Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2014-03-26

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

All times shown according to UTC.

Time Nick Message
00:25 jclift johnmark semiosis: Just noticed that GitLab is coded in Ruby.  Zero interest in using it now.
00:31 ndevos jclift: and whats gitorious written in?
01:03 bala joined #gluster-dev
01:47 [o__o] left #gluster-dev
01:50 [o__o] joined #gluster-dev
03:11 bharata-rao joined #gluster-dev
03:30 shubhendu joined #gluster-dev
03:33 kanagaraj joined #gluster-dev
03:40 itisravi joined #gluster-dev
04:04 sahina joined #gluster-dev
04:08 pk joined #gluster-dev
04:26 deepakcs joined #gluster-dev
04:40 mohankumar joined #gluster-dev
04:41 kdhananjay joined #gluster-dev
04:45 lalatenduM joined #gluster-dev
04:56 aravindavk joined #gluster-dev
05:06 hagarth joined #gluster-dev
05:11 bala joined #gluster-dev
05:12 ndarshan joined #gluster-dev
05:18 ppai joined #gluster-dev
05:26 kdhananjay joined #gluster-dev
06:12 spandit joined #gluster-dev
06:15 raghu joined #gluster-dev
07:00 lalatenduM hagarth, ping
07:02 lalatenduM pk, ping
07:03 kdhananjay joined #gluster-dev
07:12 pk lalatenduM: pong
07:12 pk lalatenduM: Everybody is in snapshot meeting...
07:13 lalatenduM pk, ohh, lets talk later then :)
07:13 lalatenduM pk, when u have time, check latest coverity run report in gluster-devel mailing list
07:18 pk lalatenduM: Replied
07:19 pk lalatenduM: This is because of afrv2 merge over the weekend
07:19 lalatenduM pk, yup, thanks :)
07:19 lalatenduM pk, thats the reason I did the coverity run to see if afrv2 introduced anything :)
07:29 pk lalatenduM: How to check if we fixed everything before sending the patches?
07:31 lalatenduM pk, thats difficult using covscan , I have an idea. once you guys fix them , let me know
08:07 hagarth joined #gluster-dev
08:21 edward1 joined #gluster-dev
09:35 bala joined #gluster-dev
09:50 sahina joined #gluster-dev
09:51 hagarth joined #gluster-dev
09:51 kanagaraj joined #gluster-dev
10:28 kanagaraj joined #gluster-dev
10:29 sahina joined #gluster-dev
10:32 shyam joined #gluster-dev
10:33 bala joined #gluster-dev
10:35 hagarth joined #gluster-dev
10:53 sahina joined #gluster-dev
10:54 kanagaraj joined #gluster-dev
10:55 nixpanic joined #gluster-dev
10:56 nixpanic joined #gluster-dev
10:58 kdhananjay joined #gluster-dev
11:03 hagarth joined #gluster-dev
11:09 jclift ndevos: Gitorious seems to be written in Ruby too. :(
11:09 ndevos jclift: thats what I thought!
11:09 jclift No good options. ;)
11:13 kshlm jclift, good options for what?
11:18 ndevos argh, rewriting writing of iovecs to /dev/fuse keeps on breaking, maybe the total size should be < PAGESIZE and not <= ?
11:23 kanagaraj joined #gluster-dev
11:27 aravindavk joined #gluster-dev
11:31 xavih ndevos: is this to handle readdir(p) requests greater than a page size or it also fails with other requests ?
11:32 ndevos xavih: yeah, for some reason it always fails if the iovec is 4096 bytes, smaller ones work fine - I must be missing something
11:33 xavih ndevos: could it be related to the size that fuse sends on a readdir(p) request ?
11:34 xavih ndevos: it determined the maximum data it want to read, isn't it ?
11:34 ndevos xavih: I'm not sure how you mean that?
11:34 xavih ndevos: what happens if you send back more data than requested ? is this a problem for fuse ?
11:35 ndevos xavih: oh, interesting, I did not notice that - the size is in the fuse_header_in (or whatever the name)?
