[00:02] mhepp (~mhepp@213.211.38.19) left irc: Remote host closed the connection
[00:42] pflanze (~chris@129.132.19.124) left irc: Ping timeout: 492 seconds
[01:02] Nick change: riel -> unriel
[01:31] JonB (~jon@129.142.112.33) joined #vserver.
[01:56] <Bertl> hi Jon!
[01:59] <JonB> hey Bertl
[02:03] Simon (~sgarner@210.54.177.190) joined #vserver.
[02:03] <Bertl> hi Simon!
[02:04] <Simon> hi Bertl!
[02:34] <Simon> hmm
[02:34] <Simon> I thought immutable-linkage-invert meant the file could be deleted but not modified?
[02:35] <Simon> touch test
[02:35] <Simon> setattr --immulink test
[02:35] <Simon> rm test
[02:35] <Simon> rm: cannot remove `test': Operation not permitted
[02:37] pflanze (~chris@217.162.112.177) joined #vserver.
[02:38] <Bertl> sounds strange ...
[02:38] <Simon> now I have a file I can't delete ;)
[02:38] <Bertl> please verify the attributes ... 
[02:38] <Simon> showattr test
[02:38] <Simon> test    00008000
[02:38] <Bertl> if it is 'only' immutable, this is correct ...
[02:39] <JonB> Bertl: is there someway i can prevent a vserver from stealing the cpu in stable? or can i only monitor it, and if so, how ?
[02:39] <Bertl> @jon you can use the 'sched' and 'nproc' options ...
[02:40] <Bertl> @simon only one flag, probably immutable ... no ILI flag ...
[02:40] <Simon> setattr --immutable test
[02:40] <Simon> showattr test
[02:40] <Simon> test    00000010
[02:40] <Bertl> fine now the other flag ...
[02:41] <Bertl> and then we try both ;)
[02:41] <Simon> oh ok
[02:41] <Simon> setattr --immutable --immulink test
[02:41] <Simon> showattr test
[02:41] <Simon> test    00008010
[02:41] <Simon> you need both?
[02:41] <Bertl> yeah, it's a little inverted, the logic ...
[02:42] <Simon> seems to work ;)
[02:42] <Bertl> you ahve to complain to Sam for that
[02:42] <Simon> gotcha ;)
[02:43] <Simon> is the ILI flag meant to be the same as the 't' flag with chattr?
[02:44] <Simon> chattr =it test
[02:44] <Simon> showattr test
[02:44] <Simon> test    00008010
[02:44] <Bertl> well, actually no, but sam allocated the next bit he found ...
[02:44] <Bertl> and this is the tail merge flag, iirc ...
[02:44] <Simon> yes
[02:44] <Simon> Note: As of this writing, the ext2 or ext3 filesys- 
[02:44] <Simon>        tems  do  not  (yet, except in very experimental patches) support tail-
[02:44] <Simon>        merging.
[02:45] <Simon> guess it doesn't matter then ;)
[02:45] <Bertl> maybe we will drop the ILI flag completely ...
[02:45] <Simon> oh? :(
[02:46] <Simon> how will we do unification?
[02:46] <Bertl> a little logic can make this useless, when we have context tagging ...
[02:46] <Bertl> xid==0 and more than one link means unified ;)
[02:47] <Simon> ah true
[02:50] <JonB> Bertl: i cant find explinations of sched and nproc on the linux-vserver.org website, shall i look somewhere else ?
[02:51] <Bertl> well, jack has some in-detail descriptions ...
[02:51] <Bertl> as this was added somewhere in ctx-12 or so ...
[02:53] <JonB> Bertl: okay
[02:53] <Bertl> ctx-6 is nproc ;)
[02:53] <JonB> Bertl: what is what that freevps thingy ?
[02:54] <Bertl> Alexeys version of vserver ...
[02:54] <JonB> Bertl: does he have other functionalities ?
[02:54] <Bertl> he has a lot more features ... and some stability issues ...
[02:55] <JonB> heh
[02:55] <JonB> wouldnt it be better to work together ?
[02:55] <Bertl> we work together where possible ...
