| [2010/02/02 00:19:34] @ Log started by gepetto | ||
| [2010/02/02 00:19:34] @ Quit: notbrien: Quit: notbrien | ||
| [2010/02/02 01:12:10] <mqr> ohadlevy: Look at, yes. My only still viable suggestion (from way back when that opened) had been to use zaml to generate the yaml for such cases. It's pure ruby, faster, and much simpler. Do you have any other ideas to suggest? | ||
| [2010/02/02 01:15:02] <ohadlevy> mqr: why not store the catalog in pson? | ||
| [2010/02/02 01:15:36] <ohadlevy> mqr: the problem is only to convert the catalog back to yaml on the client side | ||
| [2010/02/02 01:18:06] <mqr> 1) Some tools expect it in yaml, 2) pson is lossy, 3) (and this may just be my curmudgeonliness) I'm not convinced pson is going to be any more reliable in the long run, 4) until you suggested it on the ticket it hadn't really been advanced as an idea. | ||
| [2010/02/02 01:18:33] <mqr> I'm not sure what you mean by "the problem is only to convert the catalog back to yaml on the client side". | ||
| [2010/02/02 01:18:54] <ohadlevy> ok, in my test scenario, i have .25.4 master and a 25.4 client | ||
| [2010/02/02 01:18:59] <ohadlevy> the client is running RHE4 | ||
| [2010/02/02 01:19:08] <ohadlevy> the catalog is complied and transfer to the client | ||
| [2010/02/02 01:19:18] <ohadlevy> the clients gets the catalog correctly, but tries to cache it on the local disk and fail | ||
| [2010/02/02 01:20:14] <ohadlevy> commenting out line #40 of indirector/yaml.rb, solves the issue | ||
| [2010/02/02 01:20:14] <gepetto> ohadlevy: #40 is http://projects.reductivelabs.com/issues/show/40 "Puppet - Bug #40: Go through the servers and eliminate use of Puppet[] for storing variables - ReductiveLabs.com" | ||
| [2010/02/02 01:21:17] <ohadlevy> catalog is transfer in pson btw | ||
| [2010/02/02 01:41:27] <ohadlevy> mqr: make sense? | ||
| [2010/02/02 01:46:38] <mqr> ohadlevy. Yes, I understand the problem completely. | ||
| [2010/02/02 01:46:52] <mqr> What's at issue is the solution. | ||
| [2010/02/02 01:48:22] <ohadlevy> mqr: I would be happy with a solution not to store the local config, I tend not to run my clients with cached catalogs anyway, as the results might be unexpected | ||
| [2010/02/02 01:49:11] <mqr> Options, as we understood them when I created that ticket: 1) support a pure ruby way to generate yaml (e.g. zaml or something like it) 2) do nothing and hope that it doesn't come up. | ||
| [2010/02/02 01:49:25] <mqr> I liked 1 but we went with 2. | ||
| [2010/02/02 01:49:40] <ohadlevy> i thought that pson mainly addressed that | ||
| [2010/02/02 01:49:47] <mqr> Option not to store is reasonable. | ||
| [2010/02/02 01:50:04] <ohadlevy> or store it in the same format you got it | ||
| [2010/02/02 01:50:41] <mqr> pson was for transfer (and has spawned it's own problems); apparently some people need/expect to stored catalogs in yaml. | ||
| [2010/02/02 01:51:14] <ohadlevy> mqr: assuming yaml doesnt break for them too... | ||
| [2010/02/02 01:51:30] <mqr> Also, pson is lossy. I still fear that will bite us some day. | ||
| [2010/02/02 01:51:51] <mqr> The extreme yaml breakage is a RHEL4 thing. | ||
