[00:25] mathiaz, YES! [00:25] TY [00:26] mathiaz, I've been trying to get someone to look at mod_rails, so we can get the rails stack done [00:27] macd: right - I've finished reviewing the package [00:28] macd: I'll send my comments but overall it looks good [00:28] This one correct: http://revu.ubuntuwire.com/details.py?package=passenger [00:28] just to make sure were talking the same one [00:28] macd: It seems that mod_rails is very memory hungry [00:28] macd: yop - I've reviewed the passenger package [00:28] It only builds against a forking server [00:29] macd: correct - we do the same with php [00:29] so until its able to be built with perthread its kindve a stalemate [00:29] but so far mod_rails handles mem the best of all the other rails serving solutions [00:29] macd: according to the passenger documentation it should only take 10 lines of code [00:29] if that [00:29] its standard debian vhost, and a directory [00:30] macd: At the moment, Passenger does not support Apache with the worker MPM (which uses threads instead of processes). But because the application pool is implemented in a modular way, supporting the worker MPM shouldn't take more than 10 lines of code. [00:30] ohh, I was on another page [00:30] macd: ^^ from http://www.modrails.com/documentation/Architectural%20overview.html [00:30] Interesting, I'll talk to Neil the guy who packaged it, see if hes interested in looking into that [00:30] macd: I heard that nginx+thin is one of the best solution out there [00:31] Ive heard that as well [00:32] I've seen some benchmarks that dont show much of a diff really [00:32] http://ariekanarie.nl/archives/51/mod_rails-vs-thin-vs-ebb-vs-mongrel [00:32] thats just thin, not nginx added in [00:34] macd: hm - interesting [00:34] macd: but it seems that deploying apps under mod_rails is easier than the other solutions [00:34] mod_rails is by far the easiest to configure for sure also [00:35] its just a standard vhost [00:35] with the index set to a deeper location, so its fairly straightforward [00:35] * mathiaz agrees [00:35] Much easier to work that in, than what we were looking at with mongrels and multiple configuration changes [00:35] macd: there are some incompatibility with other apache module though [00:36] macd: the default virtual host creates an Alias for /doc, which broke my first test [00:36] macd: removing it and it worked like a charm :) [00:36] did you leave a note about that? [00:37] As far as the modules go, you can use them still on different vhosts, just not within your rails host [00:37] like mod_userdir, your rails app cant live in ~user/rails [00:37] macd: correct - that's documented in the passenger user guide [00:37] err could live in ~user/rails, but not ~user/ [00:38] I dont think thats a problem then right? [00:39] macd: oh no - not a showstopper [00:39] macd: it's documented [00:44] mathiaz, does passenger-memory-status work correctly? [00:47] err -stats [00:54] macd: in the sense that it reports the correct stats ? [00:54] macd: I don't know actually [00:54] in the sense it works at all [00:54] I get ruby errors [00:54] but Im sure my ruby is suspect, as its not from the repos [00:55] macd: oh yes - it works [00:55] The stats _are_ the actual ones, not the incorrect ps/top ones ;P [00:55] macd: I was able to get the stats of the server [00:56] macd: http://paste.ubuntu.com/29147/ [00:56] That looks about right [00:57] macd: http://paste.ubuntu.com/29148/ - that's after I've hit a simple rails app [00:57] http://blog.phusion.nl/2008/06/09/phusion-passenger-20-rc-1-and-ruby-enterprise-edition-released/ <-- that blog post seems to suggest the other MPM model is supported [00:58] cool so it works, I was just getting ruby errors, and dont have the buntu ruby installed [00:58] * macd makes a note to setup a fresh one [00:58] macd: hm - the blog post states that the worker mpm is supported now [00:59] macd: that's good news - however I'd like to make sure that the worker mpm is supported by passenger running the standard ruby vm [00:59] macd: not the EE vm [00:59] Thats what Im looking into [00:59] Im setting up a fresh vm [01:00] btw, are you running your vm in virtualbox? (I cant boot any intrepid kernels in virtualbox) [01:00] macd: I'm using kvm [01:00] Mine are running in s2k8 atm ;P [01:00] since they boot in there [01:01] macd: I ran into some issue with intrepid on kvm, using no-kvmclock to boot a 2.6.26 kernel worked [01:01] hello! [01:01] macd: but I heard that intrepid has also some issue with virtualbox [01:01] Can someone help me fix my usb [01:02] macd: I don't know how to boot them - you could try to boot with a 2.6.24 kernel [01:02] mathiaz, thats what I've been doing when I need them in there [01:09] Can someone help! I cant access my flash drive from the usb port [01:11] New bug: #250655 in samba (main) "samba daemon deadlocks" [Undecided,New] https://launchpad.net/bugs/250655 === freaky[t] is now known as fReAkY[t] === RoAk is now known as RoAkSoAx [04:58] what do you think about ubuntu?? [04:59] linux ubuntu I mean............ [04:59] hello [05:00] ya........... [05:01] :) === hessml|away is now known as hessml|away|away [07:56] soren: around? [07:57] nxvl: Yup. [07:57] soren: i have been looking for you [07:57] soren: i have work on some merges on the virtualization side [07:58] Cool. [07:58] and i suscribed you to them [07:58] have you noticed? [07:58] i think it was qemu and (not mine) bochs [07:58] I'm a bit behind on bug mail, so no. [07:58] exactly what i thought [07:59] :D [07:59] also [07:59] have you noticed that i applied for u-u-c? [08:02] soren: on the motu-council list [08:06] Yes, I did. [08:06] I'll get to it today :) [08:07] yeah, np [08:07] :D [08:07] and at last but not a least [08:08] are you technicaly familiar with the linux boot sequence? [08:08] SUre. [08:08] have you heard about the pymouth proyect? [08:09] project* [08:09] http://fedoraproject.org/wiki/Releases/FeatureBetterStartup [08:09] No. [08:10] well [08:10] pymouth is a lighter and kind of better replacement for rhgb [08:10] It seems to be "Plymouth", nor "pymouth", though :) [08:11] oh yes [08:11] my bad [08:11] :P [08:11] well, the thing is [08:11] the spanish fedora community is making eco on the news all over the internet (i have read at least 4 posts this week) about Plymouth [08:12] Ok. [08:13] the odd thing is that they are saying that the boot will be inmediat [08:13] Hm? Where does it say that? [08:13] soren: exactly [08:14] that is on all the spanish post about it [08:14] Link? [08:14] and of what i have understand [08:14] http://linux.adslzone.net/2008/07/15/fedora-10-arrancara-al-instante/ [08:15] they are just making quicker the grafical thing on the bootsequence, which is something we don't have on server for example [08:15] so the init part will still be slow [08:15] wouldn't it? [08:15] That certainly seems to be the case, yes. [08:15] actually the spanish post title are: "Fedora 10 start instantaneously" [08:16] They seem to think that the boot sequence is done when you have graphics. [08:16] Yeah. I wonder how they got that idea. [08:16] which is not at any point the scope of the project, as i have understand it [08:16] jajaj [08:17] it's kind of funny to see people that go around saying they know a lot, and are experts make mistakes like that [08:19] Yeah, it seems very odd. [08:20] actually i'm enjoing it [08:22] the last guy that makes the mistake is a guy that goes around saying how expert he is, and how much he knows and how less we other people know (and he has a special problem with me) and i have just posted on the almost same planet than him a technicaly detailed explanation on how that's NOT what it's comming [08:23] so tomorrow the flamewars will start on the peruvian community [08:23] dammit is really late [08:23] i need to go [08:23] soren: have a nice day! [08:25] You too, Nicolas. [08:25] Take care. [08:43] moin === MajorP47 is now known as Buddha === Buddha is now known as MajorP^ [09:19] can anyone help me on this http://www.howtoforge.com/how-to-install-ubuntu8.04-with-software-raid1 in the part where "make all drives bootable" im using 2 IDE my ubuntu have booted just that command i cant run [09:21] <_ruben> jmazaredo: you cant run grub? [09:41] when i type grub i get the grub probmpt [09:41] prompt [09:41] but when i try to type device (hd1) /dev/sdb [09:42] it says selected disk does not exist [09:43] i tried it inside the bash and also "before the boot up