11:35 xavih ndevos: a reddirp request receives an offset and a size
11:35 xavih ndevos: yes
11:36 ndevos xavih: ah, just like the network protocol :D
11:36 xavih ndevos: it seems that fuse xlator mostly ignores that size
11:36 xavih ndevos: I don't know if this could be a problem
11:37 ndevos xavih: that could be, I'll test with an artificial max iovec size of 2048 and see if the first writev() passes
11:38 ndevos xavih: in case that errors out too, I guess the size of the fuse-request really matters
11:39 xavih ndevos: yes, this test could tell us if the problem is the I/O vector size or it's really related to a maximum buffer size determined by the fuse request
11:40 ndevos xavih: thanks for the idea :)
11:40 xavih ndevos: yw :)
11:46 nishanth joined #gluster-dev
11:50 ndevos xavih: yeah, looking at the readdir(p) functions, it seems that the size is set there... I guess we'll just need to ignore the extra dirents and have them sent a 2nd time
11:51 hagarth joined #gluster-dev
11:51 xavih ndevos: yes, I'm looking at the kernel side of fuse and it seems to limit the returned to the amount specified on the request
11:51 xavih ndevos: otherwise it fails
11:52 ndevos xavih: yes, it seems so
11:52 xavih ndevos: this makes more complex the change... :-/
11:53 * ndevos in fact has fs/fuse/dev.c open too
11:53 aravindavk joined #gluster-dev
11:54 ndevos xavih: not really, the change is not optimal, where the dirents are put in the buffer, we just return prematurely, causing the offset in the next fuse-request to re-receive the last dirents
11:55 xavih ndevos: ah, ok, I was thinking to keep the unsent dirents to merge them with a future request... your solution is considerably easier... :P
11:57 ndevos xavih: keeping them would definitely be nice, but uhm, yeah, thats more difficult
11:57 xavih ndevos: yes, *much* more
11:59 itisravi joined #gluster-dev
11:59 xavih ndevos: since xlators do not have real knowledge of how much space needs every entry, the size parameter in readdir(p) calls is useless
12:01 xavih ndevos: it would be better that all xlators return as much entries as possible, and buffer them in the fuse xlator to serve them when needed
12:02 xavih ndevos: it's a small caching in the fuse xlator side
12:02 xavih ndevos: the problem is that this would need a global modification to make it really useful...
12:02 xavih ndevos: the problem is that this would need a global modification to make it really useful...
12:07 shyam joined #gluster-dev
12:10 xavih_ joined #gluster-dev
12:12 xavih_ ndevos: sorry, I've lost connection
12:12 hagarth jclift: ping
12:13 ndevos xavih_: my bluetooth keyboard lost connection too...
12:16 ppai joined #gluster-dev
12:17 kdhananjay joined #gluster-dev
12:22 jclift hagarth: 퐁
12:23 kanagaraj joined #gluster-dev
12:24 jclift kshlm: re "No good options".  Brainstorming about what would make for a better version of the Forge, instead of Gitorious.  eg GitLab, or GitHub, or ?
12:24 hagarth jclift: google translate makes it Fong for me :)
12:24 jclift Seems like both Gitorious and GitLab are written in Ruby.
12:24 jclift hagarth: It's supposed to be "Pong" ;)
12:24 ajha joined #gluster-dev
12:25 jclift hagarth: http://unicode-table.com/en/D401/
12:25 hagarth jclift: would you be able to lead the community meeting today? I hate to admit that I have one more conflict today too..
12:25 jclift hagarth: Heh, was kind of expecting that.  Yeah, no worries. ;)
12:25 hagarth jclift: haha - you know why I ping you on wednesdays ;)
12:26 jclift :)
12:26 hagarth jclift: nevertheless, thanks a ton!
12:27 jclift np at all.  I'll send out the reminder in a bit, etc
12:29 kkeithley1 joined #gluster-dev
12:35 sahina joined #gluster-dev
12:52 ppai joined #gluster-dev
12:53 kanagaraj joined #gluster-dev
12:53 mohankumar joined #gluster-dev
13:05 kanagaraj joined #gluster-dev
13:10 kanagaraj joined #gluster-dev
13:19 jobewan joined #gluster-dev
13:23 ppai joined #gluster-dev
13:46 shubhendu joined #gluster-dev
13:47 hagarth joined #gluster-dev
13:47 nishanth joined #gluster-dev
13:49 pk left #gluster-dev
13:52 kanagaraj joined #gluster-dev
14:17 jclift Gah.  Cut-n-paste prob. :(
14:25 wushudoin| joined #gluster-dev
14:26 tdasilva joined #gluster-dev
14:27 wushudoin joined #gluster-dev
14:37 jclift Interesting: http://colorillo.com/b913?2djmbd
15:00 raghu left #gluster-dev
15:00 jclift *** GLUSTER COMMUNITY MEETING TIME in #gluster-meeting ***
15:01 jdarcy joined #gluster-dev
15:35 jclift ndevos: You around?