| [2010/02/02 01:51:55] <ohadlevy> mqr: any chance we can detect the strings that can break yaml before? | ||
| [2010/02/02 01:52:01] <ohadlevy> or thats the wrong path again | ||
| [2010/02/02 01:54:01] <mqr> When I was doing zaml (I wrote most of the code in it) we tried to characterize the bad strings for syck (the c library used by ruby's yaml) but 1) they change with version, 2) there are a lot of them, 3) detecting them is as or more complicated than just writing your own yaml library. | ||
| [2010/02/02 01:54:26] <mqr> yaml isn't too bad of a format, but the c code version *cough* sucks. | ||
| [2010/02/02 01:55:28] <ohadlevy> mqr: as a quick hack, I'll suggest we have it configurable, e.g. run from cache catlog, store cache catlog etc | ||
| [2010/02/02 01:55:34] <ohadlevy> catalog | ||
| [2010/02/02 01:55:52] <mqr> That sounds very reasonable. | ||
| [2010/02/02 01:56:30] <ohadlevy> this will solve my problem until the next one ruby 1.81 :( | ||
| [2010/02/02 01:57:09] * ohadlevy wonders if i just recreate RHEL5 rpms for RHE4 and RHE3 | ||
| [2010/02/02 03:08:15] <masterzen> ohadlevy: mqr: the problem will be solved if we ever do #2892 which IMHO is plain unnecessary. | ||
| [2010/02/02 03:08:16] <gepetto> masterzen: ohadlevy: #2892 is http://projects.reductivelabs.com/issues/show/2892 "Puppet - Bug #2892: Remove unneeded YAML serialization when caching catalog in puppetd - ReductiveLabs.com" | ||
| [2010/02/02 03:08:59] <masterzen> ohadlevy: mqr: the whole issue with YAML as I see it is the amount of memory the current ruby extension uses to parse/generate. | ||
| [2010/02/02 03:09:13] * masterzen doing progress integrating yajl-ruby... | ||
| [2010/02/02 03:33:13] <ohadlevy> masterzen: so what would you consider the best approach ? | ||
| [2010/02/02 03:33:44] <masterzen> ohadlevy: I think we should do #2892, and definitely dump yaml :-) | ||
| [2010/02/02 03:33:44] <gepetto> masterzen: ohadlevy: #2892 is http://projects.reductivelabs.com/issues/show/2892 "Puppet - Bug #2892: Remove unneeded YAML serialization when caching catalog in puppetd - ReductiveLabs.com" | ||
| [2010/02/02 03:34:08] <masterzen> ohadlevy: there are no reason we add +200MB of mem to our puppet process for _nothing_ | ||
| [2010/02/02 03:35:41] @ Quit: Chiku: Read error: Connection reset by peer | ||
| [2010/02/02 03:35:46] @ Chiku joined channel #puppet-dev | ||
| [2010/02/02 04:06:03] <duritong> +100000 | ||
| [2010/02/02 04:07:04] <duritong> masterzen: ahh, the cause of #2982 is plain unnecessary and not the solving of #2982 | ||
| [2010/02/02 04:07:05] <gepetto> duritong: masterzen: #2982 is http://projects.reductivelabs.com/issues/show/2982 "Puppet Documentation - Feature #2982: Documentation for "filebucket" executable - ReductiveLabs.com" | ||
| [2010/02/02 04:07:05] <gepetto> duritong: masterzen: #2982 is http://projects.reductivelabs.com/issues/show/2982 "Puppet Documentation - Feature #2982: Documentation for "filebucket" executable - ReductiveLabs.com" | ||
| [2010/02/02 04:08:09] <masterzen> duritong: yes, bad wording. I actually meant, the operation of dumping to yaml is IMHO not necessary. | ||
