[00:13] ibms new linux commercials are fun [00:13] hmm box hasnt died yet [00:13] maybe its just the good old gcc3 problem [00:13] yeah? like those ibm linux presentation cds, which require windos 2000? [00:13] havent seen those [00:14] but hey, we are talking about a conversion so its not unreasonable to target what you are trying to convert from. [00:14] to the business people, it looks better than having to reboot your machine to a live cd [00:16] Nick change: Medivh -> Medivh- [00:17] #!/bin/bash [00:17] if pwd | grep "/usr/src/.*linux.*" &>/dev/null; then [00:17] echo "USE gcc296 TO COMPILE KERNELS" [00:17] exit [00:17] fi [00:17] exec /usr/bin/gcc "$@" [00:17] problem solved [00:17] for the absent minded like me [00:19] hmm, you know there is a /etc/alternatives/gcc ... on most distros ;) [00:20] Nick change: Medivh- -> Medivh [00:20] right but i dont mind using gcc3 for user space [00:20] its only the kernel i want to make sure i use 296 [00:21] or are you saying i could solve this problem using that somehow? [00:21] if so, fill me in :) [00:22] hmm, okay I would probably have set gcc2.96 default, so if you actually want to compile with 3.x, you have to select it ;) [00:22] hmm true, might be a better option [00:27] <4>PU el panic: Unable to continue [00:27] be0000001008081f[0000000000000000]Bank 0: be0000001008081f at 0000000000000000<0>Bank : Machine Check08>f[0000000000000000] [00:27] Kernel panic: CPU context corrupt [00:27] <0>In idle task - not syncing [00:27] Kernel panic: CPU context corrupt [00:27] Last message repeated 1 time(s). [00:27] CPU 2: Machine Check Exception: 0000000000000004 [00:27] <4> CPU 0: Machine Check Exception: 0000000000000004 [00:27] <<0>Kernel panic: Unable to con0inue [00:28] Rebooting in 60 seconds..K [00:28] there goes the gcc3 theory [00:28] well, what kernel is that? sounds interesting [00:28] that was pax+2.4.25pre6+1.3.6+netconsole [00:28] im going to remove pax and give it another spin [00:29] maybe its just a hardware problem [00:30] could be ... [00:30] I mean context corrupt sounds like it ... [00:39] Doener_zZz (~doener@p5082DC50.dip.t-dialin.net) joined #vserver. [00:47] doener_na (~doener@pD9588B3D.dip.t-dialin.net) left irc: Ping timeout: 492 seconds [00:49] so far so good [00:49] maybe its pax [00:50] that would be a dissapointment [00:50] hmm, yes for you ;) [00:51] i like not having to worry about trivial bugs when i dont have time to upgrade :( [01:16] loger joined #vserver. [01:16] yea thats what people seem to be saying [01:16] original redhat kernel seems to be stable though [01:17] Could be that this new kernel stresses the box in different and other ways. [01:28] tanjix (~ViRu_@mastercrd.tv) left irc: [03:37] nathan_ (~nathan@209-6-130-26.c3-0.sbo-ubr1.sbo-ubr.ma.cable.rcn.com) left irc: Quit: BitchX-1.0c20cvs -- just do it. [05:47] Nick change: Bertl -> Bertl_oO [07:21] nathan_ (~nathan@209-6-130-26.c3-0.sbo-ubr1.sbo-ubr.ma.cable.rcn.com) joined #vserver. [07:50] Nick change: Bertl_oO -> Bertl [07:50] hmm, nathan_? still testing? [07:53] yes [07:53] they tested the memory [07:53] and say its fine [07:53] but they tested it from user space [07:53] id like a memtest86 [07:53] im threatening to cancel the machine unless they investigate this issue [07:53] hmm, sounds reasonable ... [07:54] theoretically, software shouldnt be able to trigger an MCE right? [07:54] not as far as I know .. [07:54] MCE is something like the famous APIC error on dual celeron ;) [07:55] god this entire thing has been a horrible disaster [07:55] i checked my bill and it has already been $500 for the two billing cycles [07:55] and the machine hasnt been been used [07:55] cant afford it anymore :/ [07:55] well, they should be glad to provide a new system ... [07:56] thats what im hoping for [07:56] because if