[02:55] <Bertl> for example we agreed on some common userspace interfaces ...
[02:56] <Bertl> and if it makes sense, we share/exchange code ...
[02:56] <JonB> Bertl: okay
[02:57] <JonB> Bertl: anyway, thanks, nproc and sched does what i need
[02:57] <Bertl> wonderful ...
[02:58] <Bertl> any last questions for today ... before I go to bed? ;)
[02:58] <JonB> Bertl: and with the limit on VM too, that would be great
[02:58] <JonB> Bertl: how can you know all this
[02:58] <JonB> :)
[02:58] <JonB> Bertl: one last thing yes
[02:58] <JonB> is the sched a binary flag ?
[02:58] <JonB> you just set or not
[02:58] <JonB> not some value ?
[02:59] <Bertl> yup ..
[02:59] <Bertl> just a 'flag'
[02:59] <JonB> Bertl: okay
[02:59] <Bertl> like the others ;)
[03:00] <JonB> Bertl: nproc is a value, right ?
[03:00] <Bertl> nope, another 'flag'
[03:01] <JonB> okay, so if i dont like 1000 i have to edit the source ?
[03:03] <Bertl> no, this flag means that the number of processes (a ulimit) will be the maximum for the entire vserver ...
[03:03] <Bertl> this will be removed in the near future, and replaced by a rlimit syscall switch command ...
[03:04] <JonB> what is the ulimit usualy a maximum for? a user ?
[03:04] <Bertl> a task ...
[03:04] <Bertl> actually it is get/setrlimit()
[03:04] <JonB> how can a task have 1000 (ulimit) processes ?
[03:05] <Bertl> but bash and others call it ulimit ...
[03:05] <JonB> well, never mind, you need to sleep
[03:05] <JonB> and me too
[03:06] <Bertl> ulimit -a gives, max user processes, which is a feature per process ;)
[03:06] <JonB> Bertl: sleep tight, sweet dreams
[03:06] <Bertl> okay .. have a good night too ...
[03:06] JonB (~jon@129.142.112.33) left irc: Quit: zzzz
[03:07] <Bertl> cu all 2morrow ...
[03:07] <MrBawb> later Bertl
[03:07] Nick change: Bertl -> Bertl_zZ
[03:52] Simon (~sgarner@210.54.177.190) left irc: Read error: Connection reset by peer
[03:55] WSU (~WSU@ny.webpipe.net) left irc: Quit: Client Exiting
[04:31] Simon (~sgarner@apollo.quattro.net.nz) joined #vserver.
[05:10] pflanze (~chris@217.162.112.177) left irc: Ping timeout: 493 seconds
[10:24] ensc (~ircensc@134.109.116.202) left irc: Ping timeout: 492 seconds
[10:50] ensc (~ircensc@ultra.csn.tu-chemnitz.de) joined #vserver.
[11:13] say (~say@212.86.243.154) joined #vserver.
[11:51] <infowolfe> Bertl_zZ: ping?
[12:18] matta (~matta@68.81.235.145) left irc: Quit: Hey!  Where'd my controlling terminal go?
[12:32] Nick change: Bertl_zZ -> Bertl
[12:33] <Bertl> hi all!
[12:33] Simon (~sgarner@apollo.quattro.net.nz) left irc: Quit: so long, and thanks for all the fish
[12:33] <kestrel_> hey herbert, how are you?
[12:33] <Bertl> fine thanks ... cold seems almost cured ...
[12:33] <Bertl> how are you?
[12:33] <kestrel_> good to hear
[12:34] <kestrel_> i am good :)
[12:36] <kestrel_> i am playing around with the idea of writing a block device driver for 2.6.x
[12:36] <infowolfe> Bertl, how's it going?
[12:36] <Bertl> hmm, details?
[12:36] <Bertl> hi chip! thanks fine, what's your status?
[12:36] <kestrel_> it's for uml herbert :)...a distributed volume manager
[12:36] <infowolfe> my util-vserver ebuild is now assigned to gentoo's hardened herd
[12:36] <kestrel_> it may be a tad ambitious, but you have to start somewhere :)
[12:36] <Bertl> hmm, care to explain it a little?