| [2010/02/02 04:08:55] <masterzen> duritong: the issue is that mqr was saying that the localcache should stay in yaml for legacy tools... | ||
| [2010/02/02 04:10:42] <duritong> yeah | ||
| [2010/02/02 04:11:22] <duritong> and you try to keep it in yaml with your streaming approach? | ||
| [2010/02/02 04:17:16] <Volcane> cached catalog format should be configurable | ||
| [2010/02/02 04:17:24] <Volcane> so those with legacy tools can choose to suffer | ||
| [2010/02/02 04:20:10] <duritong> yeah would be the ideal option | ||
| [2010/02/02 04:26:15] <masterzen> duritong: I don't touch the cache part. It's the job of #2892. I'm only focusing on never loading the full response body in ram, only processing it one chunk at a time. | ||
| [2010/02/02 04:26:16] <gepetto> masterzen: duritong: #2892 is http://projects.reductivelabs.com/issues/show/2892 "Puppet - Bug #2892: Remove unneeded YAML serialization when caching catalog in puppetd - ReductiveLabs.com" | ||
| [2010/02/02 04:26:23] <masterzen> Volcane: sure | ||
| [2010/02/02 04:26:36] <duritong> masterzen: ok | ||
| [2010/02/02 05:18:30] @ Quit: ppickfor: Ping timeout: 252 seconds | ||
| [2010/02/02 05:23:09] @ ppickfor joined channel #puppet-dev | ||
| [2010/02/02 05:37:23] @ nigelk joined channel #puppet-dev | ||
| [2010/02/02 05:44:27] @ Quit: nigelk: Quit: nigelk | ||
| [2010/02/02 05:44:41] @ ppickfor1 joined channel #puppet-dev | ||
| [2010/02/02 05:44:41] @ Quit: ppickfor: Killed (NickServ (GHOST command used by ppickfor1!~ppickfor@76-10-184-216.dsl.teksavvy.com)) | ||
| [2010/02/02 05:44:44] @ ppickfor1 is now known as ppickfor | ||
| [2010/02/02 06:04:56] @ ppickfor1 joined channel #puppet-dev | ||
| [2010/02/02 06:04:56] @ Quit: ppickfor: Killed (NickServ (GHOST command used by ppickfor1!~ppickfor@76-10-184-14.dsl.teksavvy.com)) | ||
| [2010/02/02 06:05:00] @ ppickfor1 is now known as ppickfor | ||
| [2010/02/02 07:29:24] @ Quit: ppickfor: Ping timeout: 252 seconds | ||
| [2010/02/02 07:32:01] @ ppickfor joined channel #puppet-dev | ||
| [2010/02/02 09:10:03] @ Quit: rocket: Ping timeout: 252 seconds | ||
| [2010/02/02 09:38:54] @ rocket joined channel #puppet-dev | ||
| [2010/02/02 09:55:17] @ notbrien joined channel #puppet-dev | ||
| [2010/02/02 10:20:06] @ jmccune joined channel #puppet-dev | ||
| [2010/02/02 13:02:41] @ Quit: mqr: Ping timeout: 246 seconds | ||
| [2010/02/02 13:17:06] @ lak joined channel #puppet-dev | ||
| [2010/02/02 13:48:21] @ mqr joined channel #puppet-dev | ||
| [2010/02/02 13:54:25] @ Quit: mqr: Read error: Connection reset by peer | ||
| [2010/02/02 13:54:51] @ mqr joined channel #puppet-dev | ||
| [2010/02/02 14:01:47] @ Quit: lak: Quit: lak | ||
| [2010/02/02 14:08:13] @ nigelk joined channel #puppet-dev | ||
| [2010/02/02 14:11:01] <nigelk> hacim: yt? | ||
| [2010/02/02 14:11:25] @ mqr_ joined channel #puppet-dev | ||
| [2010/02/02 14:11:33] <hacim> hey nigelk | ||
| [2010/02/02 14:11:43] @ Quit: mqr: Read error: Connection reset by peer | ||
| [2010/02/02 14:11:45] <nigelk> hey. | ||
| [2010/02/02 14:11:55] <nigelk> so I think you're right that we're not the people to best package puppet emacs modes :) | ||