process" [09:47] <_ruben> dont think i ever used that device line .. but my guess is, that with 2 ide disks, both are master on seperate bus i hope, so they'd be sda and sdc [09:49] <_ruben> http://paste.ubuntu.com/29245/ .. thats how i do it [09:50] <_ruben> old doc tho [09:54] i only have 1 ide slot so 1 is master and 1 is slave [11:48] _ruben ty [11:49] got it working now sudo grub [11:51] i'm having this horrible desire to create full mail stack for ubuntu :D [11:52] including quarantine, per user configuration, anti spam and anti virus protection [11:53] and everything in mysql :) except mail, which would be in maildir, on the disk [11:53] I'm sure some people would find that useful. [11:54] all the parts are already there [11:54] Yep [11:54] we just need to combine it together [11:54] I prefer to keep mail config on the disk myself. [11:55] then just dump sql :D [11:55] I do :) [11:56] mail config like...? [11:56] users etc. [11:56] all the configs of services would be on disk, but per user configuration and quarentine would be in sql [11:56] ah... problem with users on disk is that you need to give someone a root account so he could add new users [11:56] <_ruben> ivoks: surely sounds like smth i'd want to use ;0 [11:56] <_ruben> ;) [11:57] with users in mysql you can just give away mysql access to a service (like postfixadmin) [11:57] True. The problem with putting them in SQL is that your mail system relies on the SQL server. [11:58] true, but sql servers aren't that unreliable... [11:59] we could even provide migration tools [12:00] now... that could be a killer-app for ubuntu-server [12:00] <_ruben> definatly [12:00] i might work on something for 9.04... [12:00] till then we will have all tools in main... thanks to ScottK === fReAkY[t] is now known as freaky[t] [12:34] is there a way to check the linux raid if it is copying in realtime like the command tail -f [12:35] What? [12:36] ivoks: I already have a setup preetty much like that. [12:37] soren: me too, but i was thinking in providing a meta package which would set up everything [12:37] ivoks: yeah. I started working a web interface for it, too, at some point, but I wanted to code it in Django, but wasn't too keen on getting Django into main. [12:39] soren: MailZu + postfixadmin :) [12:39] but much nicer interface would be welcome :) [12:40] one which would integrate mailzu and postfixadmin and roundcube [12:42] Yeah, that would be neat. [12:45] going for a walk and coffe... see you at 1300UTC [13:23] kirkland: hey, I tried out your encrypted Private folder mount last night, and everything worked as advertised... cool stuff [14:04] sommer: very nice! [14:04] sommer: thanks! [14:05] np [15:14] kirkland: what's the status of the ecryptfs testing blog post ? [15:14] mathiaz: I'm pushing a patch of 8 manpages upstream [15:15] mathiaz: I'd like to get those into intrepid first [15:15] mathiaz: I'll write the post in the meantime, and just hold off publishing until then [15:51] New bug: #229252 in openldap (main) "slapd gssapi failure" [Undecided,New] https://launchpad.net/bugs/229252 [15:54] New bug: #231321 in openldap (main) "ldap over ssl fails" [Undecided,New] https://launchpad.net/bugs/231321 [15:56] kees: does bug https://bugs.launchpad.net/ubuntu/+source/openldap2.2/+bug/249878 apply to openldap 2.2 (dapper) ? [15:56] Launchpad bug 249878 in openldap2.2 "CVE-2008-2952: BER Decoding Remote DoS Vulnerability" [Undecided,New] [15:58] <_ruben> bah .. stupid ass software raid cards .. and this pos card even has a 'bugged' kernel driver, limited its 8 ports to only 4 being usable [15:59] _ruben: hack that [16:01] <_ruben> lukehasnoname: not feeling like rolling my kernel .. recently a bug was filed against the (vanilla) 2.6.24 kernel addressing this problem (broadcom raidcore rc4000 series) .. card's crap anyways, tho was kinda curious what its performance would be under linux [16:01] * _ruben asks his boss for better hardware [16:01] 3ware's the touted linux raid card brand, right? [16:02] <_ruben> 3ware's sweet .. but also far more pricey .. broadcom/raidcore is cheap shit [16:03] 3ware rulez [16:03] not only cause of driver, but mostly cause of features and performance [16:03] and managability [16:04] <_ruben> yeah .. that is: from what i've heard .. no hands-on experience with em tho :( [16:05] 3ware works with smatmontools, 3ware provides CLI and web tools for management [16:06] New bug: #250847 in redhat-cluster-suite (main) "Apache predefined script in redhat-cluster-suite is not properly setup for Ubuntu" [Undecided,New] https://launchpad.net/bugs/250847 [16:07] oh, we can expect more bugs against rhcs :/ [16:08] meh thanks ivoks :) [16:09] i already decided to set up a testing env. back at home in fall and get this package into good shape [16:09] ivoks: you might want to talk to fabbione as well [16:10] of course [16:12] <_ruben> hmm .. wonder how 'trivial' it'd be to 'update' canonical's vmware-server packeges .. be it for gutsy's new (-15) kernel, or even hardy [16:12] <_ruben> and vmware-server itself ofcourse (1.0.4 -> 1.0.6) [16:30] isnt there a server team meeting today? [16:30] macd: Yes. [16:30] It's in #ubuntu-meeting. [16:30] k thx [16:39] If another user is logged into the server while I am also logged in, what utility can I use to send a message to that user via the terminal? [other than wall] [16:41] write [16:45] is anybody else having a "grub installation error" with intrepid alpha 2 (server) and also the daily server iso? [16:45] ahasenack: I did actually... amd64? [16:45] sommer: no, both kvm/qemu and a real P4 [16:46] ahasenack: ah, not sure... and I may have had a bad CD, but I haven't done more testing [16:46] ahasenack: did you try lilo? [16:46] sommer: no, is there a lilo option within the installer? [16:47] ahasenack: ya, once grub fails go back and there should be an option for lilo [16:47] sommer: ok, I can try that [16:47] ahasenack: what's the size of the disk you are installing on? [16:48] ivoks: 20G [16:49] ok [16:49] ivoks: and 1G inside kvm/qemu [16:50] that's ok [16:55] <_ruben> Generating locales... en_US.UTF-8... [16:55] <_ruben> that isnt supposed to take minutes or is it ? [16:55] <_ruben> on dual opteron 2ghz [16:58] sommer: lilo failed too [16:58] sommer: it's weird [16:58] ahasenack: is there any log entries in alt+f4 ? [16:58] it says "lilo is already the newest version", and then proceeds to setup grub [16:59] finally reports that dpkg returned an error code (1) while processing grub [16:59] and lilo-installer: says that calling apt-install lilo failed [16:59] ahasenack: mmmm... you might try asking in #ubuntu-installer and see if they have more ideas [16:59] I can probably attach these logs to a lp ticket, I was just checking if this was known [17:00] ahasenack: ya, a bug is probably a good idea, but folks in #ubuntu-installer may have more info [17:00] sommer: ok, thanks [17:01] ahasenack: np [17:03] ok, now is fine too [17:03] On the other hand, I'd be happy to answer (user-level) discussions about ruby gems here [17:04] persia, what is your take on the problem? [17:04] So the problem only affects gems that wants you to call them directly? [17:04] persia, were really up against a wall here trying to find some common ground fix [17:04] macd: gems are inherently broken, and if someone wants to use one, it ought be packaged. [17:04] persia, thats where Im at right now, just install gems from source and forget the debian way [17:04] but that doesnt do much to get a rails stack in ubuntu the debian way [17:05] I need about 10 minutes to go down the hall, but I'll be right back [17:05] Those are different issues, surely? [17:05] I'll be gone by then, but back tomorrow. [17:05] My opinion aside, the best fix is likely to have a README specifying that those wishing to use unpackaged gems with sudo installation ought adjust the path in /etc/environment. [17:05] well I guess I can wait [17:06] persia, cant we just do that with postinst? [17:06] nvm [17:06] opposite [17:06] No. Packages are not permitted to alter files that don't belong to them in an automated manner. [17:06] what about some automated way of building gems from the gem repositories? [17:06] and just have those going into -updates or -backports? [17:06] There is some discussion of allowing careful manipulation of other packages conffiles as proxy for user modification, but even this is technically a violation of policy. [17:07] persia, yeah debian policy is