Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2014-03-25

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:38 pk joined #gluster-dev
00:54 badone joined #gluster-dev
01:17 tdasilva joined #gluster-dev
01:21 tdasilva left #gluster-dev
01:24 bala joined #gluster-dev
01:50 lpabon joined #gluster-dev
02:14 pk left #gluster-dev
02:36 pk joined #gluster-dev
02:36 pk left #gluster-dev
03:24 badone joined #gluster-dev
03:26 shubhendu joined #gluster-dev
03:41 kanagaraj joined #gluster-dev
03:43 sahina joined #gluster-dev
03:49 badone_ joined #gluster-dev
03:51 itisravi joined #gluster-dev
04:01 ndarshan joined #gluster-dev
04:24 bharata-rao joined #gluster-dev
04:34 raghu joined #gluster-dev
04:59 ppai joined #gluster-dev
05:05 deepakcs joined #gluster-dev
05:17 kdhananjay joined #gluster-dev
05:29 bala1 joined #gluster-dev
05:34 hagarth joined #gluster-dev
05:48 lalatenduM joined #gluster-dev
05:49 pk1 joined #gluster-dev
06:00 aravindavk joined #gluster-dev
06:08 spandit joined #gluster-dev
06:27 mohankumar joined #gluster-dev
06:35 bala2 joined #gluster-dev
06:47 sahina joined #gluster-dev
06:50 hagarth joined #gluster-dev
06:51 kanagaraj joined #gluster-dev
07:39 kanagaraj joined #gluster-dev
07:48 bala1 joined #gluster-dev
07:59 sahina joined #gluster-dev
08:01 hagarth joined #gluster-dev
08:50 raghu` joined #gluster-dev
09:02 kanagaraj joined #gluster-dev
09:06 ndarshan joined #gluster-dev
09:06 sahina joined #gluster-dev
09:07 shubhendu joined #gluster-dev
09:09 bala1 joined #gluster-dev
09:10 ppai joined #gluster-dev
09:12 ppai jclift, ping
09:22 jclift ppai: 퐁
09:23 jclift ppai: Reading through your email now :)
09:23 ppai jclift, on a totally unrelated topic - is it possible to make/implement glupy in pypy
09:24 jclift ppai: Btw, "from gluster import glupy" won't work.  It's "from gluster.glupy import *".
09:24 ppai jclift, yes, just wanted to check if I get "module not found"
09:24 jclift ppai: I have no idea.  I'm still getting back into learning Python, and haven't looked at Pypy at all yet.
09:24 jclift ppai: Ahhh, cool. :)
09:24 * jclift is not a deeply experienced Python coder
09:25 jclift (but working on it)
09:25 jclift ppai: The print statements in heeloworld and debug-trace.py go to stdout or stderror.  I don't remember which.
09:26 jclift ppai: They show up when a person starts glusterfs in the foreground with --debug or -N.
09:26 jclift ppai: But yeah, they don't turn up in the normal log files.
09:27 ppai jclift, I don't know anything about pypy either but this video got me thinking https://www.youtube.com/watch?v=l_HBRhcgeuQ , do watch it when free
09:27 jclift 43 mins
09:29 jclift will try..
09:30 jclift ppai: That's interesting about -DDEBUG.  Yeah, I'll test that right away.  Just finished reading the morning email, so good timing. :)
10:08 jclift ppai: Wonder if these patches are causing the -DDEBUG problems? http://review.gluster.org/#/c/6957/ http://review.gluster.org/#/c/7029/
10:09 ppai jclift, I don't think so.. -DDEBUG is still there
10:10 jclift No worries. :)
10:12 ppai jclift, https://bugzilla.redhat.com/show_bug.cgi?id=1035751 is similar but not the same
10:12 glusterbot Bug 1035751: unspecified, unspecified, ---, kdhananj, MODIFIED , Brick processes crash when the test 'bug-887098-gmount-crash.t' is run after building glusterfs with DDEBUG flags
10:33 jclift ppai: Ahhh
10:34 jclift ppai: That's interesting, because the backtrace in the core from doing that with the glupy translator was something about memory accounting too
10:35 jclift ppai: Maybe the same fix will work if applied to glupy.h/.c?