[12:36] <infowolfe> Bertl, to who?
[12:36] <infowolfe> nevermind...
[12:36] <Bertl> if I don't add a @.. to the last message ;)
[12:36] <kestrel_> surely
[12:37] <infowolfe> Bertl, ok
[12:37] <Bertl> but you can elaborate too ...
[12:37] <kestrel_> the idea is thus...you allocate some physical devices on X number of nodes in a cluster to the dvm
[12:38] <Bertl> ok, uml in this context means?
[12:38] <kestrel_> then you create volumes out of this pool...it automatically mirrors the volumes across the network and (long term goal) will migrate blocks to the host that is currently using a block device, for performance
[12:38] <kestrel_> sorry, uml should have been vserver
[12:39] <kestrel_> i'm using uml to test develop it....i have bloody uml on the brain
[12:39] <Bertl> ahh ... 
[12:39] <Bertl> consider at least to try QEMU for development ...
[12:39] <kestrel_> so basicall, you would mount a vserver on a logical volume, do your thing
[12:40] <kestrel_> then if the real server dies, even though half the volume is lost, the other half is mounted on another host and the vserver restarted
[12:40] Action: kestrel_ goes looking for qemu
[12:40] <Bertl> hmm, and how would that 'half' volume help you?
[12:40] <kestrel_> it's a mirror
[12:41] <Bertl> uhum ... you should have a look at drdb ...
[12:41] <kestrel_> i have
[12:42] <Bertl> so? nothing to start with?
[12:42] <kestrel_> definitely, i am stealing some of the drbd networking code
[12:42] <kestrel_> there was a very good reason for not using drbd...
[12:42] Action: kestrel_ hmms
[12:43] <kestrel_> but of course, i can not think of it now
[12:43] <Bertl> well phillip is a friend of mine, he was in my software engineering group ...
[12:44] <kestrel_> well, the block migration thing is one thing
[12:44] <kestrel_> cool :)
[12:44] <Bertl> if you need details, I can arrange a contact ...
[12:44] <kestrel_> seems like he's making a nice little business off drbd
[12:44] <kestrel_> thanks for the offer, i might take you up on that
[12:45] <kestrel_> i'll need to go off and contemplate my navel for a while now
[12:45] <kestrel_> mmm, qemu
[12:45] <kestrel_> that looks like a MUCH better option
[12:45] <kestrel_> bloody uml is driving me insane
[12:47] <Bertl> well it has some advantages ...
[12:47] <Bertl> you ahve to configure it properly though ...
[12:52] <kestrel_> advantages such as?
[12:55] <Bertl> less impact on the system, more flexibility and debug options ...
[12:55] <Bertl> and you can run i386 code not only uml devices/interface ...
[12:56] <kestrel_> ah
[12:56] <Bertl> but I guess for development they are very similar ...
[12:58] <kestrel_> i bought linux device drivers (2nd ed.) the other day, it's a very interesting read
[12:58] <Bertl> yeah, but be careful, much changed in 2.4 and even more in 2.6 ..
[12:58] <kestrel_> yes, so i have noticed
[12:58] <Bertl> by the way, this book is available online, for free ;)
[12:59] <kestrel_> i found some nice articles on lwn about porting drivers to 2.6
[12:59] <kestrel_> yeah i know :)...but i can't read that on the train
[12:59] <kestrel_> there's something reassuring about books
[12:59] <Bertl> yeah, I have a lot of books twice, virtual and real ;)
[13:00] <Bertl> if you search for something, the electronic version is easier to handle ;)
[13:01] <kestrel_> definitely agree there
[13:01] <kestrel_> you're 100% right, having both is a must in my opinion
[13:01] <kestrel_> online for searching, paper for actual reading front to back
[13:02] <kestrel_> i've remembered some of my reasons for not wanting to use drbd
[13:02] <kestrel_> i wanted to be able to dynamically add nodes to the cluster and the new storage on that node would become available to the whole cluster
[13:03] <kestrel_> hence the block migration
[13:03] <Bertl> sounds good, just don't forget to make a bandwidth calculation first ...