you can proove that that machine is faulty, I guess you can sue them ... [07:58] lithium (~lithium@CPE005004cf14b9-CM013439901047.cpe.net.cable.rogers.com) joined #vserver. [07:58] hi lithium? [07:58] Hey Bertl. [07:59] what do you think of an MCE (Machine Check Exception)? [07:59] (except for bad, bad ... ;) [08:01] I hope that question isn't meant for me... [08:01] Things seem a little slow tonight. [08:01] sure it is, or do you always want the easy ones? [08:02] I've had no problems with MCE myself. As far as I recall, it reports overheating and such...so really, I've never had it do anything for me. [08:02] All my boxes are quite reliable and all. [08:02] i dont like computers [08:02] where did you get your fishingpole? [08:02] REI [08:03] and it has no mechanical failures and continues to work flawlessly [08:03] Here's my question though; why isn't the old userspace utils package for vserver marked as unsupported? [08:03] which one, and where should I mark it? [08:04] All sorts of people seem to complain about problems with it on the mailing list, meanwhile util-vserver is actually maintained and seems to work (for me anyway). [08:04] what id like to know is why ext2 wasnt originally implemented with a non-linear search algo [08:04] because nobody has more than 100 files on his floppy! [08:05] vserver-utils, on the front page for linux-vserver.org [08:05] i suppose thats a good argument [08:05] lithium: do you see any vserver tools on http://www.13thfloor.at/vserver/s_release/v1.24/ ? [08:06] or maybe http://www.13thfloor.at/vserver/d_release/v1.3.5/ ? [08:06] No, sir. [08:07] the main problem is not the information, it is with the packages (mainly debian packages :( ) [08:07] there are some gentoo/debian/wossname packages floating around with vserver 0.23 -0.29 [08:07] I suppose so. Being a Slackware user, I compile most stuff myself and don't bother with packages... [08:07] people just install them and of course, a new kernel, and then ... bang! [08:08] True. [08:08] but to be honest, I expected that ... and I also expected vserver to fall behind ... if Jack doesn't maintain it ... [08:08] the vserver folks just have to learn, that you can't just take a package, install it and be done ... [08:09] (not at the current scale, and not for so many distros) [08:09] Indeed. I had to work my ass off to get things going...I mean I could have used vskell and such, but I did it all by hand. [08:09] but I guess, you now know where to search for issues? [08:10] Sure do. [08:10] and if I break something with the next patch, you probably come back screaming, and good so! [08:10] I actually sent a patch to Enrico for util-vserver...so that it would actually work with Slackware. [08:11] These days 2.4.24-vs1.24 + util-vserver 0.27 seems to be a winning combination. [08:11] hey nathan, didn't realize that 1.3.5.2+ actually enforces some limits, good that I review those patches ;) [08:12] slackware! [08:13] Yep. [08:13] 'tis good stuff...did you end up rolling your over vserver 'build' script? [08:13] ie. is that what you contributed back to enrico? [08:14] No, but that's an idea. I sent him a patch for Slackware's init method, to call the right scripts for startup/shutdown. [08:15] Since I've never tried any of the vserver build scripts I don't really know what would work. [08:15] I ripped a few ideas off VSkell, but did it all by hand. [08:21] does anybody have a good idea, how to do systematic, automated vserver tool/feature testing? [08:25] That could dependent on exactly what you want to test... [08:37] well, whatever you have in mind?! [08:38] automated testing is always good ;) [08:41] Well, obviously something that creates a new vserver, starts it up and shuts it down may not be bad...but I believe that sort of test script exists. [08:41] It