kind've holding my hands back [17:07] Well, most gems might be able to be processed by an automated packaging solution. Mind you, nobody has yet written an automated packaging solution that didn't end up breaking everything after a while. [17:07] persia, for sure [17:08] Debian policy is a good thing. It makes the system work. Without it, most things would break. [17:08] persia, I agree whole [17:08] So, long-term I'd recommend the same treatment as for CPAN or the cheeseshop: get everything interesting packaged. [17:09] the bones for a rails app gem wise, rails, mysql, or pgsql, and rails deps [17:09] This further allows control over versions: my experience with Rails is that there are usually two or three version incompatibilities in any given stack of gems one may wish, and one needs to mangle local plugin directories. [17:09] persia, yeah at least were not trying to package rmagick right ;P [17:10] Doing that at a distro level provides a much better experience to the developer. While they may not get the newest features, at least everything works, and they can concentrate on their app, rather than on rails. [17:10] persia, I tend to agree [17:10] the version of rails and gems in intrepid now are new enough to have all the real fixes [17:11] * persia did a 2-month web app system in RoR over a period of 6 months including 4 rewrites because the client wanted new features before calling it a bad day, so may be a bit bitter about this [17:11] persia, I feel your pain [17:11] I had a client halfway thtough decide they wanted rails2 b/c someone said it was "cool" [17:11] macd: Of course, my recommend solution means lots of packaging work for you :) [17:11] persia, yeah fun ;) [17:12] I think when Neil gets back from vacation we should have a quick discuss about gems again [17:12] in the meantime thanks for chiming in [17:12] Could you have a look at the mod_rails package in REVU? [17:12] there are 2 changes I know need to be made, but after those are done, we'd like to get it in [17:13] http://revu.ubuntuwire.com/details.py?package=passenger [17:13] Erm, perhaps later [17:13] thx [17:13] * persia runs away in search of a short TODO list [17:13] I've gotta get some work done too [17:55] i'm having an interesting issue: I logged in today and my LAN ip was getting dhcp instead of static even though I have it set for static! this is relevent info from syslog http://pastebin.ca/1079630 [18:09] anyone running pure-ftpd? [18:13] * delcoyote hi [18:15] i'm trying to figure out how to pass arguments to pure-ftpd at startup [18:22] kmaynard: is there a /etc/default/pure-ftpd file, or something similar ? === jpds is now known as Guest44515 [19:16] sommer, nope [19:18] hello [19:20] mathiaz, kirkland: for the grub bits of degraded raid booting, I think the review should be me, evand, cjwatson [19:20] mathiaz, kirkland: for the initramfs bits of degraded raid booting, me, luke, cjwatson [19:20] kees: cool, thanks for the clarification [19:21] kirkland: i.e. bug cjwatson last. :) [19:21] kees: good point ;-) [19:22] does ubuntu-server includes the ubuntu drivers? [19:30] does #ubuntu-chat exist? [19:40] i installed ubuntuserver and it doesn't load, just reboot and reaboot, when says grub loading... [19:40] anyone? === RoAk is now known as RoAkSoAx [21:51] hi -- anyone here had experiance with having 2 subnets on the same interface (eht0 and eth0:1) [21:56] lol [21:56] hey deeps :P [21:56] how goes real life [21:57] richer and more boring [21:57] what are you trying to do? [21:58] and eitherway, you'd be better off using vlans (and linux does understand the concept of multiple vlans on a single interface) [21:58] i have two /28s, one is working fine -- just wanted to set up my 2nd to go router->server ->wireless (allowing me to control it a bit better) [21:58] so i set up my router to have 2 subnets on one interfae (all fine) [21:59] set up eth0 and eth0:1(virtual interface) [21:59] if i do ping -I eth0 www.google.com it works, but if i do ping -I eth0:1 www.google.com it fails [21:59] i can ping both eth0 and eth0:1 from other machines on the eth0 subnet (tracert to the eth0:1 shows it hopping over the router) [22:01] could be routing, do you have a default route defined for