15:36 jclift ndevos: That action item about "several new xlators (encryption, cdc, changelog, prot_client, prot_server, reddir-ahead, dht) have .so.0 and .so.0.0.0 in both release-3.5 and master branch" is done isn't it
15:45 ndk joined #gluster-dev
16:31 dneary joined #gluster-dev
16:50 kkeithley_mtg 3.4.3beta2 is in the pipeline
16:56 kanagaraj joined #gluster-dev
17:05 hagarth joined #gluster-dev
17:19 purpleidea kkeithley_mtg: looking at the rpm's finally... what does this do:
17:19 purpleidea find %{_builddir} \( -name "*.rb" -o -name "*.erb" \) -exec chmod -x {} + -exec sed -i "/^#!/{d;q}" {} +
17:19 purpleidea (the sed part i mean)
17:19 purpleidea {d;q} ?
17:20 purpleidea something about changing the shebang?
17:22 purpleidea kkeithley_mtg: also, i'm going to test this and get it working if possible today... i'd like to add the spec file "upstream", and glue it all together. what's the right workflow? what commands should i be running to do this. i haven't rpm'ed in a while
17:30 kkeithley_mtg yeah, something with the shebang. copied from the other rpm spec that I used as a template. give me a couple minutes.
17:31 kkeithley_mtg not sure what you've asking about upstream and workflow. Do you want to rebuild the rpm with newer bits?
17:35 purpleidea kkeithley_mtg: i was going to add the spec file to the puppet-gluster repo...
17:35 purpleidea 2) what commands should i be running to make an rpm, etc, etc...
17:36 shyam joined #gluster-dev
17:37 kkeithley_mtg the sed strips the shebang.
17:38 purpleidea ah, this is incorrect i think
17:38 purpleidea it will definitely break the module
17:39 kkeithley_mtg to remake the rpm use `rpmbuild -ba SPECS/puppet-gluster.spec`.  This assumes you installed the src.rpm and the various files live in ~/rpmbuild/{SPECS,SOURCES}
17:39 kkeithley_mtg if you don't want the shebang stripped from .rb files then delete that from the spec
17:39 kkeithley_mtg delete that line
17:39 purpleidea right
17:40 purpleidea kkeithley_mtg: lets say i want to use my ~/code/puppet/puppet-gluster/ dir as the basis for building everything (src.rpm too) from scratch. what do i do?
17:40 kkeithley_mtg cd rpmbuild && rpmbuild -ba SPECS/puppet-gluster.spec
17:40 purpleidea should i make an rpmbuild dir in the puppet-gluster/ dir?
17:41 purpleidea (i thought it always went in ~)
17:41 kkeithley_mtg yeah, you can do that. Depending on the version of rpmbuild there are environment vars and/or command line options to build in other locations than ~/rpmbuild
17:44 kkeithley_mtg see the gluster .../extras/LinuxRPM Makefile{,.in,.am}, e.g. --define '_topdir ...'
17:44 purpleidea error: File /home/james/rpmbuild/SOURCES/puppet-gluster-e4​f46b0ec38d6ea136cd3da9d95686aa5c5ba341.tar.gz: No such file or directory
17:44 purpleidea ok
17:44 kkeithley_mtg it's sort of arcane
17:45 purpleidea haha yeah
17:45 purpleidea see i am in the process of makefile-ing this actually :P
17:45 purpleidea (the rpmbuild process)
17:45 purpleidea puppet-gluster/rpmbuild/ sound okay to you?
17:46 kkeithley_mtg for where this lives in your git source?
17:46 kkeithley_mtg if so, yeah, that's fine by me
17:46 kkeithley_mtg biab
17:48 purpleidea (okay i have a src.rpm) how to generate the real one? rpm --rebuild <file> ?
17:52 kkeithley_mtg rpmbuild -ba builds "everything".. src.rpm in SRPMS, the "real" rpm in RPMS/noarch/ probably
17:53 kkeithley_mtg -bb builds the "binary" only, i.e. not the src.rpm
17:54 purpleidea ah
17:54 purpleidea -bs (build spec)
17:54 kkeithley_mtg yes, when I built, with what I sent you, the .rpm landed in RPMS/noarch
17:54 kkeithley_mtg because BuildArch: noarch in the spec file
17:54 purpleidea i should rtfm a bit before bugging you. i normally do, but i think i've repressed learning about rpm
17:54 kkeithley_mtg no worries
17:55 purpleidea okay i'll try to send you a git branch for review by today. i'm going to add some dependencies too for puppet, i think
17:55 kkeithley_mtg okay
17:55 purpleidea and i should probably make a separate rpm for each puppet module dependency...