probably doesn't test anything too advanced though...just that the basic vserver works. [08:42] now how to do that for, let's say 20 vserver tool versions and a dozen kernels ;) [08:43] If only we had hot-swappable kernels... [08:46] well, basically we have, with QEMU or UML ... [08:47] That would work. I've never used any of those though, but I've read about UML... [09:21] I should really be going to bed. ttyl... [09:21] night ... [09:22] lithium (~lithium@CPE005004cf14b9-CM013439901047.cpe.net.cable.rogers.com) left irc: Quit: Leaving [09:36] Nick change: Zoiah -> _Zoiah [10:20] noel- (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [10:26] okay, have a good wossname! [10:26] nathan_: good luck! [10:26] cu all ... [10:26] Nick change: Bertl -> Bertl_zZ [10:28] noel (~noel@pD9FFA206.dip.t-dialin.net) left irc: Ping timeout: 504 seconds [12:09] mcp (~hightower@wolk-project.de) joined #vserver. [13:46] hello [14:06] Nick change: _Zoiah -> Zoiah [14:40] mhepp (~mhepp@r72s22p13.home.nbox.cz) joined #vserver. [14:56] mhepp (~mhepp@r72s22p13.home.nbox.cz) left irc: Remote host closed the connection [16:58] serving (~serving@213.186.190.24) left irc: Read error: Connection reset by peer [16:59] _shuri (~shushushu@vserver.electronicbox.net) left irc: Ping timeout: 483 seconds [17:01] _shuri (~shushushu@vserver.electronicbox.net) joined #vserver. [17:25] Nick change: Bertl_zZ -> Bertl [17:25] hi everyone! [17:33] Nick change: Doener_zZz -> Doener [17:33] hi [17:33] hi [17:36] so, how's development going? have been more than busy the past few days... [17:38] well, it's going good, stabilizing 1.2 and 1.3 seems done ... [17:38] great to hear :) [17:38] I'm slowly targeting 1.4 release .... [17:39] what would you like to have in 1.4? [17:41] of interest (but low low priority) would be a possibility to add ip adresses to a running vserver... [17:43] also more control over ressources (cpu, ram) assigned to the vservers would be nice, and a way to tell more about the ressources a vserver uses, mainly cpu would be of interest (at least to me ;) [17:44] maybe some of those are already available and i've just missed 'em ;) [17:45] hmm ...the ip stuff sounds interesting, although it is impossible with the current vserver concept ... [17:46] thought so... [17:46] mainly because the chbind ip limitation is not tied to a vserver ... [17:47] hello. [17:47] hi noel-! [17:47] Bertl: you will talk about linux-vserver here: :) http://www.linuxtag.org/2004/cfp.html [17:47] Nick change: noel- -> noel [17:48] will I ? [17:48] i've been thinking about the mechanism that iproute uses, allowing to bind more than one ip to a single interface/label, and then bind the interfaces/labels instead of the ip's itself... but probably that would cause more work than necessary and it would also break the 'classic' tools like ifconfig i guess... [17:49] Bertl: just a question if you find it interesting to hold a talk and show the audience what it is.:) [17:50] ah .. I see, maybe LinuxTag isn't that far away ;) [17:55] noel: please remind me again at the end of january ... [17:57] serving (~serving@213.186.190.24) joined #vserver. [17:57] hi all [17:57] how can util-vserver tools be uninstalled ? [17:57] serving: make uninstall [17:57] thanx [17:59] Bertl: np. I will do it. [17:59] thank you! [18:04] Doener: regarding resource control, what control over CPU do you have in mind? [18:08] limiting a single vserver to 10% cpu usage for example if other vservers are also requesting cpu and let it grab all cpu if no other vserver needs it for example... [18:08] but i guess that would mean some major changes to the scheduler... [18:09] well, that is done with the O(1) scheduler in the development series ... [18:10] okay and regarding memory? what limits/actions do you have in mind