[13:03] <kestrel_> already done
[13:04] <kestrel_> i read an article (from philip coincidentall) talking about how drbd gets near native speed even when mirroring with full sync across the network
[13:04] <kestrel_> plus i'm targetting gbit networks (8-port gbit switch in .au is only $300 AUD)
[13:05] <Bertl> how much for the nics?
[13:05] <kestrel_> $50 AUD
[13:05] <Bertl> not bad ...
[13:05] <kestrel_> exceptionally cheap
[13:05] <kestrel_> yeah
[13:05] pflanze (~chris@80.218.22.190) joined #vserver.
[13:06] <kestrel_> you could have a dedicated storage network in addition to the real network at quite a reasonable price
[13:06] <Bertl> well, most new boards have GB phy anyway ...
[13:06] <kestrel_> this is true
[13:06] <Bertl> hi chris!
[13:07] <Bertl> @chip you ahve to explain that gentoo stuff to me ...
[13:15] <Bertl> @kestrel okay, so what's your development status on that, alpha? or just planning?
[13:16] <kestrel_> planning
[13:16] <kestrel_> my current debate is whether to use the device-mapper interface or not...it appears to do a fair bit of what i want already, which is convenient
[13:17] <Bertl> probably the best solution to use it ... 
[13:18] <Bertl> you are 2.6 only or do you plan to backport later?
[13:18] <kestrel_> i'm going to concentrate on 2.6, as i suspect it will take me quite some time to get it working and by that time 2.6 should be mainline
[13:19] <kestrel_> but a lot of the implementation will be in user space, so as long as the kernel <-> userspace interface doesn't change, hopefully a backport wouldn't be too bad
[13:19] <Bertl> mainline yes, stable ... *G*
[13:19] <kestrel_> are you guys concentrating on a 2.6 port or are you sticking with 2.4 for the forseeable future?
[13:19] <kestrel_> heheh, yeah
[13:20] <kestrel_> the block device interface in 2.6 seems to be a lot cleaner than in 2.4
[13:20] <kestrel_> no global lock for one
[13:23] <Bertl> well, I have the 2.6 test patch, and as I said, if at least 3 people start 2.6 vserver testing, I'll continue porting/implementing it ...
[13:23] <Bertl> okay, brb ...
[13:25] <kestrel_> mmmm, qemu wants to statically link X11, but slackware does not have a static libX11.a
[13:25] <kestrel_> that's handy!
[13:25] <kestrel_> bit of a catch 22
[13:31] <virtuoso> G'day.
[13:33] <kestrel_> hi there
[13:33] <virtuoso> Bertl: 2.6.0-test9 worked just fine with your patch for 5 or 6 days. I played with quota tuning and overquota on it.
[13:33] <virtuoso> Bertl: Now I'm compiling -test10 with your patch.
[13:36] <Bertl> hey, cool ... 
[13:37] <Bertl> @kestrel you can disable X11 entirely, if you don't want it ...
[13:37] <Bertl> @virtuoso which patch, by the way ;)
[13:38] <virtuoso> Bertl: qh. Are there any other patches already? :)
[13:38] <Bertl> yup, there is a 'test' vserver version ...
[13:38] <Bertl> but I guess the server is currently offline, or the entire network :(
[13:39] <virtuoso> Brr. I messed up with things.
[13:39] <kestrel_> bertl: do you know how, off the top of your head?
[13:39] <Bertl> how so?
[13:39] <virtuoso> I thought I didn't find a vserver patch for 2.6 on 13thfloor..
[13:39] <virtuoso> A week ago or so.
[13:39] <Bertl> maybe it wasn't there ... a week ago or so ...
[13:40] <Bertl> @kestrel, no sorry, probably a config option ...
[13:46] <kestrel_> disabling sdl support seemed to do it
[13:50] Nick change: Bertl -> Bertl_oO
[14:18] serving (~serving@213.186.189.120) left irc: Ping timeout: 492 seconds
[15:03] virtuoso (~shisha@195.131.114.115) left irc: Quit: 2.6.0-test10
[15:12] virtuoso (~shisha@ip114-115.adsl.wplus.ru) joined #vserver.
[15:27] say_ (~say@212.86.243.154) joined #vserver.
[15:27] say (~say@212.86.243.154) left irc: Read error: Connection reset by peer
[16:09] serving (~serving@213.186.189.225) joined #vserver.
[16:41] Xorader (~xor@81.18.128.249) left irc: Quit: Leaving
[16:47] kestrel_ (~athomas@192.65.90.115) left irc: Quit: reboot
[16:47] morrigan (~morrigan@MAIL.13thfloor.at) joined #vserver.
[16:53] kestrel_ (~athomas@192.65.90.115) joined #vserver.
[17:06] kestrel_ (~athomas@192.65.90.115) left irc: Quit: Hey!  Where'd my controlling terminal go?
[17:09] morrigan (~morrigan@MAIL.13thfloor.at) left irc: Quit: ircII EPIC4-1.1.12 -- Are we there yet?
[17:09] surriel (~riel@riel.netop.oftc.net) left irc: Ping timeout: 483 seconds
[17:17] kestrel_ (~athomas@192.65.90.115) joined #vserver.
[17:19] Nick change: unriel -> riel
[17:47] morrigan (~morrigan@212.16.62.51) joined #vserver.
[17:47] morrigan (~morrigan@212.16.62.51) left irc: Client Quit
[17:48] AGoe (~agoeres@80.184.230.79) joined #vserver.
[17:48] <AGoe> hello everybody.. anybody in?
[17:49] AGoe (~agoeres@80.184.230.79) left irc: Client Quit
[17:50] <apw> hi
[18:01] Nick change: Bertl_oO -> Bertl
[18:02] <Bertl> hi .. I'm here now ;)
[18:06] <virtuoso> Bertl: Can you give me a little hint?
[18:06] <virtuoso> Bertl: I'm looking for the code in the kernel that does compression/decompression of the kernel image. Especially decompression.
[18:12] <Bertl> hmm ... you mean the initial boot loader ...
[18:13] <virtuoso> Yes.
[18:13] <Bertl> 2.4 or 2.6?
[18:13] <virtuoso> Do they differ in that?
[18:14] <Bertl> yes ...
[18:14] <virtuoso> Hmm. Better both. :)
[18:14] <Bertl> look at arch/i386/boot/*
[18:14] <virtuoso> I'm already there.
[18:15] <Bertl> compressed/misc.c on 2.6
[18:16] <Bertl> this is the same for i386 on 2.4 ...
[18:16] <virtuoso> Ok, thanks.
[18:18] <virtuoso> This guy didn't waste too much time on commenting his code. :)
[18:18] <Bertl> well, what do you expect?
[18:19] <Bertl> you should have a look at cscope if you do not use/know it already ...
[18:20] <virtuoso> I'll try it.
[18:21] <virtuoso> I usually use gview with ctags.
[19:11] pflanze (~chris@80.218.22.190) left irc: Quit: [x]chat
[19:19] riel (~riel@riel.netop.oftc.net) left irc: Ping timeout: 492 seconds
[19:34] BobR (~georg@149.148.78.14) joined #vserver.
[19:36] BobR (~georg@149.148.78.14) left irc: Client Quit
[19:37] BobR (~georg@149.148.78.14) joined #vserver.
[19:43] Nick change: BobR -> BobR_oO
[20:07] Nick change: BobR_oO -> BobR
[21:38] BobR (~georg@149.148.78.14) left irc: Ping timeout: 492 seconds
[22:15] riel (~riel@66.187.230.200) joined #vserver.
[22:40] AGoe (~agoeres@80.184.230.79) joined #vserver.
[22:40] <AGoe> hi herbert..
[22:40] <AGoe> time for some questions?
[22:57] AGoe (~agoeres@80.184.230.79) left irc: Quit: de cetero censeo aliquem necesse dormire
[23:38] micah (micah@192.101.188.235) joined #vserver.
[23:40] micah_ (~micah@micha.hampshire.edu) left irc: Ping timeout: 483 seconds
[00:00] --- Wed Nov 26 2003