename `i` and `k` to `pos` and `attempt`?
      • shykes
        Judging by my confusion, I guess it would be better. Sorry for the last minute change
      • dominikh
        no problem. should've renamed those variables anyway
      • shykes
        why do you want to make it at most once the address space?
      • dominikh
        because if we loop through it once, and haven't found any free IP, there is no free IP.
      • shykes
        eg. why not "for inUse {"
      • dominikh
        if we do that, we end up with an infinite loop and the allocator locks up
      • shykes
        Ah, I see, that is in between each select
      • dominikh
        yup
      • shykes
        thought it was interleaved with it
      • dominikh
        1) try to get IP 2) send out IP/error or get back a released IP. 2 will always trigger 1 again
      • shykes
        Got it
      • dominikh
        pushed the variable rename
      • shykes
        The behavior is exactly what I was hoping for
      • Ok
      • dominikh
        throw a /8 at it and you'll probably never have to reuse an IP, without wasting 1GB of memory ;)
      • shykes
        Yeah that will be nice :)
      • testing with a /8
      • dominikh
        for what it's worth, my very first experience with docker was "why does it need 1GB of memory to run 'echo hello world'?"
      • shykes
        Yeah... We tried to optimize the experience on a vary narrow target first
      • (eg. out-of-the-box ubuntu quantal with apt-get install lxc)
      • dominikh
        hehe. yeah, not-so out of the box Gentoo with manually set up bridge here. aka the "unsupported" clientele
      • shykes
        not anymore :)
      • It's really awesome to receive such high-quality external contributions - we got a lot!
      • damnit can't resolve github.com while building
      • Oh...
      • My host is also on 10.0.0.0/8
      • dominikh
        heh
      • shykes
        facepalm
      • Awesome
      • Ok - merging
      • dominikh
        yay :)
      • shykes: bummer, my elaborate commit message got lost in the merge
      • shykes
        Yeah I was just looking at that
      • There was a conflict which I had to resolve manually
      • I don't understand what went wrong
      • The PR hasn't auto-closed
      • and your commit is not in the log
      • dominikh
        you created a whole new commit. not a merge commit, that is, but a normal commit, with my diff
      • shykes
        Ok - can you walk me through how not to do that?
      • dominikh
        I'm trying to figure out how you did it in the first place
      • shykes
        I'm a relative git newb. Recent hg convert
      • git merge your-branch
      • indicated conflict
      • resolved the conflict
      • did git add + git commit which is what the conflict warning told me to do
      • dominikh
        hm, one sec
      • shykes
        Maybe you should resolve the conflict yourself by merging master into your branch
      • then I can just auto-merge
      • dominikh
        really wondering why it didn't create a merge commit for you
      • shykes
        I also had to run gofmt -w
      • (I have a pre-commit hook which fails if gofmt doesn't pass)
      • maybe that's what caused it?
      • dominikh
        mhm, not sure. hm, did you just "unpush" the merge?
      • shykes
        let's assume, completely hypothetically, that I did
      • dominikh
        haha
      • shykes
        hey I'd rather break a pull or 2 than not give fair credit for a great contribution