there? [18:10] ah ok, so on 2.6 such control will be possible? [18:10] well it is already in 2.4.24-ck1 ... [18:10] oh ok... [18:10] but yes 2.6 will have this too ... [18:12] mainly the same as cpu, so that you have limits in stress situations, but when no other vserver needs it then one alone may grab whatever it needs... [18:13] okay, I see ... guess that is what the RSS limits will do soon ... [18:14] cool... for now that's all i can think of ;) [18:14] good, that means we are on 'your' track ... [18:15] noel (~noel@pD9FFACA5.dip.t-dialin.net) left irc: Remote host closed the connection [18:15] noel (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [18:17] noel (~noel@pD9FFACA5.dip.t-dialin.net) left irc: Client Quit [18:18] and i guess also on the track of a lot of hosters that provide vserver's to customers... having a limiting mechanism (or other way round a way to guarantee a specific amount of ressources) is always good for those [18:18] guess so ...what it needs is some documentation ... or tools to simplify the use ... [18:22] interesting to me seems, that whenever the resource limiting topic comes up, nobody mentions I/O resources ... [18:24] Doener: how would you like a resource overview to look like? [18:25] i/o hasn't been much of a problem by now... [18:25] with overview you're referring to the last point i made? [18:25] yup [18:26] ideal case would be that top when started inside a vserver would display the cpu usage etc. of that vserver [18:27] hmm, okay, so more stealth tech, than overview right? [18:28] well... in a way... the overview part is that customers can view 'their' usage in the common way... [18:28] the real 'overview' would need some other kind of vtop i guess... [18:29] yeah, it basically depends what you 'want' to know ... [18:30] I could imagine something you could bill a customer, like the vserver-stat output gives for running processes ... [18:30] something like cpu cycles burned in vserver XY [18:31] similar to network interface accounting [18:31] that's not exactly what i'm thinking of right now, but also interesting... [18:32] well, it was an example for 'just another' perspective ;) [18:34] what i mean is that the user, while working in the vserver, can use top to see what he actually 'burns' right now, as he is probably used to this on a dedicated server... [18:34] I would really appreciate to get lots of ideas/suggestions on the mailing list (doesn't mean we implement everything, but it's always nice to know) [18:35] Doener: what about similar stuff, like the uptime? do you use that? did you test it? [18:41] hmm... i don't see what you mean... [18:42] noel (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [18:42] well, the uptime command, or uptime inside a vserver, doesn't reflect the vserver uptime, only the host uptime (in 1.2x) [18:42] noel- (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [18:42] noel- (~noel@pD9FFACA5.dip.t-dialin.net) left irc: Read error: Connection reset by peer [18:43] yupp, i know that, we emulate it using filetime from the utils... [18:44] just uptime not la of course ;) [18:44] well 1.3.x does contain a 'virtual' uptime ... [18:45] it almost made it into 1.2x, but I got some reports that some tools did behave a little odd, and no positive feedback ... so it stayed in development branch ... [18:47] guess i should set up a 1.3.x machine in the near future... [18:48] would be nice to have some feedback on devel branch ... guess this would allow to get more 'useful' features into stable, in less time ... [18:55] noel (~noel@pD9FFACA5.dip.t-dialin.net) left irc: Quit: Client exiting [19:04] okay, I'm off for a hot bath ... [19:04] Nick change: Bertl -> Bertl_oO [19:37] noel (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [19:40] more testing required? [19:46] James__ (~chatzilla@h24-71-63-164.ok.shawcable.net) joined #vserver. [19:46] hello all [19:51] noel (~noel@pD9FFACA5.dip.t-dialin.net) left irc: Remote host closed the connection [20:02] noel (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [20:10] serving (~serving@213.186.190.24) left irc: Read error: Connection reset by peer [20:17] serving (~serving@213.186.190.24) joined #vserver. [20:31] Nick change: Bertl_oO -> Bertl [20:31] hi all, had a hot bath, and will get me something to eat now ... [20:32] maharaja: testing is always welcome ... vs1.24 for eaxmple but 1.3.6pre too ... [20:32] brb ca 15 min [20:32] Nick change: Bertl -> Bertl_oO [20:52] ensc (~ircensc@ultra.csn.tu-chemnitz.de) joined #vserver. [20:52] hello [20:52] Nick change: Bertl_oO -> Bertl [20:52] hi enrico ... [20:53] how was your weekend, except for 'unproductive'? [20:54] new 'vunify' should be now working [20:54] hmm, with gcc 2.96 ;) [20:54] for this week, new initscripts are planned and then I will give the alpha branch into beta-test [20:55] no, gcc296 does not have any chance there... [20:55] SSilver2k2 (~bubba@66.176.31.110) joined #vserver. [20:55] hi SSilver2k2! [20:55] SSilver2k2 (~bubba@66.176.31.110) left irc: Client Quit [20:55] SSilver2k2 (~bubba@66.176.31.110) joined #vserver. [20:55] ensc: did you have a look at the vlimit issues? [20:56] code is cluttered enough by additional and unneeded '{ ... }' blocks [20:56] which issues? the cli interface? [20:56] well, yes, haven't checked why but it seems it doesn't communicate the limits to the kernel correctly ... [20:57] im looking at freevps, and it says its based off some stuff from linux-vserver, id rather use a ture free open source project, would i be better off sticking with vserver? [20:57] hmm, hard to tell ... [20:58] currently freevps is free (it seems so), has some features and IIRC some stability issues too ... [20:58] but it is focused on RH kernel releases 2.4.18-x.y.z [20:59] thats the thing. it seems to use an out of date kernel [20:59] im used to running .23/.24 [20:59] hmm .. vs1.3.5/6 is running on 2.4.25-pre6 ;) [21:00] bertl: what distro are you using? [21:00] Bertl: I detected a copy&paste error in src/vlimit.c, line 196 ('soft' instead of 'hard'). Can this cause your problems? [21:00] SSilver2k2: mandrake .. and a little debian [21:01] ah, im used to runnign gentoo. [21:01] ensc: very likely ... [21:01] could you send me a patch? [21:03] SSilver2k2: well I guess you won't be happy with a RH-2.4.18 kernel then ;) [21:03] Bertl: *shrusg* i dont mind switching distros as long as it works. its all linux on the inside [21:04] that is a word! [21:04] Bertl: http://savannah.nongnu.org/cgi-bin/viewcvs/util-vserver/util-vserver/src/vlimit.c.diff?r1=1.5&r2=1.6 [21:05] even switching to bsd isn't a bad thing [21:05] ensc: hey great, I like that, what do they use? [21:05] i had a major bsd vs linux-discussion last night [21:05] Bertl: who is 'they'? [21:05] savannah? [21:06] yup [21:06] the html diff thingy ... [21:06] a cgi diff-parser [21:06] this is the usual viewcvs [21:07] hmm, can this be used to view diffs too? [21:07] I mean without the source? [21:08] ??? [21:08] hmm, maybe we should move the kernel stuff really to a cvs repository ... [21:08] which source? [21:08] okay, forget it ... I'm an idiot ... [21:09] was misinterpreting what I see, kind of 'color blinded' ;) [21:10] i came to the conclusion, that something like cvs has a lot of advantages [21:11] my only reasoning against cvs (or arch) at the moment, is the amount of data ... [21:12] I'm currently using 14GB of hardlinked directories ... [21:12] yes, while porting Linux for an arm device I use CVS too. With 512 MB RAM it was fun since complete sources could be hold in cache, but on a 392 MB it was a horror... [21:13] bertl: ok, that's