[09:14] <frankban> morning gmb
[09:14] <gmb> Hi frankban
[11:52]  * gmb -> lunch
[12:58] <frankban> gmb: thanks for the review
[13:01] <gmb> np
[13:08] <gary_poster> benji frankban gmb call in 2
[13:09] <gmb> ack
[14:17] <benji> gary_poster: I'm about to send this to the -tech list: https://pastebin.canonical.com/61706/; feedback?
[14:21] <benji> gmb and frankban: have you used the terminal broadcasting hack?  If so (or not, even), do you have any feedback on my future message to the tech list (https://pastebin.canonical.com/61706/) or on the wiki page describing the hack (https://dev.launchpad.net/yellow/RemoteTerminalBroadcasting)?
[14:22] <frankban> benji: not used, taking a look
[14:25]  * gmb neither
[14:25]  * gmb looking
[14:28] <gmb> benji, Both look good to me. Looking forward to having something to try this with properly :)
[14:36] <gary_poster> benji, sorry, didn't see notification. looking.  Also, I'm looking at testrepository to see feasibility of adding a -v
[14:36] <gary_poster> more important is to help you if that is valuable
[14:36] <gary_poster> lemme know
[14:36] <benji> gary_poster: the slave just started so I'm looking at it now; I'll know more in a few minutes
[14:38] <gary_poster> benji, pastebin looks good.  one possible change: terminal size doesn't need to match.  viewer terminal must be same size or bigger.  Maybe what you said is simpler for an email.  I think it is more nuanced on the web page itself
[14:39] <benji> gary_poster: I changed it on the wiki; I /think/ that under some circumstances being exactly the same size is important
[14:39] <gary_poster> huh, ok
[14:50] <frankban> benji: maybe I am missing something, but, in the first scenario, if 2 users have ssh access to a machine, isn't it easier to just use tmux with a shared socket?
[14:50] <benji> frankban: it may; how do you do that?
[14:56] <gary_poster> whee
[14:56] <gary_poster> benji, am back post-crash-and-update if you were looking for me
[14:57] <benji> gary_poster: I wasn't ;)
[14:57] <gary_poster> :-) k
[15:01] <frankban> benji: http://readystate4.com/2011/01/02/sharing-remote-terminal-session-between-two-users-with-tmux/
[15:01] <benji> frankban: thanks!  I'll take a look.
[15:01] <gmb> Every day's a school day on the Launchpad Yellow Squad :)
[15:02] <benji> frankban: that's a little different than what we're doing here, in one good way and one bad way.  Good: both users can control the session.  Bad: both users have to be able to SSH into the host box, which is often an issue (either of security or hassle)
[15:02] <gary_poster> heh :-)
[15:03] <gary_poster> but that's your first case though, yeah?
[15:03] <gary_poster> the heh was to school day comment
[15:03] <gary_poster> the first case comment was an aside to benji's reply to frankban
[15:03] <benji> gary_poster: nope; using the tmux (or screen) approach means that we'd have to *work* on chinstrap, not just use it as a shared conduit
[15:03] <gary_poster> :-)
[15:04] <gary_poster> ah!
[15:06] <frankban> benji: ah, I see
[15:07] <gary_poster> benji, I'd expect a lot of comments of the sort "why don't you just use X".  Positioning the tool maybe should come first (i.e., why not screen, why not tmux, and so on).  Just a thought
[15:08] <benji> gary_poster: good point; by "positioning" do you mean describing it relative to the other solutions?  I tried to do that vs. hangouts, but I only glanced at screen/tmux.
[15:09] <benji> gary_poster: oh, that is what you meant (if only I could read)
[15:09] <gary_poster> Maybe "I was interested in finding a screen-equivalent that didn't do X; a tmux-equivalent that didn't do Y; a hangout/skype screen sharing that allowed Z; and a tool that was easy to set up and use.  Unix to the rescue."
[15:14] <benji> gary_poster: I like that, but I already wrote it another way and I'm lazy ;) https://pastebin.canonical.com/61714/
[15:14] <benji> gary_poster: the bit starting at "We've tried using screen" is new.
[15:21] <gary_poster> benji, second sentence needs editing (probably just delete "The one drawback is" and capitalize "That."  Reading more...