| [2010/02/02 14:12:06] <nigelk> I'm just unsure what's the least painful strategy atm | ||
| [2010/02/02 14:12:29] <hacim> yeah, me neither | ||
| [2010/02/02 14:12:35] <hacim> i've tested out the puppet-el package, and it works | ||
| [2010/02/02 14:13:50] <nigelk> were we ever supplying the emacs modes in a useful location? | ||
| [2010/02/02 14:13:57] <hacim> we were, it was working before too | ||
| [2010/02/02 14:14:02] <nigelk> ok | ||
| [2010/02/02 14:14:04] <hacim> it was part of the puppetmaster package | ||
| [2010/02/02 14:14:10] <nigelk> so perhaps we should commit to puppet-el for the lifetime of 0.25.x? | ||
| [2010/02/02 14:14:18] <nigelk> and for rowlf look to using emacs-goodies? | ||
| [2010/02/02 14:14:52] <hacim> i'm fine with that, russ is the one who does most of the work on the major-mode and it doesn't change very often | ||
| [2010/02/02 14:15:01] <nigelk> yep, my thoughts exactly | ||
| [2010/02/02 14:15:06] <nigelk> ok. will mail that to the list | ||
| [2010/02/02 14:15:15] <hacim> i was thinking maybe a suggests/recommends on the vim/emacs stuff in the other binary packages might be prudent? | ||
| [2010/02/02 14:16:06] <nigelk> I think a suggests is the right thing to do | ||
| [2010/02/02 14:16:33] <hacim> think both puppetmaster/puppet packages should have them? | ||
| [2010/02/02 14:16:38] <nigelk> hrm | ||
| [2010/02/02 14:16:58] <nigelk> maybe... recommends for vi/emacs on puppet and suggests on puppetmasterd? does that make sense? | ||
| [2010/02/02 14:17:25] <nigelk> hrm. I don't know actually. | ||
| [2010/02/02 14:18:12] <hacim> yeah, me neither :) | ||
| [2010/02/02 14:18:49] @ Caesar joined channel #puppet-dev | ||
| [2010/02/02 14:18:55] <nigelk> Hail Caesar! | ||
| [2010/02/02 14:19:02] <Caesar> pizza pizza | ||
| [2010/02/02 14:19:03] <hacim> i think maybe suggests on both would be fine | ||
| [2010/02/02 14:19:14] <nigelk> Caesar: recap | ||
| [2010/02/02 14:19:15] <nigelk> hacim: i was thinking maybe a suggests/recommends on the vim/emacs stuff in the other binary packages might be prudent? | ||
| [2010/02/02 14:19:26] <nigelk> nigelk: maybe... recommends for vi/emacs on puppet and suggests on puppetmasterd? does that make sense? | ||
| [2010/02/02 14:19:26] <nigelk> nigelk: hrm. I don't know actually. | ||
| [2010/02/02 14:19:26] <nigelk> hacim: yeah, me neither | ||
| [2010/02/02 14:19:29] <hacim> i can imagine situations where some people are going to want to have it installed on clients, some on the puppetmaster and some on their desktops where they have neither | ||
| [2010/02/02 14:19:48] <hacim> all hail! | ||
| [2010/02/02 14:20:36] <Caesar> Given Recommends are treated as Depends in most cases, if you don't want it installed by default, it has to be a Suggests | ||
| [2010/02/02 14:20:44] @ antarus joined channel #puppet-dev | ||
| [2010/02/02 14:21:17] <hacim> right | ||
| [2010/02/02 14:21:51] <nigelk> If we recommend it, will that auto-pull in vim/emacs ? | ||
| [2010/02/02 14:21:56] <nigelk> that doesn't seem desirable | ||
| [2010/02/02 14:22:26] <hacim> yeah, that is why I'm suggesting Suggests :) | ||