17:56 kkeithley_mtg that's up to you, however you want to do it.
17:56 purpleidea is %{_datadir} /usr/share/ ?
17:57 kkeithley_mtg I think so. let me confirm
17:57 purpleidea "however i want to do it" - nope - you're the rpm wizard. i'm the lowly upstream guy.
17:57 kkeithley_mtg lol
18:00 purpleidea kkeithley_mtg: i'm getting the feeling that if this is based off an rpm, i should probably add rpm generation natively into a makefile instead of pulling from github each time... any stock recommendations or versioning pattern?
18:00 purpleidea or where should i pull the version number from, for example?
18:01 purpleidea (and how to template that into the spec file for instance)
18:02 kkeithley_mtg /usr/lib/rpm/macros is where _datadir is defined, and yes, it's /usr/share
18:03 kkeithley_mtg versions can be old style -x.y.z-r  or use the git commit tag like I did in the example I gave you.
18:03 kkeithley_mtg it's up to you.
18:04 purpleidea kkeithley_mtg: right. okay, i like your commit idea... but do i need to pull the sha1 into the Version: field somehow?
18:06 kkeithley_mtg so, it's not my idea. There's a contingent in Fedora that like using the commit hash because it makes it harder for people to play games with replacing tarballs behind the scenes and that sort of thing.
18:06 kkeithley_mtg Let me see if I can find an example
18:06 purpleidea haha
18:07 purpleidea as long as i can "make rpm", it's totally automatic, and everyone is happy, then i don't care especially
18:08 kkeithley_mtg but it's not pervasive enough yet that I have any of those kinds of rpms on my boxes.  Maybe just use the commit hash for the git src tarball and use x.y.z-r like I did in the example?
18:08 kkeithley_mtg yeah, seems reasonable to me
18:08 purpleidea kkeithley_mtg: do i store the version number if a file, or pull it from a git tag, or what?
18:09 purpleidea in* a file
18:09 kkeithley_mtg oh, no.  you get to invent whatever version you want. You can do it with a tag in github
18:10 purpleidea right, but i'd like to pull this number into the .tar file name, the spec file, and anywhere else it's needed from a _single_ source...
18:11 kkeithley_mtg right. So if you set a tag then you can get it from github using the versioned tarball.
18:11 kkeithley_mtg hang on for an example
18:12 kkeithley_mtg e.g. if you set a tag like V2.0.0, you can download it from github from https://github.com/%{name}/%{name}/archive/V%{vers​ion}/%{name}-%{version}.tar.gz
18:12 jdarcy joined #gluster-dev
18:13 purpleidea i think i'll add making the tar ball a local operation so that i can hack on planes, etc...
18:13 jdarcy left #gluster-dev
18:13 purpleidea so in my makefile i'll tar -cf ... $(version) ... but where should i pull the $version from?
18:14 kkeithley_mtg hard-coded in your spec file
18:14 purpleidea :(
18:15 purpleidea can't i get the data from a single place to get used everywhere?
18:15 purpleidea even puppet-gluster/VERSION
18:15 purpleidea and in the makefile $(shell cat VERSION) for example...
18:15 purpleidea but how would i get that into the spec file?
18:16 kkeithley_mtg you could do like gluster and have a foo.spec.in, then use sed to generate the foo.spec
18:16 purpleidea oh yuck
18:17 purpleidea needing cmake to build a puppet rpm is sort of silly
18:18 kkeithley_mtg no, you don't need cmake. If you $(shell cat VERSION) in the makefile, what's wrong with running sed over a template spec file?
18:19 kkeithley_mtg or you might be able to rpmbuild --define 'version $whatever' in the Makefile
18:19 purpleidea ah okay. this i like
18:19 purpleidea kkeithley_mtg: true about sed.
18:20 purpleidea kkeithley_mtg: if i use --define 'version $foo' what do i use in the Version: field? %{version} ?
18:20 kkeithley_mtg the --define '_topdir ...' was arcane, because I couldn't find that documented.