[15:22] <gary_poster> I mean
[15:22] <gary_poster> delete "that" too
[15:22] <gary_poster> and capitalize "sharing"
[15:23] <gary_poster> benji, "lessoned" -> "lessened"
[15:23] <benji> gary_poster: good catch, I wanted to use a contrasting word, so I deleted "The one drawback is that" and added "Unfortunately, "
[15:23]  * benji fixes spelling.
[15:24] <gary_poster> I would mention talking about tmux and...what's that canonical thing in the same space that has an asian name meaning foldable temporary wall...
[15:24] <gary_poster> in the same area as screen
[15:25] <gary_poster> since AIUI the arguments are the same
[15:25] <benji> b... something
[15:26] <gary_poster> gmb, do you remember the name of that canonical I-am-not-tmux thing?  juju uses it too; looking there
[15:26] <benji> byobu
[15:26] <gary_poster> that's it
[15:26] <gary_poster> sorry gmb nm :-)
[15:26] <gary_poster> otherwise looks good benji
[15:27] <gary_poster> gmb, I know you have gone down this maze of twisty passages before: following code path through subunit + testtools + testrepository = think about doing something else instead :-)
[15:31] <gmb> gary_poster, benji: Sorry, was getting a cup of tea. Can I assume that I can just raise an eyebrow at this conversation and move on, or is there still a question dangling...?
[15:32] <gary_poster> gmb, you can move on, but the very last comment was directly addressed to you and had nothing to do with the foregoing...and is not particularly important
[15:32] <gmb> Oh, er. Right.
[15:33] <benji> heh
[15:33] <gmb> gary_poster, And yes, it's a knotty and horrible place to be.
[15:33] <gary_poster> :-)
[15:33] <gmb> Oftentimes the only way to navigate out of it is to ask one of {lifeless,jml} for a map.
[15:33] <gary_poster> ah
[15:33] <gary_poster> ok
[15:34] <gmb> lifeless, of course, is like the guy at a crossroads who gives you really long-winded directions and then, when he's finished, goes back and corrects himself on one of the more esoteric but fundamentally important things that you've already forgotten he said.
[15:34] <gary_poster> lol, yes
[15:34] <gary_poster> well described
[15:34] <gmb> Note that I may be jaded about asking him for questions when I'm not capable of physically stopping him.
[15:34] <gmb> s/questions/answers/
[15:34] <gary_poster> heh
[15:35]  * gary_poster goes to branch subunit...
[15:36] <gmb> Best of luck. We'll let your children know you love them if you don't come back.
[15:36] <gary_poster> lol thank you
[15:36] <benji> lol
[15:36] <benji> gary_poster: email to -tech away
[15:36] <gary_poster> cool benji
[15:37] <benji> gary_poster: the tests are very quietly running at http://ec2-107-21-150-52.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/0/steps/shell_8/logs/stdio while I work on the tracking issue report
[15:37] <gary_poster> cool benji.  yeah, no errors so far...
[16:05] <benji> gary_poster: that's an interesting exception coming from the stdlib tempfile module: http://ec2-107-21-150-52.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/0/steps/shell_8/logs/stdio
[16:05] <gary_poster> benji, smells like another overlayfs error on the face of it, yeah?
[16:06] <benji> gary_poster: yep, that's what I was thinking
[16:09] <benji> gary_poster: tracking issue report: https://pastebin.canonical.com/61724/
[16:09] <benji> let me know if there is something else you'd like on it
[16:24] <gary_poster> benji, for the test failure, also looks like pub_config.htaccessroot was not made, and that's the problem that led to the traceback, so we can't simply expect to run "tempfile.mkstemp" to trigger the problem.  tracking issue: excellent thanks.  I have the RT info, as you'd guess.
[16:25] <benji> gary_poster: re. failure: hmmm; re. report: cool
[16:52] <frankban> gary_poster (or someone else): the MP for shell toolbox is ready: https://code.launchpad.net/~frankban/python-shell-toolbox/helpers/+merge/96180
[16:53] <frankban> ... and I am sorry for the long diff
[17:07] <gary_poster> frankban, interesting, so you opted to put run later down in the file to keep things alphabetized?
[17:07] <frankban> gary_poster: yes, i've reordered functions inside the module
[17:07] <gary_poster> cool
[17:32] <gmb>  gary_poster, Er, I *think* I just uploaded a new version of charm-tools to my PPA... If this works I can push it to the yellow PPA, too...