10:36 ppai jclift, I'm trying that right now, will see
10:36 jclift ppai: Cool. :)
10:41 ppai jclift, haha, works :)
10:41 ppai jclift, or so I think
10:43 ppai jclift, can u apply this and confirm ? http://fpaste.org/88338/
10:45 jclift ppai: Will do
10:50 shubhendu joined #gluster-dev
10:52 kanagaraj joined #gluster-dev
10:52 ndarshan joined #gluster-dev
10:52 sahina joined #gluster-dev
10:53 bala1 joined #gluster-dev
10:54 jclift ppai: Yep, that stops the crash here too.
11:03 ppai joined #gluster-dev
11:06 ppai jclift, i'm back, thanks
11:11 jclift :)
11:14 lpabon joined #gluster-dev
11:19 pk1 joined #gluster-dev
11:27 aravindavk joined #gluster-dev
11:41 edward1 joined #gluster-dev
11:47 sas joined #gluster-dev
11:56 sasundar joined #gluster-dev
12:11 jclift ppai: It's taking a bit longer to test than expected, as I'm hitting some teething problems on F20 with my new testing repo
12:12 jclift ppai: Still working on it, but it might be a few hours
12:12 ppai jclift, no problem
12:18 itisravi_ joined #gluster-dev
12:19 deepakcs joined #gluster-dev
12:20 jclift Note to self, the build.gluster.org build.sh script doesn't yet work on F20, due to treating warnings as errors.
12:33 tdasilva joined #gluster-dev
12:40 jclift lpabon: ping.  Is gluster-jenkins2 in Rackspace yours?
12:41 kkeithley1 joined #gluster-dev
12:41 jclift lpabon: Not seeing an easy way to determine vm ownership in the Rackspace UI. :/
12:43 kkeithley_mtg issue with vm ownership?
12:44 jclift kkeithley_mtg: Just trying to see who starts vm's, as there's a new jenkins one in there
12:44 kkeithley_mtg gluster-jenkins2 is mine
12:44 jclift kkeithley_mtg: Just for curiosity/awareness, nothing actually critical :)
12:44 jclift kkeithley_mtg: Cool. :)
12:45 kkeithley_mtg intended to be a slave for build.gluster.org
12:45 jclift kkeithley_mtg: No worries.
12:46 jclift kkeithley_mtg: Have you used the Rackspace API yet for things?
12:46 * jclift is just about to start looking at it
12:46 kkeithley_mtg for core gluster.  Someone who knows jenkins might help me set it up as a slave. I don't know if hagarth has cycles to spare
12:46 kkeithley_mtg jclift: no, I haven't tried the API
12:46 jclift lpabon knows jenkins
12:47 jclift And someone else mentioned they do too, just not remembering who that was...
12:48 kkeithley_mtg We wanted a slave running NetBSD too, but RackSpace hasn't got an image. (And FreeBSD is a type of Linux? lol)
12:48 jclift kkeithley_mtg: I'm thinking to try spinning up VM's remotely, to try each of the difference instance types running the regression tests say 10 times (1 time per 10 new vms), to work out the best bang for buck instance type we should use
12:49 jclift kkeithley_mtg: Yeah, noticed FreeBSD under Linux too ;)
12:49 jclift For NetBSD, I'm suspecting we just need to get NetBSD working on a VM once, and then snapshot it.
12:49 jclift kkeithley_mtg: From that snapshot we should be able to clone and start up new vms
12:50 kkeithley_mtg Oh, okay. I can snapshot the NetBSD vm I had running here
12:50 kkeithley_mtg s/had/have/
12:50 jclift Running in Rackspace?
12:51 kkeithley_mtg er, no.
12:51 jclift Ahh.
12:51 jclift I'm meaning that the Rackspace GUI has tools to snapshot a VM that's been setup
12:52 jclift There _might_ also be tools to upload customer vm snapshots.