that subnet? [22:01] each subnet has a default gateway set in the /interfaces config [22:01] other than that -- no [22:02] wget -O /dev/stdout -q -4 --bind-address=xxx.xxx.xxx.xxx http://whatismyip.org [22:02] see what comes out from both [22:02] where xxx. is the ip etc [22:02] uh huh [22:04] the eth0 one works [22:04] the eht0:1 doesn't [22:06] sounds like it could be routing related, can you ping another host in the same subnet though eth0:1 [22:06] ? [22:06] sec i find out [22:07] eyp [22:07] yep, i can ping the gateway of that subnet from it [22:07] (from eth0:1 i can ping both subnet gateways) [22:08] given that both are gateways are on the same machine [22:08] buzzons, aren't you normally supposed to set the physical interfaces to no address if you're using virtual interfaces? [22:08] and you dont have any proper segregation between them, thats to be expected [22:08] Deeps :: the gateway ip's of the router [22:08] it can also ping out of the subnet (to the other one) [22:08] so instead of having eth0 and eth0:1, you'd have eth0:0 and eth0:1 [22:09] humm.. maybe? [22:09] that could be it, otherwise my money's on it being routing related [22:09] houw would i fix the routing issue if it is that? [22:09] Deeps: i think routing issues might be why you're supposed to set it up the way i suggested [22:09] sounds logical [22:10] try it dana's way, saves me the hassle :) [22:10] route add default gw subnet1 eth0 [22:10] route add default gw subnet2 eth0 [22:10] ? [22:10] ** [22:10] route add default gw subnet2 eth0:1 [22:10] where subnet1/2 is replaced with the ip of the gateway for that subnet? [22:11] or... should i try with the eth0:0 and eth0:1 -- was only doin it the way i was due to a guide on the net [22:14] tried adding two default gw routes -- no luck on the eth0:1 [22:17] I'll try what dana said -- may work -- but can't tonight [22:17] thank you for the help [23:47] kirkland: the ssl memory leak in apache, was that only in intrepid, or was it found in all releases? [23:48] kees: hmm, let me check the bug to refresh my memory [23:48] kees: do you have the # handy? [23:48] https://issues.apache.org/bugzilla/show_bug.cgi?id=44975 [23:49] issues.apache.org bug 44975 in mod_ssl "memory leak with mod_ssl and zlib compression" [Normal,Needinfo] [23:49] https://bugs.edge.launchpad.net/ubuntu/+source/apache2/+bug/224945 [23:49] Launchpad bug 224945 in apache2 "[SRU] memory leaks in apache2 when running mod_ssl" [High,Fix released] [23:50] kirkland: you mentioned something about gutsy, but then openssl wasn't the root cause, so I wanted to double-check. [23:51] kees: ah, yes, definitely not openssl's problem [23:51] kees: absolutely a problem with Apache, that was fixed upstream [23:52] kees: and we cherry picked that fix for Hardy [23:52] kirkland: yeah. I'm just trying to understand if it was a problem with feisty/gutsy too. (a CVE was assigned for this issue, as it turns out) [23:53] kees: to be honest, I never tested Feisty/Gutsy [23:53] kees: however, if you have a KVM, it's absolutely trivial to test [23:53] kees: see https://bugs.edge.launchpad.net/ubuntu/+source/apache2/+bug/224945/comments/13 [23:53] kirkland: https://bugs.edge.launchpad.net/ubuntu/+source/apache2/+bug/224945/comments/8 says you tried dapper through gutsy? [23:53] Launchpad bug 224945 in apache2 "[SRU] memory leaks in apache2 when running mod_ssl" [High,Fix released] [23:54] kees: duh...... [23:54] kees: okay, then i believe myself :-) [23:54] kees: yeah, openssl was a red herring [23:54] kees: the problem was definitely in Apache [23:54] kirkland: but I wasn't sure if that meant you ran some openssl-specific test or the one from comment 13 [23:54] kees: all of the testing I did was with ab (apache bench) [23:55] kirkland: okay, sweet, so it sounds like gutsy needs this backported too. [23:55] this is an interesting corner-case... [23:55] I have a security-affecting fix already in -updates... and I have to do a backport to gutsy. I guess I need to pocket-copy the -updates one, and then publish the gutsy fix. [23:55] hm [23:58] woohoo.... patch sent upstream for 20 new/modified ecryptfs manpages!!! [23:59] * kirkland 's documentation duties are hereby DONE for a while ;-) [23:59] mathiaz: ping