[17:34] <gary_poster> gmb, cool.  Do you want me to check something or other?  Uh, and charm-tools...oh, so you mean, this includes the new binary of python-charm-tools or whatever it is?
[17:35] <gmb> gary_poster, Well, AIUI, all we actually needed to do was and a python charm helpers package, which bac already did. My job was to package it up for us to use, which I think I just acheived.
[17:36] <gmb> ... after spending some time fighting build deps on precise, which was fun.
[17:37] <gary_poster> gmb, right I forgot bac had gotten pretty far on that
[17:38] <gmb> gary_poster, The good news is that once you're system's not being a dork, bzr builddeb is mind-blowingly excellent.
[17:38] <gary_poster> so, yeah, gmb, I think we ought to put it in yellow, and then propose it to clint
[17:38] <gmb> I must remember to by jamesw a beer.
[17:38] <gary_poster> :-)
[17:38] <gmb> gary_poster, Okay, let me make sure it's uploaded to mine first... it's in the ether somewhere at the moment.
[17:38] <gmb> (i.e. dput says "yay", Launchpad says "No")
[17:38] <gary_poster> ack
[17:45] <gary_poster> frankban, approved with comments https://code.launchpad.net/~frankban/python-shell-toolbox/helpers/+merge/96180
[17:45] <frankban> thank you gary_poster
[17:46] <gary_poster> thank you!
[17:48] <gary_poster> benji, sabdfl is following you :-)
[17:53] <gmb> Argh, silly man. Trying to upload a binary to a PPA. Nevergunnahappen.
[17:53]  * gmb rebuilds with -S
[17:54] <gmb> \0/ accepted.
[17:55] <gmb> gary_poster, I've copied the package to the Yellow PPA.
[17:56] <benji> gary_poster: heh, it seems so
[17:56] <gary_poster> gmb, yay!  So...if we need to do this again, is the knowledge on bac's page, or in your brain, or some combination?
[17:56] <gmb> gary_poster, I'm going to update bac's page.
[17:56] <gary_poster> awesome, thanks gmb
[17:57] <gmb> how do you strikethrough an entire wiki page again? ;)
[17:59] <gmb> Ah, although builddeb mightn't help us with the no-such-package-exists-yet case, actually.
[18:02] <gary_poster> gmb, are you going to have the MP for the change to charm tools today?  If so, I'd love to have a URL
[18:02] <gary_poster> If not, np, but I won't wait for it.
[18:02] <gmb> gary_poster, I will, yes.
[18:03] <gary_poster> great, thank you.  Lemme know and I will include it in our review response to ~charmers
[18:06] <gary_poster> benji, we have changes for the master and slave charm, right?  If you can make MPs for them, that would be good to get them landed before I ask the ~charmers to approve our charm.
[18:06] <gary_poster> Or ask me to make MPs, pointing me to the branches
[18:06] <benji> gary_poster: I can do MPs momentarily.
[18:06] <gary_poster> thank you
[18:10] <gmb> gary_poster, https://code.launchpad.net/~gmb/charm-tools/add-charm-helpers/+merge/96204
[18:10] <gary_poster> thank you gmb, great
[18:11] <gmb> np.
[18:11]  * gmb -> dinner, but will check back later w.r.t build status of the package, etc.
[18:11] <gary_poster> gmb, could you quickly remove "7	+# NOTE TO LAUNCHPAD DEVELOPERS: This is a bootstrapping file from the
[18:11] <gary_poster> 8	+# setuptools project.  It is imported by our setup.py."
[18:11] <gary_poster> :-)
[18:12] <gary_poster> well, do it later
[18:12] <gary_poster> :-)
[18:14] <gary_poster> benji, do you recall details as to why we rejected the idea of combining the master and slave charm into one?  My memory is that we felt that overlap was not significant, and that the idea of flexibly becoming the master or the slave did not jibe with our need to have a time-consuming preparation of slaves, and...there was something else about how a charm decided to be the mysql master or slave that didn't make s
[18:14] <gary_poster> ense for us.  Any other recollections from you?
[18:15] <benji> gary_poster: the "no overlap at all" but was the biggest, but also our desire to do as little dynamic setup as possible was another.  With the two-in-one approach everything would have been dynamic
[18:16] <gary_poster> right, thanks benji
[18:46] <benji> gary_poster: it turns out that all of my changes to the slave have already been landed or superseeded, so the trunk is good to go as-is
[18:46] <gary_poster> benji, even to args?