| [2010/02/02 14:22:44] <nigelk> ok. lets just suggests it for all binary packages | ||
| [2010/02/02 14:23:03] <nigelk> Caesar: we also kind of decided on supporting puppet-el for the lifetime of 0.25.x | ||
| [2010/02/02 14:23:14] <nigelk> and looking to move it to emacs-goodies for the next major version | ||
| [2010/02/02 14:23:21] <hacim> well... i wouldn't add the suggests to the puppet-el or puppet-vim binary packages | ||
| [2010/02/02 14:23:29] <nigelk> well no :) | ||
| [2010/02/02 14:23:49] <hacim> i hate to be exacting, but just wanted to be clear :) | ||
| [2010/02/02 14:24:04] <Caesar> nigelk: ok | ||
| [2010/02/02 14:24:11] <Caesar> What about the vim stuff? | ||
| [2010/02/02 14:24:16] <nigelk> hacim: no, it's better to be clear :) | ||
| [2010/02/02 14:24:21] <Caesar> We're going to keep that and package it correctly, right? | ||
| [2010/02/02 14:24:27] <nigelk> yes. | ||
| [2010/02/02 14:24:31] <Caesar> Right | ||
| [2010/02/02 14:24:39] @ Quit: mqr_: Ping timeout: 252 seconds | ||
| [2010/02/02 14:24:46] <hacim> assuming we know what correct is | ||
| [2010/02/02 14:25:02] <Caesar> I thought we'd figured it out? | ||
| [2010/02/02 14:25:16] <hacim> i think I got it right with the recent changes... it works and it is being done like some other packages | ||
| [2010/02/02 14:25:38] <hacim> the documentation is a bit opaque so its hard to be certain | ||
| [2010/02/02 14:28:47] <hacim> nigelk: what do you think about a suggests on the puppet-common binary? | ||
| [2010/02/02 14:29:03] <hacim> i am leaning towards no | ||
| [2010/02/02 14:29:10] <nigelk> hacim: I vote no | ||
| [2010/02/02 14:30:12] <hacim> ok, i just pushed those changes | ||
| [2010/02/02 14:31:02] <hacim> hm, this might mean NEW processing | ||
| [2010/02/02 14:31:19] @ mqr joined channel #puppet-dev | ||
| [2010/02/02 14:32:08] <Caesar> No might about it | ||
| [2010/02/02 14:32:15] <Caesar> If it's a new binary package, it'll wind up in NEW | ||
| [2010/02/02 14:32:20] * Caesar checks the state of NEW | ||
| [2010/02/02 14:32:50] <Caesar> It's not *too* bad | ||
| [2010/02/02 14:33:03] <Caesar> Could be better, has definitely been worse | ||
| [2010/02/02 14:33:18] @ jes5 joined channel #puppet-dev | ||
| [2010/02/02 14:33:38] <Volcane> well he/it claims the music annoyance is turned off | ||
| [2010/02/02 14:33:41] <Volcane> oops | ||
| [2010/02/02 14:34:24] <nigelk> he/it ? | ||
| [2010/02/02 14:34:33] <Volcane> :) | ||
| [2010/02/02 14:34:52] <Volcane> it might be a miss behaving bot :) | ||
| [2010/02/02 14:34:53] @ jes5 left channel #puppet-dev () | ||
| [2010/02/02 14:45:36] @ Quit: mqr: Ping timeout: 246 seconds | ||
| [2010/02/02 14:47:45] @ jes5 joined channel #puppet-dev | ||
| [2010/02/02 14:48:03] <nigelk> Volcane: if it were a miss behaving bot wouldn't it be she/it? | ||
| [2010/02/02 14:54:45] <hacim> nigelk: so where are we at for uploading? | ||
| [2010/02/02 14:55:25] <hacim> i can do it now, or should we wait for people to read their email? | ||
| [2010/02/02 15:25:20] <nigelk> hacim: I think we're good to go with uploading | ||