pretty much [21:13] and I have (just verified that) 622 kernel versions ... [21:14] most of them would be about 100k in diff ... others up to 5M ... [21:15] but I was thinking of checking in the 'broken out patches' into some cvs/arch thingy ... [21:15] do you know 'patcher'? [21:15] http://www.mn-logistik.de/unsupported/pxa250/patcher [21:16] not yet ... [21:16] really nice... [21:16] (although it is written in perl) [21:16] quilt ( http://savannah.nongnu.org/projects/quilt/) is a similar tool [21:17] SSilver2k2 (~bubba@66.176.31.110) left irc: [21:17] hmm, sounds good ... well I have some tools written myself .. to aid in the kernel patching tasks ... but I have to get a closer look at that ... [21:18] http://www.nongnu.org/quilt/ [21:34] ensc: http://vserver.13thfloor.at/Experimental/vswitch.h (last paragraph) [21:35] Bertl: what are 'flags'? [21:35] the ext2 flags? [21:35] well, I guess this should be more vserver specific ... [21:35] I had the attributes of vproc and the IUNLINK flag in mind ... [21:37] mmh, with this interface I have to do everytime 2 syscalls: first one to get old values, second one to set new ones [21:38] this is not good for two independent things (ext2 flags, and quota ctx) [21:38] well, would you prefer to set the new ones first, and return the old ones later? [21:39] look at it like vserver flags, it's not required that IUNLINK is stored in the ext2 flags, not even xattr flags ... could be acl or inode too ... [21:40] yes, but when I want to set quota-ctx, I do not need to know the old value [21:40] the 'current' implementation happens to use extended attributes for that ... [21:40] okay, what about a separate set of set/get xid only? [21:41] or one combined get and sets for each feature/set? [21:41] ok; or a set-mask [21:41] how is it done with stat info? [21:42] ?? [21:42] well stat returns lots of information, some of it can be set with chown/chgrp/utime ... [21:43] ah, you mean that get() reutrn all information but you have distinct set_XXX() methods for the attributes? [21:43] yes [21:44] ok, no problem for me, but this will add 2 new structs ;) [21:44] do we care? [21:44] would could use unions too [21:44] by the way, it will only add one at most ;) [21:45] set_fxid could be done without ... [21:45] and the flags too .. so both could be passed with dev/ino only ... [21:46] remember the id field? [21:46] ah, yes [21:47] I'll try to add the 000 as a flag too ... [21:48] (the chroot barrier ;) [21:53] Bertl: what's about a set- and an unset-mask? With a special value like '-1' for xid the distinct setting of flags and xid would be allowed [21:54] hmm, would you prefer such an interface over separated calls? [21:55] yes; at least for the set-/unset-mask. [21:56] I do not know if special values are a good idea... [21:56] # vlimit --ctx --help 100 -H --5 100 [21:56] vlimit: unrecognized option `--ctx' [21:56] Try 'vlimit --help" for more information. [21:56] # vlimit --version [21:56] vlimit 0.27.194 -- limits context-resources [21:56] This program is part of util-vserver 0.27.194 [21:57] is this fixed? [21:57] Usage: vlimit [-c|--ctx ] [-a|--all] [-MSH -- ]* [21:57] I will fix the help-text... [21:58] hmm, why the help text? [21:59] program will be 6 bytes shorter ;) [21:59] Dear Customer, [21:59] To verify if it is a hardware issue or not we are going to swap out your server for a new server. Your drives will be moved over to the new server. [21:59] hooray! [21:59] Action: nathan_ dances around [21:59] hip hip, hooray. [22:00] congrats ... [22:00] if it still crashes, who wants all my computers? ill give up. [22:01] ensc: I thought you wanted to be compatible to chcontext? [22:01] nathan_: me, me ! [22:01] ok, I will remove --ctx from chcontext too ;) [22:01] good idea ... replace it by --xid 8-) [22:02] nathan_: we'll build a monument for you, after you gave up, and donated all your computers to the vserver developers ... [22:03] you'll be remembered as 'the one who went fishing' ;) [22:04] haha that sounds good [22:05] when the box comes back and hopefully proves stable im going to look into that issue of entering the context and it failing for some reason [22:06] Bertl, while true; do vserver blah restart; done & while true; do vserver blah exec "ps"; done [22:06] Bertl, do you see any condition that would be true where the exec would fail with those two running concurrently? [22:07] ill have a more concrete example when its back up [22:08] ensc: okay, please reload ... [22:08] nathan_: hmm, don't know what vserver restart and exec really do, guess that depends on the tools ... [22:08] Bertl: what is 'mask'? [22:08] the mask for the flags ... [22:09] on read, it means: what flags are supported ... [22:09] on write, it means: change those flags ... [22:09] one flag reserved for xid ;) [22:09] Bertl, im wondering if it is just a state problem with the vserver-utils and not a chcontext issue, ill have to see what its getting for the xid [22:10] a 'set_mok [22:10] ok [22:11] nathan_: try for separate servers, that should tell .. [22:11] Bertl, what do you mean? [22:11] separate vservers? [22:12] make it foo and bar instead of blah ;) [22:13] ensc: does this interface make sense to you? [22:14] Bertl: yes, it is ok [22:17] (okay reload again) [22:18] what is with the ili flags? [22:24] hum, obviously I forgot that .. 8-) [22:31] ensc: okay, what about that? [22:33] yes, it is good [22:34] actually it should be something like 0x00FF for the vx_flags and HIDE isn't strictly a flag ... [22:34] but I guess we can 'adapt' that later ... [22:35] should be more like WEAK = 0x00, ADMIN = 0x01, WATCH = 0x02, ADMIN+WATCH = 0x03, HIDE = 0x04 [22:38] Simon (~sgarner@apollo.quattro.net.nz) joined #vserver. [22:44] Shade (shade@cpe109.bb101.cablesurf.de) joined #vserver. [22:44] Nick change: Shade -> Sh[a]de [22:44] hi [22:44] hi [22:45] Bertl: how are u? [22:45] fine thanks, how a re you? [22:45] to much work but fine too [22:45] i have a question about enricos tools [22:46] Sh[a]de: which one? [22:46] ok i have all installed with vs1.24 [22:46] in ./configure i had this msg [22:46] checking ext2fs/ext2fs.h usability... no [22:46] checking ext2fs/ext2fs.h presence... no [22:46] but one sec [22:47] you need e2fsprogs-devel, or working kernel headers [22:47] if i install "e2fslibs-dev" the message is ...yes at ext2fs.h but with error [22:47] need the error msg? [22:47] https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=112448 [22:47] e2fslibs-dev doesn't work for me either. [22:48] I have to use the linux 2.4 headers. [22:48] ok ensc... but how can i fix this error? [22:48] enrico, can we solve those issues once and for all, by doing a fallback? or have you done so already? [22:48] with the comment in buttom? [22:49] edit line 980 as suggested by comment #1 [22:49] ok line 980 of ext2fs.h ? [22:49] Bertl: new tools will have this fixed since e2fs headers are used by C prgoams only [22:49] Sh[a]de: yep [22:49] ok i test this brb [22:50] thank ensc [22:50] +s [22:50] ensc: okay, thanks ... [22:51] ensc: this one: /usr/include/ext2fs/ext2fs.h ? [22:51] Sh[a]de: yes [22:53] hmmm if i search this line: _INLINE_ errcode_t ext2fs_resize_mem(unsigned long old_size EXT2FS_ATTR((unused)), i don't find anything... [22:55] what is at line 980? [22:55] /* [22:55] ok, what is the error messagE? [22:55] sec [22:57] http://www.wwip.de/vserver/outputs/error_ext2fs.txt [22:57] suhcoolbro (~Suh@216-161-89-245.ptld.qwest.net) joined #vserver. [22:58] hi suhcoolbro! how's life? [22:58] Sh[a]de: can you look into config.log, or publish this file? [22:58] np i publish it one