[18:46] <benji> gary_poster: as for setuplxc: I might have been a little light on this MP, but I think the diff is small enough for me to get away with it: https://code.launchpad.net/~benji/launchpad/more-integration/+merge/96211
[18:46] <gary_poster> benji, I saw it; will be happy to get to it very soon
[18:46] <benji> gary_poster: surprisingly, yes; I think frankban must have done it
[18:47] <gary_poster> ah ok...there were two changes to args...one was ssh key
[18:47] <gary_poster> what was the other
[18:48] <benji> gary_poster: user (-u buildbot)
[18:48] <benji> maybe I accidentally pushed to the slave trunk and that's how those got there... I don't know (being able to commit to the trunk isnt' good if we want to do reviews)
[18:49] <gary_poster> :-)
[18:49] <gary_poster> ok thanks benji
[18:49] <gary_poster> was there anything for the master?
[18:50] <gary_poster> I thought there was
[18:50]  * gary_poster checks notes
[18:50] <benji> nope
[18:50] <gary_poster> No
[18:50] <gary_poster> Right ok
[18:50] <gary_poster> I'll go look at setuplxc
[18:50] <gary_poster> btw benji, thoughts on https://docs.google.com/a/canonical.com/document/d/1UuW2T7Vhmp4zqa5Rz31TErg46EnO936Yehddp8p10AY/edit welcome
[18:51] <benji> gary_poster: I've already corrected a typo and am almost done reading it
[18:51] <gary_poster> Cool thank you :-)
[18:52] <benji> gary_poster: bikeshed: "DONE?" sounds weak, I'd say either "DONE" or remove it alltoghether so it's just "BLOCKED ON..."
[18:53] <benji> gary_poster: looks good
[18:54] <gary_poster> benji fair enough, changed.  The deployment plan might actually be done; not sure if we need more approval than Robert.  Since I don't know that, I guess it is not done either way :-)
[18:54] <benji> :)
[18:58]  * benji works on bash completion for juju for a few more minutes before delving back into the tests-on-slave issues.
[18:59] <gary_poster> benji, ow: http://ec2-107-21-150-52.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/0/steps/shell_8/logs/stdio
[18:59] <benji> gary_poster: yeah, I'm not sure what's going on there; I /think/ there was a timeout or other ssh failure between the host and container that caused at least a restart, but it may have caused some of that flood
[19:00] <gary_poster> huh
[19:00] <gary_poster> maybe so
[19:21] <gary_poster> benji, setuplxc approved with trivial comment
[19:21] <gary_poster> https://code.launchpad.net/~benji/launchpad/more-integration/+merge/96211
[19:36] <benji> gary_poster: I now know as much about bash completion as I do about zsh (which isn't saying much) :)
[19:36] <benji> gary_poster: if you're looking for something to pair on, I'm going to start looking at those test failures now
[19:48] <gary_poster> benji, cool.  I'll be available in 12 or so
[19:48] <benji> gary_poster: k
[20:13] <gary_poster> benji, I'm in goldenhorde, and for some reason my camera is working again.  <shrug>
[21:12] <gary_poster> benji, what we did with the sleep is in the wrong place
[21:12] <gary_poster> it needs to be on the ssh side
[21:12] <benji> gary_poster: ah! you're right
[21:13] <gary_poster> benji flacoste suggests protocol keep alive
[21:13] <gary_poster> an ssh setting
[21:14] <benji> gary_poster: that rings a bell... let me look
[21:29] <gary_poster> benji ProtocolKeepAlives 240
[21:30] <benji> gary_poster: I'm not sure that does what we want; that's more about keeping an interviening party from cutting the connection (like some firewalls do)
[21:31] <gary_poster> benji, we can move the xvfb card to landing, right?  it will be done when that branch I reviewed lands right?
[21:31] <benji> gary_poster: right, we believe xvfb is happy now
[21:31] <benji> do we have a card for "make all the tests pass"?
[21:31] <gary_poster> benji, no that's the final stage
[21:53] <benji> gary_poster: broken pipe error with ProtocolKeepAlives (actually, ServerAliveInterval which is it's real name, ProtocolKeepAlives is an alias for what was once a Debian-only feature)
[22:15] <gary_poster> benji, :-/