12:52 jclift Haven't checked for that yet though.
12:52 kkeithley_mtg not following you. How do we get a NetBSD vm setup. Seems like an chicken-and-egg thing
12:53 kkeithley_mtg netbsd vm inside, e.g., a rackspace rhel/fedora vm?
12:53 jclift Step 1) Install standard full virtualised (hvm) linux.  Step 2, create 2nd partition on its disk.  Step 3, install FreeBSD to it's partition.  Step 4, update grub to point to new partition
12:53 jclift Step 5, reboot and hope :)
12:54 jclift More just hoping that approach will work.  Haven't yet tested it.
12:54 jclift Can't see why it wouldn't for a HVM based vm though
12:54 jclift For PVM it probably wouldn't
12:57 jclift Interesting.  Rackspace has "Auto scale" feature.  It seems to be based on Heat.
12:57 * jclift knows some of the Heat guys already
13:07 lpabon jclift: yes, gluster-jenkins is mine... i probably can place my initials as a prefix
13:14 jclift lpabon: jenkins2 is the one I was curious about :)
13:14 lpabon oh, hmm, let me see
13:14 jclift lpabon: It's kkeithley_mtg 's
13:15 jclift lpabon: Would you have time to help kkeithley_mtg set it up as a Jenkins slave?
13:16 lpabon Sure, but most likely by friday, kkeithley_mtg and I are in an all day meeting for the next three days
13:16 jclift Cool, np. :)
13:16 kkeithley_mtg yeah, friday is fine
13:20 hagarth joined #gluster-dev
13:33 deepakcs joined #gluster-dev
13:51 sasundar joined #gluster-dev
14:02 primechuck joined #gluster-dev
14:22 wushudoin| joined #gluster-dev
15:05 jobewan joined #gluster-dev
15:38 ndk joined #gluster-dev
16:23 mohankumar joined #gluster-dev
16:55 hagarth joined #gluster-dev
18:04 lpabon joined #gluster-dev
18:06 ndevos jclift, johnmark, anyone: what would be the best place for documentation of a GlusterForge project?
18:07 ndevos forge.gluster.org has a wiki, but we also have http://www.gluster.org/community
18:07 lpabon ndevos: any glusterforge project?  I always like a doc directory in the repo itself
18:07 ndevos lpabon: yes, but that is not very user friendly...
18:07 lpabon wiki is where documentation goes to die :-)
18:08 lpabon it depends, if the repo viewer automatically formats the documentation then it is a nice way to show doc...like Github does
18:08 ndevos oh, yes, that would definitely be my preference!
18:09 lpabon i guess, my suggestion would be... use github :-)
18:10 ndevos well, I dont like to spread the bits around too much, you need to keep them in sync too :-/
18:10 lpabon yep.  For gluster-swift, our main repo is gluster.org, then i push to github and forge using a shell script
18:12 ndevos well, I dont like that too much... and it confuses users too, they try to send pull-requests on github then
18:12 lpabon IMHO, I would use github because it provides a nice integration with Travis-ci.org.  It works really well.  That way projects can have automatic CI environments
18:12 * ndevos doesnt know what Travis-ci.org is, and if he needs that
18:13 lpabon Oohh Travis-ci is rainbows and ponies :-D... check this out:
18:13 lpabon https://github.com/lpabon/cmockery2  any commit or pull request is automatically tested by Travis-ci.org here: https://travis-ci.org/lpabon/cmockery2
18:14 lpabon Two main benefits of having a project be in Github: 1. Doc looks good.  2. Free CI :-)
18:15 lpabon ...just my $0.02
18:15 ndevos indeed, looks very nice :)
18:15 lpabon and it is really easy to setup.  Check out my setup file:  https://github.com/lpabon/cmockery2/blob/master/.travis.yml
18:15 ndevos but, I think we could do that with a scalable Jenkins setup too...
18:16 lpabon that part is hard, because someone would need to give access to that user.  Then we would need to create the project, and so on.