18:22 kkeithley_mtg Version: is a way of defining version (I think, I'm guessing)  so if you use --define 'version 3.4.5' then you wouldn't have a Version: line?  not sure
18:22 kkeithley_mtg not sure if that's going to be tricky to get through Fedora package review, but these are two different things, so probably doesn't matter
18:22 kkeithley_mtg doesn't matter for now
18:23 purpleidea kkeithley_mtg: okay i'll test
18:23 purpleidea here's another one: packages generated are fc20.src ...
18:24 kkeithley_mtg yes, when you do this on a Fedora (20) box. I'd imagine you could rename it without the fc20 in the src.rpm name
18:24 purpleidea how do i know what that string is going to be in advance, and can i use this spec file for fedora _and_ centos, and i'm trying to figure how to trick the makefile into having proper dependencies so it expects the right file...
18:25 purpleidea kkeithley_mtg: ^^ yeah! if i could just remove the fc20 from the src rpm that would be best!
18:27 kkeithley_mtg I routinely take fc20.src.rpms onto RHEL and CentOS boxes. Just rename it in the makefile. I don't see a -o option or anything like it. As for getting the right behavior wrt Fedora and CentOS for dependencies, that can be tricky sometimes.
18:27 kkeithley_mtg E.g. the glusterfs.spec has a metric shitton of %if %{rhel}  and %if %{fedora} cruft.
18:28 kkeithley_mtg What are you thinking of as issues for dependencies?
18:28 purpleidea kkeithley_mtg: if i can have the rpmbuild command spit out a file that doesn't have the fc20 in there, then i can generate it directly with make, without having to run mv commands in the makefile...
18:29 purpleidea for dependencies i need to depend on 6 other puppet modules basically...
18:31 awheeler_ joined #gluster-dev
18:31 kkeithley_mtg So, I'm not aware of a command line option to rpmbuild to name the src.rpm that's produced, nothing like -o foo.
18:32 purpleidea kkeithley_mtg: any way to get the -1 release variable in?
18:32 kkeithley_mtg If the puppet modules you depend on are available in RPMs, then use a Requires: clause in the spec file
18:33 kkeithley_mtg if --define 'version 3.4.5' works then --define 'release 1' should work too
18:34 kkeithley_mtg and sadly (I just tried)  --define 'version 3.4.5' does not work.
18:35 lpabon joined #gluster-dev
18:35 purpleidea ah crap
18:35 purpleidea even with {%version} inside?
18:36 lpabon semiosis: ping
18:36 kkeithley_mtg you must have a Version: x.y.z line
18:37 purpleidea kkeithley_mtg: sed it is then!
18:37 lpabon @help
18:37 glusterbot lpabon: (help [<plugin>] [<command>]) -- This command gives a useful description of what <command> does. <plugin> is only necessary if the command is in more than one plugin. You may also want to use the 'list' command to list all available plugins and commands.
18:37 purpleidea sed it ain't so
18:37 lpabon @help list
18:37 glusterbot lpabon: (list [--private] [--unloaded] [<plugin>]) -- Lists the commands available in the given plugin. If no plugin is given, lists the public plugins available. If --private is given, lists the private plugins. If --unloaded is given, it will list available plugins that are not loaded.
18:37 lpabon @help later
18:38 glusterbot lpabon: Error: There is no command "later".
18:38 purpleidea @tell later lpabon hey
18:38 purpleidea ^ like this i think
18:38 purpleidea @tell later lpabon hey
18:38 glusterbot purpleidea: Error: I haven't seen later, I'll let you do the telling.
18:38 lpabon awesome, thanks purpleidea :-D
18:38 purpleidea @tell lpabon hey
18:38 glusterbot purpleidea: The operation succeeded.
18:38 purpleidea this: @tell lpabon hey
18:38 kkeithley_mtg @later tell lpabon hey
18:38 glusterbot kkeithley_mtg: The operation succeeded.
18:38 purpleidea @later tell lpabon hey
18:38 glusterbot purpleidea: The operation succeeded.
18:38 purpleidea or this @later tell lpabon hey
18:38 purpleidea to do it later
18:39 lpabon @tell later semiosis I do not have admin access to jenkins, you need to ask either a2 or hagarth
18:39 glusterbot lpabon: Error: I haven't seen later, I'll let you do the telling.