sec [22:59] http://www.wwip.de/vserver/outputs/config.log [23:00] mmh, they are screwing it up more and more with each version... :( [23:00] Sh[a]de: can you install a stable version of e2fsprogs-devel? [23:01] hmm [23:01] i i find one, sure [23:01] this one is from apt-get in debian [23:02] +f [23:03] in debian it calls: e2fslibs-dev [23:03] e2fsprogs-devel is not found [23:04] suhcoolbro (~Suh@216-161-89-245.ptld.qwest.net) left irc: Quit: NO CARRIER [23:07] e2fslibs-dev (1.27-2) [23:07] The headers and static libraries for ext2fs-aware tools-development. [23:08] [ Distribution: stable ] [ Section: devel ] [23:08] the only one :( [23:08] hmm, these seem to be very old. 1.34 is last stable version and it should work [23:08] ok is there a way to compile it for debian from sourcE? [23:09] the .deb only this one :/ [23:10] oh whait [23:10] Package: e2fslibs-dev (1.34+1.35-WIP-2003.12.07-2, 1.34+1.35-WIP-2003.08.21-3) [23:10] still unstable from debian [23:10] install working (non-2.6) kernel-headers and call configure as 'ac_cv_header_ext2fs_ext2fs_h=no ./configure ...' [23:13] suhcoolbro (~Suh@216-161-89-245.ptld.qwest.net) joined #vserver. [23:15] hmmm [23:15] what version u mean? [23:15] the new one or the "old" one from apt-get? [23:16] *confused* [23:17] I am completely unfamilarly with Debian and do not know which packages provides what. The kernel headers are at /usr/include/linux/ext2_fs.h [23:17] the 'ac_cv...' statements turns off usage of e2fsprogs header [23:18] ok and the statement above should fix this error? [23:18] it enforces usage of kernel headers [23:22] ok i have many unresolved dependency problems, i fix it and come back here... brbr [23:22] up till now thanks ensc [23:25] this would lead into the infinite one. gives another solution to remove the problem? [23:25] too many dependency problems about stable/unstable [23:26] so i can not install 1.34 e2fslibs [23:27] Sh[a]de: ignore e2fslibs; just install working kernel headers and use 'ac_cv...' [23:28] hmm ok and where can i get working kernel headers? (sry this is not my normaly way, this goes deep in linux. not my competence and need some help ;) [23:28] you need /usr/include/linux/ext2_fs.h [23:29] from kernel <2.5 [23:29] k [23:35] uhm now i have really problems [23:35] never install unstable in woody *g [23:35] +packs [23:35] Sh[a]de: do you need a download address for Fedora? ;) [23:35] nope *g but thanks [23:36] mc: error while loading shared libraries: libext2fs.so.2: cannot open shared object file: No such file or directory [23:36] :( [23:38] Sh[a]de: okay, enrico, please correct me, but AFAIK you do not need a working version of ext2 stuff, you only need working headers, which could be used for compiling util-vserver ... [23:38] and you could put those headers in the util-vserver package too ... [23:39] Ok I have understood this, this isn't a problem. [23:39] Bertl: I do not want to replicate this stuff [23:40] well, what I wonder is: why do the ext2 tools/utils work with those 'broken' headers? [23:41] they are in C; the 1.27 C++ issues (usage of reserved C++ words) were fixed in 1.28. But Debian unstable is unstable and ships both broken kernel headers, and CVS/BK snapshots of e2fsprogs which are broken again [23:48] ok i try to include headers from <2.5 [23:50] ozan (~ozan@dsl81-215-19222.adsl.ttnet.net.tr) joined #vserver. [23:50] hello all... [23:56] noel- (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [23:56] noel_ (~noel@pD9FFACA5.dip.t-dialin.net) joined #vserver. [23:57] noel_ (~noel@pD9FFACA5.dip.t-dialin.net) left irc: Read error: Connection reset by peer [23:57] noel (~noel@pD9FFACA5.dip.t-dialin.net) left irc: Read error: Connection reset by peer [23:59] Nick change: noel- -> noel [23:59] hi noel? [23:59] hi ozan! [00:00] --- Tue Jan 20 2004