18:16 ndevos right
18:16 lpabon Jenkins does not provide a very easy method to be self-managed
18:16 ndevos oh, that can hardly be called a 'configuration file'
18:17 lpabon pretty sweet!  super easy to setup, compared to setting up jenkins and slaves and accounts, etc...
18:18 lpabon although.. Jenkins *does* provide excellent monitor tools for major projects, but for side projects, IMHO, i think Github-Travis is a great combination
18:24 ndevos that surely looks like it... and I would likely use it for non-gluster projects, but I really dislike having two git repos
18:25 edward1 joined #gluster-dev
18:58 edward1 joined #gluster-dev
19:16 edward1 joined #gluster-dev
19:21 lalatenduM joined #gluster-dev
19:22 lalatenduM hagarth, ping
20:41 tdasilva left #gluster-dev
20:57 johnmark lpabon: someday we're going to have multiple repos on git.gluster.org and automatic-mirroring on github, forge, etc.
20:57 johnmark ndevos: ^^^
20:57 johnmark someday... maybe... *sigh*
20:58 lpabon johnmark: lol, i just use a shell script on my system, simple enough
20:58 ndevos johnmark: okay, I guess I'll wait with any documenting until thats ready :D
20:58 johnmark ndevos: nooooo
20:58 lpabon johnmark: did you see the travis integration with github? That could work *really* well with glusterfs sub-projects (if we get unit and simple functional tests)
20:59 johnmark lpabon: yeah, well, it would also be nice if the forge could auto-mirror to github
20:59 johnmark we could do that, but it has to be a manual setup
20:59 lpabon or....
20:59 johnmark lpabon: no, I have not. That's a good idea though
21:00 ndevos maybe travis-ci.org could get extended to not only be integrated with github?
21:01 johnmark ndevos: would be nice
21:01 lpabon yeah, if it just pointed to a git repo, that would be nice
21:01 johnmark lpabon: or... ?
21:01 johnmark lpabon: yup
21:01 ndevos well, it seems to use some hook, but that could be hooked into gitorious too
21:01 lpabon im sure we could talk to them about it
21:02 ndevos lpabon: because you <3 automated tests, I think you're the right person to start that discussion ;)
21:02 lpabon :-D . i do.  I called my first child "unit test" :-)
21:03 ndevos lol!
21:14 johnmark LOL
21:14 johnmark awesome
21:20 semiosis johnmark: how attached are we to gitorious?
21:20 * semiosis not a fan
21:20 semiosis although maybe it just needs to be upgraded
21:40 johnmark semiosis: yeah, we're going to talk to them about that
21:40 johnmark semiosis: I'm becoming less of a fan wiht every passing day :(
21:41 semiosis my latest annoyance with it: i go to the projects page, enter my project name in the search box, get 78 results none of which are my project
21:41 semiosis meanwhile i use gitlab every day at work and <3 it more each day
21:45 kshlm joined #gluster-dev
21:45 kshlm joined #gluster-dev
21:45 kshlm joined #gluster-dev
21:50 johnmark yeah, I'm taking a good long look at gitlab
21:50 johnmark but there's no hosted service that offers it
21:50 johnmark that I know of
21:50 johnmark speaking of... we need a better way to navigate projects
21:59 a2 i'm sure you mentioned this before but i forget: what was the problem with github?
22:07 a2 gilab seems to have side by side view
22:07 a2 (of patches/diffs)
22:10 kshlm joined #gluster-dev
22:11 jclift Sounds like there's a market opportunity for setting up hosted GitLab.
22:11 * jclift tries to feel enthused about filling it
22:11 jclift Fails...
22:11 jclift :/
22:18 jclift a2: I think the reason we originally went with Gitorius is because we wanted to support a fully open source system, since GitHub isn't 100% open source (just parts of it).
22:18 jclift GitLab on the other hand is 100% Open Source
22:18 jclift But there doesn't seem to be a place offering hosting for that (yet).