| [2010/02/02 15:28:03] <hacim> nigelk: ok, i'll do it now, unless you stop me | ||
| [2010/02/02 15:30:06] @ nigelk_ joined channel #puppet-dev | ||
| [2010/02/02 15:30:27] @ Quit: nigelk: Read error: Connection reset by peer | ||
| [2010/02/02 15:30:27] @ nigelk_ is now known as nigelk | ||
| [2010/02/02 15:49:48] @ fsweetser joined channel #puppet-dev | ||
| [2010/02/02 15:53:00] * Volcane wtf's at teyo.mail | ||
| [2010/02/02 15:53:26] <nigelk> Volcane: indeed :) | ||
| [2010/02/02 15:54:25] @ lak joined channel #puppet-dev | ||
| [2010/02/02 16:10:48] @ Quit: ppickfor: Quit: Leaving. | ||
| [2010/02/02 16:28:04] <jmccune> Is the master branch stable now that 0.25.x is out? | ||
| [2010/02/02 16:28:19] <jmccune> If so, I'll update the relevant developer documentation on the wiki | ||
| [2010/02/02 16:45:49] @ mqr joined channel #puppet-dev | ||
| [2010/02/02 16:49:20] @ mpdehaan joined channel #puppet-dev | ||
| [2010/02/02 17:05:54] <mpdehaan> ReinH, mind granting me ChanServ ops so we can have some EST coverage? :) | ||
| [2010/02/02 17:06:49] @ chadh joined channel #puppet-dev | ||
| [2010/02/02 17:06:51] <jamesturnbull> nigelk: Volcane oops Teyo | ||
| [2010/02/02 17:07:23] <nigelk> MOAR AUSTRALIANZ PLZ | ||
| [2010/02/02 17:08:32] <mpdehaan> maybe a dingo ate your config management platform | ||
| [2010/02/02 17:09:25] * tmz reads that in Elaine's voice from Seinfeld | ||
| [2010/02/02 17:09:28] <tmz> haha. | ||
| [2010/02/02 17:35:32] @ jmccune is now known as _jmccune | ||
| [2010/02/02 17:36:42] @ _jmccune is now known as jmccune_ | ||
| [2010/02/02 17:39:47] @ Quit: jmccune_: Quit: leaving | ||
| [2010/02/02 18:24:43] @ Quit: nigelk: Remote host closed the connection | ||
| [2010/02/02 18:25:15] @ nigelk joined channel #puppet-dev | ||
| [2010/02/02 18:30:13] @ jmccune joined channel #puppet-dev | ||
| [2010/02/02 18:41:06] @ Quit: jmccune: Quit: brb | ||
| [2010/02/02 18:41:39] @ jmccune joined channel #puppet-dev | ||
| [2010/02/02 18:42:24] @ jmccune left channel #puppet-dev () | ||
| [2010/02/02 18:43:43] @ jmccune joined channel #puppet-dev | ||
| [2010/02/02 18:47:28] @ Quit: jmccune: Quit: brb | ||
| [2010/02/02 18:47:51] @ jmccune joined channel #puppet-dev | ||
| [2010/02/02 18:50:06] @ jmccune is now known as jmccune_ | ||
| [2010/02/02 19:11:09] @ Quit: lak: Ping timeout: 260 seconds | ||
| [2010/02/02 19:11:40] @ lak joined channel #puppet-dev | ||
| [2010/02/02 19:55:07] @ Quit: lak: Quit: lak | ||
| [2010/02/02 20:18:18] @ Quit: mpdehaan: Ping timeout: 252 seconds | ||
| [2010/02/02 20:24:25] @ Quit: notbrien: Quit: notbrien | ||
| [2010/02/02 20:43:38] @ Quit: mqr: Quit: mqr | ||
| [2010/02/02 20:46:26] @ Quit: jes5: Quit: Leaving. | ||
| [2010/02/02 21:49:06] @ mqr joined channel #puppet-dev | ||
| [2010/02/02 21:53:49] @ Quit: mqr: Ping timeout: 264 seconds | ||
| [2010/02/02 22:06:25] @ mqr joined channel #puppet-dev | ||
| [2010/02/02 22:12:17] @ Quit: mqr: Ping timeout: 265 seconds | ||
| [2010/02/02 22:41:01] @ Quit: nigelk: Quit: nigelk | ||
| [2010/02/02 23:39:12] @ notbrien joined channel #puppet-dev |
Generated by irclog2html.py 2.6 by Marius Gedminas - find it at mg.pov.lt!