18:39 lpabon @later semiosis I do not have admin access to jenkins, you need to ask either a2 or hagarth
18:39 lpabon doh
18:39 purpleidea @later tell semiosis lpabon wants access to jenkins
18:39 glusterbot purpleidea: The operation succeeded.
18:40 lpabon @later tell semiosis I do not have admin access to jenkins, you need to ask either a2 or hagarth
18:40 glusterbot lpabon: The operation succeeded.
18:40 lpabon there we go
18:41 purpleidea kkeithley_mtg: what is the recommended dir name to store rpm releases in my source dir? tar/ ?
18:41 purpleidea releases/ ?
18:46 purpleidea kkeithley_mtg: o shit check this out, run: git archive -o latest.tar.bz2 HEAD inside a git tree
18:52 lalatenduM johnmark, any update from ANL guys?
18:53 kkeithley_mtg no recommendation?   I don't know anyone else that does this.
18:54 kkeithley_mtg git-archive, neat trick
18:54 purpleidea ya
18:54 purpleidea okay, i'm at the "templating the .spec file part" what commands do you recommend?
18:56 kkeithley_mtg sed -e s/__VERSION__/$(shell cat VERSION)/ -e s/__RELEASE__/$(shell cat RELEASE)/ < puppet-gluster.spec.in > puppet-gluster.spec
18:56 kkeithley_mtg or something similar
18:58 purpleidea ok
19:00 johnmark lalatenduM: not to me
19:00 purpleidea is it bad practice to store the spec.in file in the SPECS dir?
19:00 johnmark saw your note
19:00 purpleidea (which will be in git)
19:01 lalatenduM johnmark, cool
19:04 * semiosis got the message
19:04 purpleidea kkeithley_mtg: what is the correct command to turn the .src rpm into the main rpm if you only have the .src rpm?
19:05 kkeithley_mtg rpm -i $src.rpm && cd rpmbuild -ba SPECS/$spec
19:06 kkeithley_mtg if you have a tar file that contains a spec file, you can do `rpmbuild -ta $tarfile`
19:07 purpleidea kkeithley_mtg: so which way should i do it? build the tar, and then the rpm's, or build the tar, and then use that as the source for the rpms...
19:07 purpleidea (and what do i change the source0 to?)
19:07 kkeithley_mtg Source0 is the same either way
19:08 purpleidea kkeithley_mtg: but my tar's won't come from github because i'm generating them myself with git archive
19:09 kkeithley_mtg right, well, pick one way you like.   build a tar, sed the spec, rpmbuild -ba   is one.   sed the spec, build the tar,  rpmbuild -ta is the other.   Six of one AFAIC
19:10 purpleidea damn i think i've underappreciated the work of package maintainers
19:10 purpleidea and the pain involved...
19:10 semiosis paintainer*
19:10 semiosis (package maintainer)
19:10 purpleidea semiosis: right
19:11 purpleidea i used that today
19:11 purpleidea kkeithley_mtg: if i include the spec in the tar, and then do -ta does it have to have a certain path, or will rpmbuild just "find any .spec file" ?
19:11 kkeithley_mtg yeah, I really feel for Debian/Ubuntu packagers. semiosis "walked" me through it once.
19:12 purpleidea "kkeithley: why the blindfold, semiosis ?" - "semiosis: because it's scary"
19:12 kkeithley_mtg the spec file should be in the $cwd when you tar cf $foo.tar .
19:12 kkeithley_mtg lol
19:12 purpleidea kkeithley_mtg: so on one hand, it should be in rpmbuild/SPECS/ and on the other it should be in .
19:13 purpleidea my brain hurts
19:13 purpleidea i don't think i've given you enough beer to put up with my rpm issues
19:13 kkeithley_mtg wait, no...
19:14 kkeithley_mtg er, yes, okay.  If you're going to rpmbuild -ba SPECS/puppet-gluster.spec then yes. You need .../rpmbuild/SPECS and .../rpmbuild/SOURCES.    The tarball in .../rpmbuild/SOURCES and the spec in .../rpmbuild/SPECS. That's it.
19:15 semiosis debian packaging is the most convoluted, undocumented, weirdest software framework i've ever seen
19:16 purpleidea kkeithley_mtg: okay. and what changes if i want to -ta ?
19:16 purpleidea semiosis: you mean it's worse than rpm?
19:16 kkeithley_mtg way worse
19:16 semiosis i dont know rpm at all
19:16 purpleidea damn
19:16 semiosis but i believe kkeithley_mtg
19:17 purpleidea in other news, i was talking to jordan about https://github.com/jordansissel/fpm/
19:17 purpleidea but haven't tried it yet
19:18 semiosis if you want to get your packages into distros fpm wont help
19:18 semiosis distros want source packages to build using the standard tools/workflow
19:18 purpleidea oh well :(
19:33 lalatenduM johnmark, we definitely need documentation for Openstack + glusterfs in gluster.org , may be with RDO too e.g documentation should be linked to http://openstack.redhat.com/Using_G​lusterFS_for_Cinder_with_RDO_Havana
20:10 purpleidea kkeithley_mtg: still around? i'm getting near being done... the %install section doesn't work with:
20:10 purpleidea + install -d -m 0755 /home/james/code/puppet/puppet-gluster/r​pmbuild/BUILDROOT/puppet-gluster-0.0.2-1​.fc20.x86_64//usr/share/puppet/modules/
20:10 purpleidea + cp -r puppet-gluster-0.0.2 /home/james/code/puppet/puppet-gluster/rpm​build/BUILDROOT/puppet-gluster-0.0.2-1.fc2​0.x86_64//usr/share/puppet/modules/gluster
20:10 purpleidea cp: cannot stat 'puppet-gluster-0.0.2': No such file or directory
20:11 purpleidea + install -d -m 0755 /home/james/code/puppet/puppet-gluster/r​pmbuild/BUILDROOT/puppet-gluster-0.0.2-1​.fc20.x86_64//usr/share/puppet/modules/
20:13 kkeithley_mtg I'm around. I have to lead a discussion in a bit.
20:14 shyam left #gluster-dev
20:14 purpleidea okay, no worries. if you know what to do, that's great, if not, it's not a rush for me.
20:15 kkeithley_mtg looking at mine
20:17 purpleidea kkeithley_mtg: what i have now: https://github.com/purpleid​ea/puppet-gluster/tree/rpm
20:17 purpleidea still some things to fix of course
20:18 kkeithley_mtg what does `ls BUILD` show?
20:18 purpleidea ls rpmbuild/BUILD
20:18 purpleidea puppet-gluster-0.0.2/
20:18 kkeithley_mtg hmmm
20:18 purpleidea the script fails in the %install part of the spec file
20:19 purpleidea in fact, the cp -r part of the spec file
20:19 kkeithley_mtg right
20:19 kkeithley_mtg but it'
20:19 kkeithley_mtg but it's there
20:19 purpleidea when i traced the script i think it's already in that directory... but it's confusing what's correct...
20:19 purpleidea it tries to run this generate file to make the rpm: http://paste.fedoraproject.org/88948/64640139
20:19 kkeithley_mtg change the cp -r line to   "pwd && cp -r ...."
20:20 purpleidea + pwd
20:20 purpleidea /home/james/code/puppet/puppet-gluste​r/rpmbuild/BUILD/puppet-gluster-0.0.2
20:20 purpleidea + cp -r puppet-gluster-0.0.2 /home/james/code/puppet/puppet-gluster/rpm​build/BUILDROOT/puppet-gluster-0.0.2-1.fc2​0.x86_64//usr/share/puppet/modules/gluster
20:21 kkeithley_mtg your fpaste shows that it did a `cd puppet-gluster-0.0.2`
20:21 kkeithley_mtg and ^^^ confirms it.
20:21 purpleidea right
20:21 purpleidea which is wrong, but i didn't add that...
20:21 kkeithley_mtg where did that cd puppet-gluster-0.0.2 come from
20:21 purpleidea the generated /tmp/ rpm script ?
20:21 kkeithley_mtg right
20:22 purpleidea (which i fpasted)
20:22 kkeithley_mtg yup
20:22 jclift Heh, I remember .deb packaging Salasaga many years ago.  .deb is the worst system yet
20:22 jclift (for packagers)
20:22 kkeithley_mtg please fpaste your spec file
20:23 purpleidea kkeithley_mtg: https://github.com/purpleidea/puppet-g​luster/blob/rpm/puppet-gluster.spec.in
20:23 purpleidea kkeithley_mtg: oh... should the cp -r have a %{} prefix?
20:24 purpleidea kkeithley_mtg: oh... should the cp -r have a %{the_build_dir} prefix?
20:24 kkeithley_mtg no,
20:24 kkeithley_mtg it's right. same as my original. and my original works.
20:24 kkeithley_mtg let me try something
20:26 kkeithley_mtg hmmm
20:28 kkeithley_mtg where is that cd puppet-gluster-0.0.2 coming from. weird
20:30 purpleidea i do not know. the only other thing i really changed was source0 because i don't have a server path for it yet...
20:30 kkeithley_mtg actually, please fpaste the spec file that's been run through sed
20:30 purpleidea http://paste.fedoraproject.org/88959/58658171
20:30 kkeithley_mtg yeah, that change to Source0 shouldn't affect anything
20:34 purpleidea kkeithley_mtg: the only other weird thing i do, is i rename to .src rpm in the makefile to remove the fc20
20:35 kkeithley_mtg but that'd be after rpmbuild ran
20:36 kkeithley_mtg oh, nm. My rpmbuild script has cd too.
20:42 kkeithley_mtg basically mine does `cd puppet-gluster-1.0.0, cp -r puppet-gluster-$hash ....`    what's your tarball contents? Should be puppet-gluster-0.0.2/*  I bet it's not. IOW create the tar file from the parent dir of puppet-gluster-0.0.2. create tarball with `tar czf puppet-gluster-0.0.2/`
20:42 kkeithley_mtg make sense?
20:42 purpleidea ah! yeah that probably makes sense... i guess git archive doesn't include that dir! i'll poke it.
20:43 kkeithley_mtg use `git archive --prefix=....`
20:44 purpleidea kkeithley_mtg: nope, that's a real prefix to each file name :(
20:44 kkeithley_mtg oops
20:44 jclift purpleidea: Btw, initial code for spinning up instances on Rackspace via API: https://forge.gluster.org/glusterfs-rackspac​e-regression-tester/glusterfs-rackspace-regr​ession-tester/blobs/master/create_servers.py
20:44 purpleidea o wait
20:45 purpleidea --prefix works, just needs a trailing slash
20:51 purpleidea w00t... something happened. now to investigate if something good happened :)
20:55 lpabon joined #gluster-dev
20:56 kkeithley_mtg so if you run rpmbuild with --noclean you can kinda see how it would install by looking in BUILDROOT/puppet-gluster-0.0.2/....
20:59 purpleidea i'm not 100% sure i understand what you're trying to tell me (because i'm stupid)
20:59 purpleidea in other news, i think this all works!
20:59 kkeithley_mtg probably, if it didn't bail out with an error
21:00 purpleidea sweet!
21:00 purpleidea well i have to do a few more things:
21:00 purpleidea 1) actually test the rpm
21:00 purpleidea 2) add a mkdir -p rpmbuild{...} to my makefile
21:00 kkeithley_mtg where you're building, you've got a SPECS and SOURCES dir, also building should have created a BUILD and BUILDROOT dir.   If you look in the BUILDROOT/puppet-gluster-0.0.2/... you can see what it'll look like when installed.  Or just install it and see what files landed where
21:00 purpleidea 3) add all the build depends stuff, and provides, etc, and do all the other rpms
21:00 purpleidea kkeithley_mtg: right
21:01 purpleidea so i just opened the rpm with file-roller and i see that it's all there in the right folders
21:01 kkeithley_mtg excellent
21:01 purpleidea so that's good enough for me for now :)
21:01 jclift :)
21:01 jclift Git commit time
21:01 purpleidea jclift: already have the rpm branch up
21:02 purpleidea jclift: you slow!
21:02 jclift ;)
21:02 jclift Atm yeah.  A bit frain bried ;)
21:02 purpleidea i still need to add the other dependencies and misc things, so not really done yet, but yeah, close
21:02 purpleidea jclift: who is kkeithley_mtg's boss?
21:02 kkeithley_mtg rwheeler is my boss
21:03 purpleidea kkeithley_mtg: oh shit. is it bad to send him an email saying you rock?
21:03 kkeithley_mtg no need for that
21:04 kkeithley_mtg but he's in #gluster usually. Just not now
21:04 tdasilva left #gluster-dev
21:04 purpleidea you just solved my rpm woes, rpm is one of those turing-complete, not solveable things
21:04 purpleidea kkeithley_mtg: thanks again!
21:04 kkeithley_mtg yw
21:05 purpleidea time to go be a scout leader! later everyone!
21:06 jclift Have fun dude :)
21:06 kkeithley_mtg DYB
21:06 purpleidea kkeithley_mtg: haha yeah
21:07 purpleidea DOB
21:07 kkeithley_mtg +1
22:28 johnmark DOB?

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