22:19 semiosis johnmark: gitlab is cake to admin.  i'll run it on a rhcloud instance
22:19 semiosis once a month i run upgrade scripts
22:19 semiosis there is gitlab.com for hosted gitlab
22:19 jclift semiosis: With GitLab it used to have an issue of not letting projects be public to non-members
22:20 semiosis used to
22:20 jclift semiosis: So there was a guy that used to have a fork called Public GitLab.
22:20 semiosis gone
22:20 jclift semiosis: So they fixed that?  If so, then cool.
22:20 semiosis here is gitlab self-hosted on gitlab
22:20 semiosis https://gitlab.com/gitlab-org/gitlab-ce
22:20 semiosis clearly public
22:21 semiosis all public projects on gitlab.com: https://gitlab.com/public
22:23 jclift Seems a fairly artificial divide between Community Edition and Enterprise Edition: https://www.gitlab.com/gitlab-ee/
22:23 jclift Git hook rules are only in Enterprise Edition
22:23 jclift However, that's not a show stopper
22:24 semiosis i've never even looked at the EE
22:24 semiosis it's pretty new
22:24 jclift semiosis: Just looking for a hosted option.
22:24 semiosis https://www.gitlab.com/cloud/
22:25 semiosis thats it
22:25 jclift Reading now
22:26 semiosis i thought gitorious was self-hosted
22:26 semiosis if we're not self hosting, why dont we just use github then?
22:27 semiosis only reason I use gitlab at work is because we have 100s of private repos that would cost way too much on github
22:27 * jclift would prefer GitHub too
22:27 jclift My understanding is that we went with Gitorius because it was the best 100% open source solution available at the time
22:28 johnmark ok, for the umpteenth time
22:28 jclift ;D
22:28 semiosis ha
22:28 johnmark github has lots of metadata that is *not* gitable
22:28 johnmark think of all the issue tracking, and other stuff that goes into a github project
22:28 johnmark once you stuff all your things into github, it ain't coming back
22:28 johnmark it's a data roach motel
22:28 johnmark and that model pisses me off
22:28 jclift Walled garden
22:28 johnmark why do you think I left splunk after a year?
22:29 johnmark exactly
22:29 semiosis hehehe
22:29 johnmark I'm fine with mirroring stuff on github
22:29 jclift So, we have our own walled garden instead
22:29 johnmark it's great for pushing code out to many many poeple
22:29 johnmark no, we don't have a walled garden
22:29 jclift At least its _our_ garden
22:29 johnmark LOL
22:29 johnmark you can take every single piece of data that's on gitorious and apply it elsewhere
22:29 jclift (slowly)
22:29 johnmark ...with some tolerance for pain :)
22:29 johnmark LOL
22:29 johnmark right. fair enough
22:29 semiosis like... to github ;)
22:30 semiosis j/k
22:30 johnmark haha! sure :)
22:30 johnmark having said that, I *do* want to continue to use github, and I would *love* to explore ways to mirror lots of content on github
22:30 johnmark but not at the expense of our community's integrity
22:30 jclift So.  Would it be faster to a) learn how to use each of the pieces in Gitorius data dump when Gitorius goes broke, or b) Create the non .git repo bits for a project on GitHub
22:30 jclift I'd suggest b)
22:31 johnmark depends on what pieces you want to use
22:31 jclift Agree
22:31 johnmark if you only want to use the repo parts
22:31 johnmark and the CMS parts that are git-able
22:31 johnmark then yeah, that's low risk
22:31 johnmark but then we have the other problem - creating any type of community around our projects
22:31 semiosis gitlab wikis are git repos
22:31 johnmark it' sreally hard with github
22:32 johnmark right, and that' sa model I'd like to move to
22:32 johnmark and actually, gitorious wikis are also git-able
22:32 johnmark but they're also pretty screwed up
22:32 johnmark in their implementation
22:32 jclift GitHub API seems fairly powerful.  Can create/update/close/etc issues
22:33 semiosis +1
22:33 johnmark ah yes, the "open API"
22:33 jclift So, someone's probably written import/export scripting for it
22:33 johnmark which isn't open
22:33 semiosis but we have BZ for issues
22:33 johnmark because you still have to implement the backend for it
22:33 johnmark and believe me, no one hates BZ more than me
22:33 johnmark oh wait, I'm wrong - the people who have to use BZ every day hate it much more than I do :)
22:33 semiosis haha
22:34 johnmark but yeah, it's a conundrum
22:34 johnmark I'm personally thinking about gitlab
22:34 johnmark or even moving some things to launchpad
22:34 johnmark (shhhhhh)
22:34 johnmark like issue tracking, for example
22:34 johnmark a2: I think gitlab is definitely worth investigation
22:35 johnmark I have a call to setup with the gitorious folks this week or next
22:35 jclift The main thing which GitHub has, which Gitorious and GitLab don't... is huge user community
22:35 johnmark if any of y'all want to be on it, let me know
22:35 johnmark this is true - which is why I am an advocate for placing code on github
22:35 johnmark just like I used to advocate placing code and releases on sourceforge  back in teh day
22:36 johnmark but *not* using it as your primary dev platform
22:36 jclift johnmark: My point about the GitHub API is that it's not as much of a walled garden as some people like to make out.  The non-git stuff is likely able to be extracted via API
22:36 johnmark jclift: fair enough. if you can find a way to create a shared "gluster.org" experience on github, let me know
22:37 semiosis https://github.com/blog/674-introducing-organizations
22:37 johnmark and I admit to not having looked at it for some time - things may have changed
22:37 jclift My viewpoint is pretty simple.  For the forge, we should be giving our Gluster projects the very best chance of awesome outcomes.  Gitorious doesn't do that (we all know and agree, and there's work being done to fix, etc).
22:37 johnmark jclift: ok
22:37 jclift So, +1 for GitLab as a step in the right direction
22:38 jclift But the gold standard for objective stated above is GitHub
22:38 jclift The others are just pretenders
22:38 semiosis well imho gitlab is the only real competition for github
22:38 semiosis they're catching up so fast
22:39 jclift They do have a rep for evolving quickly
22:39 jclift Anyway, I don't hugely care atm.  Short term I have other tasks that I need to get sorted. :)
22:39 semiosis truth
22:40 johnmark jclift: ok, let's make a deal
22:40 johnmark semiosis: just saw that the organizations part is free for oss porjects
22:40 johnmark jclift: if we can use github, then let's do it, but you're personally responsible for keeping a 100% mirrored environment somewhere
22:41 * johnmark waits...
22:41 jclift You need to find someone that gives enough of a shit to actually code the missing bits
22:42 jclift I'm only half caring enough :)
22:42 johnmark LOL!
22:42 johnmark I see...
22:42 johnmark well, it's worth investigating
22:42 johnmark and enough people have complained that it's an issue
22:43 johnmark and it woudl solve other problems - travis-ci, for example
22:43 johnmark so yeah, I can see the benefits
22:43 jclift Yeah.  I've done basic "GitHub Organisations" stuff for another project before (BrowserQuest), but we didn't really investigate it for something the scale that the forge would beed to be
22:43 johnmark ok
22:43 jclift I can take a look at it again later, to see if it would have the capabilities we need
22:43 johnmark ok
22:43 johnmark doh... gotta run for now.
22:43 johnmark Will be back later tonight
22:44 semiosis me too, shortly
22:44 jclift I kind of doubt it, as I don't remember there being much in the way of branding possible, etc.
22:44 johnmark yeah, looks like you can stick up a logo
22:44 jclift However, there might be suitable workarounds possible
22:44 johnmark but we could potentially build a front end to it, seeing as how everythign is API-driven
22:44 jclift eg forge.gluster.org be our own front-end, with API
22:44 jclift yeah
22:44 johnmark jclift: fair enough. I'll put it on my list of things to do
22:44 johnmark right
22:46 semiosis gitlab's answer to travis-ci, fwiw: https://www.gitlab.com/gitlab-ci/
22:46 semiosis i havent used it
22:47 semiosis at work we just have jenkins with the usual git plugin polling gitlab repos
22:48 * semiosis afk

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary