1139270689 J * grant mep@p50919D50.dip0.t-ipconnect.de
1139271110 Q * grant_ Ping timeout: 480 seconds
1139271131 M * mugwump hmm, I see that someone else has forwarded a PID container patch
1139271264 M * mugwump Kirill
1139271364 J * derjohn ~derjohn@80.69.37.19
1139272869 M * gerrit ebiederm_oO, dhansen:  do you still want a list?  I can have one at OSDL in minutes if you have a name for it.
1139272897 M * gerrit ebiederm_oO, dhansen: I don't know if you are persuing vger as an alternative.  either works by me
1139272934 M * dhansen gerrit: it seems like the container name is sticking.  linux-containers?
1139273005 M * dhansen gerrit: I'll send some mail to solicit...
1139273024 N * Bertl_zZ Bertl
1139273028 M * Bertl morning folks!
1139273051 M * dhansen Bertl: hola
1139273092 M * Bertl hey, everything fine?
1139273187 M * cehteh moin
1139273317 M * cehteh btw, for LinuxTag how about vserver t-shirts or even better businness shirts for us?
1139273353 M * dhansen Bertl: everything is fine except the size of my inbox from all of the LKML discussion! :)
1139273368 M * Bertl lol
1139273634 M * mugwump heh, just been catching up myself
1139273641 M * mugwump I'm wondering if I need to s/vx/container/g
1139273790 M * Bertl and what about nx and dx?
1139273927 M * mugwump hmm, good point
1139273943 M * mugwump I'll leave it as-is for now
1139274774 M * gerrit dhansen:  let me know when you have some consensus - would be good to get this going soonest
1139274789 M * dhansen gerrit: yup, will keep you informed if I get any LKML responses
1139275205 Q * Smutje Remote host closed the connection
1139275219 J * Smutje ~Smutje@xdsl-87-78-84-163.netcologne.de
1139276010 M * mugwump in summary, what does the migrate syscall do?
1139276032 M * Bertl move the caller process from one context into another
1139276056 M * mugwump ok, and vx_info?
1139276069 M * mugwump ah, just return the vx_info information
1139276364 M * mugwump Bertl, I renamed __create_vx_info to create_vx_info
1139276371 M * mugwump to be consistent with the others
1139276382 M * Bertl hmm?
1139276431 M * mugwump well, free_vx_info, unhash_vx_info, etc were all not __'ed
1139276445 M * mugwump create_vx_info seemed to be one of those bunch in kernel/vserver/context.c
1139276458 M * mugwump ie, the 'internal' API used by the syscalls
1139276489 M * mugwump whereas the __'ed ones were even more internal functions used by those internal API functions
1139276561 M * Bertl well, hmm, in this case you probably want to change __loc_vx_info too, no?
1139276647 J * Aiken_ ~james@tooax6-061.dialup.optusnet.com.au
1139276691 M * mugwump ah, right.  I'm not porting any legacy sections at all yet
1139276944 Q * Aiken Ping timeout: 480 seconds
1139277147 M * Bertl mugwump: but yeah, I'm fine with that patch ...
1139277188 M * Bertl s/patch/change/
1139277211 M * mugwump cool... just putting the final touches on 1b
1139277219 N * ebiederm_oO ebiederm
1139277238 M * mugwump switch looks pretty tidy with just VCMD_get_version, VCMD_task_xid, VCMD_vx_info, VCMD_ctx_create :)
1139277349 M * Bertl uhum, well, yes ...
1139277960 M * ebiederm Hey Bertl.
1139278409 J * Smutje_ ~Smutje@xdsl-87-78-7-94.netcologne.de
1139278519 Q * Smutje Ping timeout: 480 seconds
1139278519 N * Smutje_ Smutje
1139280150 J * mkhl MiLLeNNiuM@200-148-40-42.dsl.telesp.net.br
1139280348 M * Bertl welcome mkhl!
1139280792 M * ebiederm Bertl: Did you see the patches I sent out?
1139281266 M * Bertl yup
1139281367 M * ebiederm Cool. 
1139281370 M * ebiederm Any comments/
1139281374 M * ebiederm Any comments?
1139281405 M * ebiederm I'm glad they at least made it through.
1139281596 M * Bertl well, you cc-ed me :)
1139281636 M * ebiederm So far I have gotten remarkably few comments.
1139281665 M * Bertl as usual ...
1139281677 M * ebiederm Although unfortunately the vserver list did not like my patches.
1139281719 M * Bertl from the last ten lkml postings more than the half got ignored at best
1139281730 M * Bertl s/the/my/
1139281777 M * Bertl ebiederm: btw, what is it with that one patch is the follow up of another?
1139281779 Q * Loki|muh Remote host closed the connection
1139281780 J * Loki|muh loki@satanix.de
1139281793 M * Bertl ebiederm: is that something the script does?
1139281835 M * ebiederm Bertl:  It can be.  In this case I did it manually.  It helps when patches come in a strange order to tie everything together.
1139281850 M * Bertl ebiederm: just because it looks terribly supid i almost all threaded readers/views
1139281879 M * Bertl when you do not see the subject as it hangs out 60 chars to the right
1139281897 M * ebiederm There is a point there.
1139282001 M * Bertl how does a pspace protect against accidential pid reuse of the 'wpid' if the process in the sub-space is gone?
1139282014 Q * SNy Ping timeout: 480 seconds
1139282024 M * ebiederm Bertl:  That doesn't currently happen.
1139282033 M * Bertl s/is gone/has left the pspace/
1139282108 M * Bertl how does ptrace work when transitioning into a sub pspace?
1139282160 M * Bertl lol @ alpha hack
1139282259 M * ebiederm I need to look at ptrace a little more.
1139282284 M * ebiederm I don't think there are any interesting cases but I haven't tested that case yet.
1139282333 M * ebiederm Basically it is just ptraceing a process know to the processes children as pid 1.
1139282428 M * Bertl hmm, seems your patches didn't make it to lkml.org either (yet?)
1139282517 M * ebiederm Possibly not the mirrors but I have gotten my patches back through lkml.
1139282539 M * Bertl okay, to simplify it for me, do you have a tar of your patches?
1139282592 M * ebiederm Sure I can make one up quickly.
1139282601 M * Bertl excellent, tx!
1139282617 M * ebiederm Just a moment.
1139282958 Q * zobel Ping timeout: 480 seconds
1139283264 M * ebiederm http://www.xmission.com/~ebiederm/files/pidspace-6-Feb-2006.tar.gz
1139283316 M * ebiederm Sorry for the delay.  I guess I am a little tired.
1139283316 J * zobel zobel@zobel.irc.ftbfs.de
1139283335 M * Bertl ebiederm: np, rome wasn't build on a day either :)
1139283341 M * Bertl wb zobel!
1139283363 M * ebiederm I'm glad I didn't get the contract then.  I think I would have tried :)
1139283424 M * Bertl hmm, which reminds me on the network stuff, any plans there?
1139283528 M * ebiederm Bertl:  Just as soon as the pidspace stuff settles down.
1139283578 M * Bertl hmm, settles down means?
1139283625 M * Bertl because I fear if 'settles down' == 'gets included', then it might take a year or two :)
1139283645 M * Bertl http://www.13thfloor.at/VPID/ebiederm-20060206/
1139283673 M * ebiederm Roughly when this round of the conversation dies down.
1139283721 M * ebiederm So a week or so I expect.
1139283737 M * Bertl btw, what do you think how 'guest' init should be treated?
1139283760 M * ebiederm Should be treated in which way?
1139283763 M * Bertl same as host init? better than other processes?
1139283775 M * Bertl regarding oom kill, signalling and such
1139283814 M * ebiederm Bertl:  For signalling it should be ``better'' for oom kill it should get no special considerations.
1139283840 M * Bertl in the oom kill case, the entire guest will go with it, no?
1139283845 M * ebiederm Basically guest init should get it if it is something you can write a test case for.  
1139283909 M * ebiederm Bertl:  oom kill is unlikely to pick the guest init.  And if it does killing an entire guest is actually a fairly reasonable thing.
1139283923 M * Bertl you do (in patch 3) introduce the is_init()
1139283930 M * ebiederm Yes.
1139283948 M * Bertl does it yield true for host or guest init or both?
1139283984 M * ebiederm So the expression is wid == 1.  wid is the parents idea of your pid.  Only one processes has a parent who thinks it's pid == 1.
1139284079 M * Bertl okay, so only the host init
1139284088 M * ebiederm Exactly.
1139284094 M * Bertl and the parent is the kernel init task
1139284100 M * ebiederm Yes.
1139284220 M * ebiederm Bertl:  oom_kill is like an oops.  Something that should never happen but the kernel tries to handle gracefully anyway.
1139284256 M * Bertl ah, well, I use strict-no-overcommit on all my systems, so yeah, I agree
1139284347 M * Bertl but, reality shows that a) folks do not do that (even on server systems) and b) there are new kinds of OOM in a vserver scenario
1139284395 M * ebiederm Agreed.  But there is no correct thing that can be done when you hit an oom, situation.
1139284417 M * ebiederm And disk space for swap to avoid it is fairly cheap.
1139284431 M * Bertl but there are a bunch of incorrect things you can do in a OOM case for a guest :)
1139284450 M * Bertl (especially if it is a _local_ OOM :)
1139284526 M * ebiederm When the bean counters say it the guest that is out of memory?
1139284537 M * Bertl please, we have no bean counters ...
1139284559 M * ebiederm So what is a local OOM?
1139284562 M * Bertl you are confusing VZ with Linux-VServer here 
1139284584 M * Bertl ebiederm: but yes, when the guest hits OOM on his resources
1139284745 M * ebiederm Well if it is just guest and not the entire system the guest mmap/sbrk should simply fail.
1139284770 M * Bertl huh?
1139284783 M * Bertl if that would be that simple, it would work for the host too, no?
1139284823 M * ebiederm The difference is that if it is just the guest and not the host there is still memory for the kernel to continue.
1139284870 M * Bertl the OOM happens not at mmap/sbrk time, it happens at page fault time
1139284889 M * Bertl i.e. when the page is first 'used'
1139284899 M * ebiederm Duh.
1139284908 M * ebiederm I was afraid I was forgetting a case.
1139284977 M * ebiederm I guess in that case the process should get an appropriate nasty signal.
1139285004 M * Bertl yes, but not if it is (guest) init
1139285020 M * ebiederm Now that is an interesting case.
1139285022 M * Bertl at this level, we want to pick a nasty process in the guest, except init
1139285078 M * ebiederm Hmm.  If the signal is not immediate death and init does not handle it what will happen?
1139285109 M * ebiederm I think init will just retry the page fault.
1139285125 M * ebiederm Quite possibly giving another process the opportunity to continue and be killed.
1139285144 M * ebiederm It has been a long time since I looked at that part of the mm.  I am curious now which signal that is.
1139285226 M * Bertl Vudumen: ping!
1139285264 M * ebiederm Odd.  I don't see the bit of code I recognize.
1139285361 M * ebiederm As sending the kill signal from the mm.
1139285367 M * ebiederm The closest I see is a SIG_BUS.
1139285374 M * ebiederm I wonder if rmap code killed it.
1139285395 M * ebiederm Ok. now to trace do_page_fault...
1139285396 M * Bertl         printk("VM: killing process %s\n", tsk->comm);
1139285400 M * Bertl                 do_exit(SIGKILL);
1139285407 M * ebiederm That would do it.
1139285475 M * Bertl so, guest init requests a page (being over the guest limits) and just dies :)
1139285514 M * ebiederm Unfortunate.  However that is better than the kernel panic that would usually occur.
1139285577 Q * mkhl Quit: 
1139285580 M * Bertl ebiederm: well, unfortunate and not usable for contexts ...
1139285583 M * ebiederm Which file is that in?
1139285591 M * Bertl arch/*/mm/fault.c
1139285715 M * ebiederm reading...
1139285798 M * ebiederm Is this a problem you have actually encountered?
1139285807 M * ebiederm I can certainly see the potential for mischief.
1139285932 M * Bertl well, it happens all the time when folks set memory limits (for guests)
1139285967 M * ebiederm Ok.
1139285974 M * Bertl currently guest init (in linux-vserver) is protected like the real one
1139285986 M * Bertl which is probably not the ultimate solution, IMHO
1139286041 M * ebiederm I see.
1139286099 M * ebiederm Well if I don't drop in an is_init call there the test is pid == 1.  And everything will work as it does currently.
1139286125 M * ebiederm That is something that will need to be carefully examined when the guest memory limits are merged into mainstream.
1139286159 M * Bertl btw, what code is planned for such a merge?
1139286182 M * ebiederm Good question.  
1139286212 M * ebiederm It isn't something I am likely to use so I haven't explored that yet.
1139286229 M * Bertl but you know that there will be such a merge?
1139286272 M * ebiederm I know resource limits on guest will be needed. 
1139286324 M * ebiederm Once the core code goes in it is obvious that resource limits will be an issue people will want.
1139286346 M * ebiederm I had suspected that CKRM would be it...  But I really don't know.
1139286358 M * Bertl well, ckrm didn't quite cut the mustard ... so now kernel folks are looking at 'beancounters'?
1139286400 M * Bertl which are not as flexible, but intruduce roughly the same overhead ...
1139286418 M * ebiederm What was the problem with CKRM?
1139286445 M * Bertl instability, no chance to make it work for all resources without extreme modifications
1139286490 M * ebiederm Ah.  A bad implementation.
1139286494 M * Bertl the problem there was, they simply wanted to make it so feature rich and complex ...
1139286510 M * Bertl with classes and subclasses and policy in the kernel
1139286536 M * Bertl so, this isn't gonna fly at all, limits have to be lean and simple ...
1139286558 M * ebiederm Actually for memory the first thing we need to do is to implement RLIMIT_RSS.
1139286563 M * Bertl (well, and efficient, of course)
1139286587 M * Bertl ebiederm: well, that's already there (in all vserver like projects)
1139286602 M * ebiederm Does it properly page out pages when you cross the limit?
1139286617 M * Bertl depends on the implementation
1139286632 M * Bertl the problem I see here is that this is actually not what you really want 
1139286652 M * Bertl (well, it's the poor man's approach)
1139286677 M * Bertl need an example?
1139286685 M * ebiederm Sure.
1139286696 M * Bertl consider a bunch of guests (maybe 10)
1139286711 M * Bertl each of them is limited to 128MB of RSS (in pages)
1139286720 M * ebiederm ok
1139286728 M * Bertl the 'total' memory is limited to 256MB (with swap)
1139286745 M * Bertl now you have those guests running on a 3GB machine
1139286778 M * Bertl and each guest's memory footprint is, let's say 160MB
1139286790 M * Bertl what will happen?
1139286803 M * ebiederm One moment.
1139286844 M * ebiederm About half the memory in the machine will remain usable only by the page cache.
1139286865 M * Bertl and the machine will trash like hell
1139286877 M * ebiederm All 160MB for each guest will remain in memory.
1139286900 M * Bertl context switch, new guest, 20MB swapout, next guest, 20MB swapout
1139286902 M * ebiederm But the guests will only get 128MB so they will experience a high page fault rate.
1139286931 M * Bertl so, while the host amchine still has lot of memory 'free' it will get unusable
1139286943 M * ebiederm To disk?  The pages should entire the swap cache but I doubt they would actually be swapped.
1139286959 M * ebiederm s/entire/enter/
1139287019 M * Bertl okay, s/swapout/page out/
1139287066 M * ebiederm I meant a quality implementation should not write the pages to disk after it allocated the swap pages for them.
1139287071 M * Bertl another thing which has to be considered is context tagging the pages 
1139287098 M * Bertl as you have to know which page belongs to which context ...
1139287117 M * ebiederm Which with the page cache can be a real challenge.
1139287144 M * Bertl well, you have to extend the page struct, leading to some issues
1139287206 M * ebiederm Are you talking about just anonymous memory or more than that?
1139287209 M * Bertl btw, something completely different: do you have an estimation how 'expensive' a current-> actually is nowadays?
1139287241 M * Bertl ebiederm: well, basically all pages which might get paged/swapped out
1139287245 M * ebiederm Well it would very depending on the architecture.
1139287260 M * Bertl s/very/vary/?
1139287262 M * ebiederm Bertl:  The filesystem cache?
1139287286 M * Bertl is only partially affected (i.e. for write outs)
1139287325 M * ebiederm Bertl:  What is the purpose of the limits and controls?
1139287349 M * Bertl you mean guest limits in general?
1139287375 M * ebiederm Yes.  Which problem are they trying to solve.  But especially guest memory limits.
1139287391 M * Bertl well, the limits have several purposes:
1139287405 M * Bertl - avoid various DoS
1139287417 M * Bertl - allow to have quality classes
1139287438 M * Bertl - regulate resource usage in general
1139287451 M * Bertl (the regulate includes accounting)
1139287537 M * ebiederm The DoS attacks are certainly a good excuse.
1139287568 M * Bertl well, indeed, just think memlock() :)
1139287677 M * Bertl ebiederm: ad modified find_task_by_pid()
1139287690 M * ebiederm Yes
1139287692 M * ebiederm ?
1139287731 M * Bertl I really favor this approach, but I should mention that it will break out of kernel stuff
1139287732 J * stefani ~stefani@c-24-19-46-211.hsd1.wa.comcast.net
1139287742 M * Bertl welcome stefani!
1139287751 M * stefani hola. 
1139287761 M * Bertl ebiederm: IIRC, a bunch of modules broke when we did that ...
1139287792 M * ebiederm Cool!
1139287821 M * Bertl well, I agree that those modules are probably dubious anyway ... but I just wanted to mention it
1139287828 M * ebiederm Actually that is one of the better reasons for doing it.  If modules are calling find_task_by_pid then the need to become aware of which context they are tracking pids.
1139287837 M * ebiederm Thanks.
1139287883 M * Skram how can i stop a server if someone deleted the /etc/vservers folder?
1139287888 M * Skram do i have to remake them?
1139287930 M * Bertl Skram: you can kill the entire context, but you will have to rebuild them if you want to start it again
1139287937 M * Skram nope
1139287963 M * Bertl so just figure the xid with vserver-stat and vkill the context
1139287966 M * Skram i need to stop it
1139287967 M * Skram okay
1139287976 M * Skram 10008   16  56.6M   5.6M   5m37s91   1m44s76   8d01h16 
1139287980 M * Skram it shows blank
1139287986 M * Skram well for the NAME
1139287987 M * Bertl that's probably the one :)
1139287990 M * Skram so the context is 10008
1139287991 M * Skram yeah
1139287993 M * Skram I know it is
1139288010 M * Skram so how can I stop it without crashing the server,, and remaking the /etc/vserver/name?
1139288020 M * Bertl vkill --xid -s 9 -- 0
1139288027 M * Bertl vkill --xid -s 9 -- 1
1139288031 M * Bertl oops
1139288037 M * Skram i need to look up the usage then i guess.
1139288038 M * Bertl vkill --xid 10008 -s 9 -- 1
1139288041 M * Skram okay cool ;)
1139288049 M * Bertl try first with '0'
1139288076 M * Skram there we go
1139288077 M * Skram thanks
1139288083 M * Bertl you're welcome!
1139288098 M * Skram yey.
1139288100 M * Skram ;)
1139288105 M * Skram now... 
1139288111 M * Skram to chaneg the name of a vserver..
1139288120 M * Bertl which name?
1139288127 M * Skram i just have to "mv" the /etc/vserver/name and /vserver/name ?
1139288133 M * Skram when its down of course
1139288144 M * Skram the name vserver-linux sees it as (NOT the hostname)
1139288147 M * Bertl so the name used on the host for starting
1139288159 M * Skram right
1139288184 M * Bertl yes, basically moving should be fine, but you have to adjust the symlinks too
1139288199 M * Skram oh yes.
1139288204 M * Skram run and vdir
1139288210 M * Bertl yep, precisely
1139288211 M * Skram i cant do this while its running, right?
1139288226 M * Bertl well, you can, but it might give unexpected results
1139288242 M * Skram :(
1139288253 M * Bertl but maybe it works quite fine, you have to test
1139288286 M * Bertl ebiederm: btw, do you work/talk with the other virtualization projects too?
1139288312 M * ebiederm Bertl:  This is as far as I have gotten so far.
1139288334 M * ebiederm I have some email and other correspondence with the IBS guys.
1139288350 M * ebiederm s/IBS/IBM/
1139288373 M * ebiederm I don't seem to be on good terrms yet with the OpenVz folks.
1139288379 M * Skram hmm
1139288395 M * Bertl okay, well, I certainly appreciate that you 'work' with us ...
1139288400 M * Skram so what do they do? have they come out with tools for linux-vserver
1139288421 M * Bertl Skram: hmm, no, I don't think so ...
1139288430 M * ebiederm I have a preference for true community projects which I don't yet percevie OpenVz or IBM are at the moment.
1139288461 M * ebiederm At the same time I'm not trying to discriminate.
1139288483 M * Skram What do they do?
1139288484 M * Skram hah
1139288520 M * Bertl they released a simplified version of the Virtuozzo product for beta testing into the public (as open source)
1139288539 M * Skram oh
1139288545 M * Skram to be an alt for VServer?
1139288561 M * Bertl while they claim that the kernel is identical, they also say that VZ is much more powrful than the OpenVZ version
1139288590 M * Bertl yes, basically they do the same as we do, give or take a few features
1139288590 M * Skram im confused
1139288608 M * Bertl the difference is more the intention behind the projects
1139288681 M * Skram yeah
1139288691 M * ebiederm They may be comming around.
1139288764 M * ebiederm Kirill Korotaev posted some patches today but he didn't CC me, just the IBM guys.  I have requested he not do that again.
1139288815 M * Bertl well, if you followed the marketing campain they did recently (for their brand new OVZ product^Wproject)
1139288842 M * Bertl and their plans to 'finally' get VZ into the kernel ...
1139288982 M * ebiederm Rik van Riel's reply was good.  He (like I suspect most of the other kernel developers) is waiting for the technical debate between the various approaches so we can figure out what to merge into the kernel.
1139289009 M * Bertl anyway ... just wanted to say that I appreciate your hanging around here ...
1139289014 M * ebiederm Welcome.
1139289111 M * ebiederm I will let the current debat happen on lkml and then I think we will need to establish a new mailling list.
1139289138 M * Bertl osdl would be nice ...
1139289148 M * Bertl well vger too
1139289323 M * ebiederm I will look into it.  But as I actually haven't had any complaints except the vserver list I'm not going to worry about it until we can agree on the basic form of an implementation.
1139289419 M * ebiederm Bertl:  Can you think of a better name than is_init?  is_protected?  is_killable?  Hmm.  That last one has potential.
1139289632 M * ebiederm is_immortal?
1139289644 M * Skram man
1139289650 M * Skram someone deleted /var/db
1139289669 M * ebiederm ouch!
1139289749 M * Skram yeah
1139289751 M * Skram how can i remake?
1139289756 M * Skram just emerge wolrd?
1139289762 M * Skram i really need to get this figured out
1139289774 M * Skram its a production server and im gonna kill the admin
1139289805 M * ebiederm What is usually kept in /var/db?  
1139289811 M * ebiederm It isn't a directory I have on my machine.
1139289881 M * Bertl ebiederm: well, do you want to use it just for host init?
1139289892 M * Skram ebiederm: gentoo stuff
1139289893 M * Skram ;)
1139289922 M * Bertl ebiederm: because imho the is_init is quite accurate ... but it might need a is_super_init() in your case
1139289940 M * Bertl (or maybe is_kernel_init()
1139289970 M * ebiederm Bertl:  That might work.  What I was aiming for was the property I cared about.  Instead of infering that property from the fact it was the init process.
1139290029 M * Skram emergeing gzcc ahhhh
1139290037 M * Bertl ebiederm: do you plan to provide deltas to your patches at a certain point?
1139290048 M * Skram *gcc
1139290068 M * ebiederm Bertl: I might.  Mostly I plan on having a dimminishing small patchset until everything is included.
1139290115 M * Bertl well, the thing is, I toy with the idea of putting your stuff into the next devel release (well after the next one actually)
1139290122 M * ebiederm I was working on that earlier and but since we were having the global discussion and then Linus started chiming in I figured I needed to display the whole picture.
1139290157 M * ebiederm How often do you have a devel release?
1139290168 M * Bertl so, I wonder if that will lead to never ending catching up or if you plan to provide updates ...
1139290185 M * Bertl devel releases are once a month at most
1139290207 M * Bertl but the experimental ones can happen twice a day or more
1139290209 M * ebiederm Ok.  So you are talking febuaray or march?
1139290236 M * ebiederm Truthfully I intend to have pieces in mm by then.
1139290237 M * Bertl well, the next devel will be soon, probably this or next week
1139290301 M * Bertl so the question is more, if I put in your 20060206 patches, do I have to tear them out on the next release?
1139290306 M * ebiederm Truthfully a lot hinges on where the discussion on the kernel list goes in the next several days.
1139290319 M * Bertl (your next release that is)
1139290333 M * mugwump yeah, where's the benevolant dictator leadership these days
1139290355 M * ebiederm My general intention is just to refine and forward port my patches. 
1139290399 M * ebiederm mugwump:  The benevolent dictator said we do this in the kernel with (context, id) tuples what more could you want?
1139290407 M * Bertl ebiederm: so .. let's put it this way, if I decide to do that, would you provide deltas between your changes?
1139290474 M * ebiederm Bertl:  But the problem is that the direction could completely change in the next couple of days.
1139290507 M * ebiederm So please hold off on your decision for just a little bit.
1139290560 M * ebiederm In the practical sense before anyone starts including my patches I want to walk through them one last time, and remove my one known bug.
1139290587 M * ebiederm I actually intend my patches to be so simple that you can just grab the functionality and not worry about tracking deltas.
1139290609 M * ebiederm But again.  The next couple days where we decide which direction stuff gets into the kernel are crucial.
1139290669 M * Bertl okay, fair enough ...
1139290738 M * ebiederm Truthfully for the devel release you are targetting if all goes well you might not have a choice but to incoporate some of my patches :)
1139290750 M * ebiederm That is my goal!
1139290796 M * Bertl well, you know, there is still patch -R :)
1139290850 M * Bertl again, something different but related: do you plan to 'maintain' the virtualization pieces in mainline?
1139290872 M * Bertl i.e. do you plan to become a 'virtualization subsystem' maintainer?
1139290886 M * ebiederm At least generally.
1139290908 M * ebiederm I don't actually expect those pieces to really need much maintenance once they are finally merged.
1139290930 M * ebiederm I certainly don't plan to allow something that gets into mainline to bitrot.
1139290945 M * Bertl hmm, are you talking about vpid or virtualization in general?
1139290993 M * Bertl I mean, it takes quite some time to maintain all the pieces distributed over the kernel
1139290994 M * ebiederm I am talking fairly generally here.
1139291024 M * Bertl (and I guess I know what I'm talking about here :)
1139291032 M * ebiederm Bertl:  The maintenance time should only be because the code has not yet been merged.
1139291055 M * ebiederm If the code needs active maintenance (once merged) the code needs to be refactored.
1139291100 M * Bertl so you assume that with putting it into mainline, it will just start to work and get used?
1139291140 M * ebiederm Well I hope the user base will pick up.
1139291161 M * ebiederm What I assume is that it will be a bug for anyone to submit patches that will break it once it is merged.
1139291171 M * Bertl user base being the various vortualization projects, yes?
1139291176 M * Bertl *virt*
1139291197 M * ebiederm The users of the various virtualization projects yes.
1139291209 M * Bertl ah, so the end users ...
1139291221 M * jkl man, i'm in big fcuk in trouble
1139291222 M * jkl hehe
1139291228 M * Bertl ebiederm: hmm, for them, new tools will be required, no?
1139291236 M * Bertl jkl: hmm, why?
1139291249 M * jkl broke something, as usual
1139291254 M * jkl this one is pretty bad though
1139291256 M * ebiederm Bertl:  Yes.  But hopefully for the last time.
1139291297 M * Bertl who is going to write them, btw?
1139291304 M * jkl '/sbin/runscript' is segfaulting
1139291310 M * jkl can't starting a damned thing
1139291316 M * jkl *start
1139291331 M * Bertl library issues? or just broken binary?
1139291361 M * jkl havent figured that out yet
1139291378 M * ebiederm Bertl: If I am lucky who is going to fixup the user space tools won't be my problem. :)
1139291392 M * jkl im having trouble reading through strace
1139291396 M * ebiederm I suspect my case is sufficient different from the virtualization projects that I won't be able to share tools.
1139291419 M * ebiederm Plus my current test tool is only a couple hundred line wrapper around sys_clone so I don't expect things to be very difficult.
1139291424 M * Bertl case means your 'actual work'
1139291441 M * jkl hm
1139291447 M * ebiederm Bertl:  Correct.
1139291450 M * jkl  /dev/unrandom
1139291542 M * Bertl unrandom looks interesting :)
1139291577 M * ebiederm I wonder if that is a symlink to /dev/zero I can't think of anything less random!
1139291624 M * Bertl /dev/full?
1139291648 M * ebiederm mmm.  That was a good meal.
1139291752 M * jkl actually urandom
1139291764 M * jkl some init.d scripts got wiped out during an upgrade
1139291765 M * ebiederm Anyway after the namespaces are accepted into the kernel I will be focusing on migration.
1139292333 M * jkl hm, what should i do about this?
1139292335 M * jkl access("/etc/ld.so.preload", R_OK)      = -1 ENOENT (No such file or directory)
1139292352 M * Bertl that's fine .. no bug no issue
1139292368 M * jkl it is?
1139292386 M * mugwump hmm, how do I put a bzImage inside a qemu disk?
1139292410 M * Bertl jkl: yes, it's one of the hooks of your libc allowing to hack stuff ...
1139292428 M * Bertl mugwump: on a partition, but why would you want to?
1139292459 M * mugwump I guess my question is, I've got a custom kernel and a qemu image with a demo kernel built into it...
1139292461 M * jkl i can't figure out why this is segfaulting
1139292468 M * mugwump and I want to boot my custom kernel not the demo one
1139292492 M * Bertl mugwump: just specify the custom kernel on your command line
1139292507 M * mugwump aha
1139292509 M * Bertl jkl: gdb?
1139292519 M * jkl strace no good?
1139292543 M * jkl here is strace, if you get a second http://pastebin.com/542817
1139292588 M * jkl no gdb :(
1139292680 M * Bertl well, segfault seems to happen in your executable
1139292719 M * jkl it's a part of baselayout-vserver
1139292737 M * mugwump can qemu boot amd64?
1139292744 M * Bertl mugwump: yes
1139292766 M * Bertl mugwump: but not efi stuff
1139292796 M * mugwump efi?
1139292818 J * Dr4g ~Dr4g@82-40-43-86.cable.ubr06.uddi.blueyonder.co.uk
1139292822 M * jkl ugh, it segfaults on other vservers too
1139292825 M * jkl thats not the problem
1139292826 M * jkl dam
1139292861 M * jkl what the heck did hollow do to baselayout
1139292862 M * mugwump Bertl, do you have a working amd64 qemu .config handy?  I'm missing kernel options
1139292895 M * Bertl mugwump: not at the moment, my test machine is unreachable since a few days ...
1139292983 M * Dr4g hey Bertl :)
1139292997 M * Bertl hey Dr4g!
1139293067 J * balbir ~balbir@59.145.136.1
1139293079 M * Bertl welcome balbir!
1139293149 P * stefani parting (is such sweet sorrow)
1139293161 Q * andrew_ Ping timeout: 480 seconds
1139293308 M * ebiederm Well it has been a long day.  Have a good night folks.
1139293315 N * ebiederm ebiederm_zZ
1139293538 M * Hollow jkl: fyi: runscript.c has never been changed for baselayout-vserver...  it's exactly the same as in normal baselayout... so what the heck damn shit fck
1139293704 M * jkl heh
1139293711 M * jkl whats up
1139293716 M * jkl i have to go to a backup
1139293721 M * jkl i have no idea what happened
1139293723 M * jkl im clueless
1139293751 M * jkl _nothing_ excpet ssh starts 
1139293760 M * mugwump Is openvz based on virtual pids?
1139293774 M * Hollow jkl: which version of baselayout is it?
1139293812 M * jkl the most recent one in portage as of yesterday
1139293827 M * jkl it might not even be the problem
1139293828 M * Hollow arch or ~arch?
1139293831 M * jkl arch
1139293855 M * Hollow hm
1139293893 M * jkl 1.12.0_pre8-r2
1139293906 M * Hollow ah, that's ~arch
1139293911 M * jkl ugh
1139293914 M * jkl how did i manage that
1139293956 M * Hollow well, it works for my vservers since half a year now..
1139293978 M * jkl im sure it does
1139293987 M * jkl otherwise you'd be hearing from a lot more people
1139294002 M * Hollow jkl: so, when does it segfault?
1139294014 M * jkl well, runscript segfaults
1139294016 M * anonc Bertl: quick question - has anything changed regarding syslog facilities between 2.0 and 2.1? I use metalog (patched so as not to use the klog component) which runs fine on 2.6.12.4-vs2.0 but the same executable does log anything when running on 2.6.15-vs2.1.0.4 Interestingly syslog-ng work fine without any issues
1139294018 M * jkl but that's normal apparently
1139294036 M * Hollow jkl: yeah, if you run it just from the command line, it will segfault.. but you're not supposed to do so ;)
1139294074 M * jkl yeah
1139294083 M * mugwump dammit, why is partition table support so easy to disable
1139294086 M * Bertl anonc: klog was virtualized, but what do you mean with 'does log anything'?
1139294092 M * jkl it's trange why all the sudden nothing can start
1139294098 M * jkl i get OK 's across the board
1139294100 M * jkl but nothing is there
1139294109 M * jkl ive tried zapping everything with no luck
1139294131 M * jkl i never had the symlinks for a lot of the dummy files in /etc/init.d
1139294142 M * jkl it was part of etc-update
1139294152 M * Hollow ic.. seems like you updated from a very old version..?
1139294158 M * jkl must have
1139294173 M * jkl this is the first one i built
1139294185 M * jkl i had ~80 conf files to update
1139294190 M * Hollow i assume you have init style = gentoo then?
1139294196 M * jkl i have tried both
1139294201 M * jkl plain and gentoo
1139294230 M * anonc Bertl: strangely enough nothing gets written by programs which should be writing to syslog (eg slapd) however using 'logger' to send messages to it results in them being written properly - also init being started and stopped also gets logged successfully. I suspect user error on my part but was just wondering what had changed ( I recall the klog part being mentioned before)
1139294267 M * Hollow ok, you can skip gentoo, it's obsolete.. can you try to run "rm -rf /var/lib/init.d" in the vserver and try again
1139294296 M * jkl ok, i was afraid to do that
1139294324 M * Hollow afraid?
1139294330 M * jkl ha, yeah
1139294331 M * Hollow it won't do you any harm ;)
1139294339 M * jkl should i do that while it's stopped?
1139294355 M * Hollow hm, yeah.. but if no service got started, it probably doesn't matter *g*
1139294475 M * jkl well
1139294481 M * jkl i see everything trying to start 
1139294490 M * jkl but all that remains after it's done is sshd
1139294509 M * jkl and it thinks everything is started correctly
1139294510 A * mugwump sighs at his not-booting qemu
1139294526 M * Hollow jkl: ok, can you start the services manually?
1139294529 M * Bertl mugwump: is it not booting or just not displaying anything?
1139294532 M * Hollow btw..
1139294539 M * Hollow jkl: how do you see everything starting?
1139294546 M * mugwump Bertl: it boots, but can't mount root
1139294547 M * Hollow with plain init style, you won't see anything
1139294565 M * jkl Hollow: started with plain, entered the vserver and watched psauxw
1139294571 M * Hollow ah
1139294572 M * Bertl mugwump: ide fs, initrd with correct fs? modules in the initrd?
1139294576 M * Bertl mug	could you upload the boot log?
1139294579 M * jkl i tried to start famd 
1139294581 M * jkl no luck
1139294590 M * mugwump Bertl: how do I get the boot log?
1139294593 M * jkl but i got OK's for portmap and famd
1139294597 M * jkl they just arent running
1139294611 M * mugwump ah, -d
1139294613 M * Bertl mugwump: ah, you are using graphics mode?
1139294634 M * Hollow jkl, Bertl: well.. will portmap and/or famd work inside a vserver?
1139294651 M * Bertl to some extend, yes
1139294664 M * jkl has been
1139294677 M * jkl courier imap needs it
1139294678 M * mugwump I'm just using the defaults, but it is popping up an X window
1139294686 M * Hollow jkl: certainly not ;)
1139294696 M * jkl well, part of it does then
1139294697 M * Hollow but that's another question..
1139294698 M * jkl or something
1139294709 M * Hollow no, i'm running courier all over the place, never installed portmap or fam
1139294710 M * mugwump ah, I suppose that's what -nographic would be about
1139294712 M * jkl yes, it is another question
1139294736 M * jkl Hollow: it's for monitoring folders that the user doesnt explicitly load in the client afaik
1139294737 M * Hollow jkl: so, did you try to start portmap manually?
1139294748 M * Bertl mugwump: needs support on the kernel side (for serial console)
1139294759 M * jkl yeah
1139294762 M * jkl nothing
1139294774 M * Hollow with the init script, or the portmap binary?
1139294781 M * jkl init script
1139294795 M * Hollow then c&p the command line the init script uses and run it, and see what happens
1139294804 M * jkl yeah, ok
1139294824 M * jkl looks to be running
1139294825 M * mugwump http://rafb.net/paste/results/XwXoHM17.html  # qemu boot log
1139294837 M * Hollow that's odd
1139294840 M * jkl i was able to bring the whole mailserver back up in this fashion earlier
1139294843 M * mugwump I don't see it seeing the partition tables...
1139294850 M * jkl but everything thus got started as root
1139294856 M * jkl and i had serious permission problems
1139294864 M * jkl i dont knwo why all these init scripts are failing
1139294900 M * jkl famd comes up fine right after that (using the bin)
1139294956 M * jkl i think i missed too many updates
1139294985 J * Aiken__ ~james@tooax7-231.dialup.optusnet.com.au
1139294993 M * Hollow jkl: ok, just to be sure, it's not a ~arch bug, try `CONFIG_PROTECT_MASK="/etc/init.d /etc/inittab" emerge =sys-apps/baselayout-vserver-1.11.14-r1` it's ~arch too, but will be stable the next days..
1139295012 M * mugwump Bertl: actually, ignore that.  I'm going to build my own disk image and see how I get on
1139295034 M * jkl i remember seeing an update for /etc/inittab when i did etc-udpate
1139295037 M * jkl could that be the problem
1139295051 M * jkl the changes looked harmless
1139295063 M * jkl but then again what do i know :)
1139295085 M * Hollow it may be a problem if you upgraded sysvinit after baselayout-vserver, it would overwrite the inittab then
1139295093 M * Bertl mugwump: you can use my scripts for that, if you like
1139295095 M * Hollow *written
1139295106 M * jkl i was doing emerge -u world
1139295116 M * jkl so if it's not smart enough to order them correctly
1139295116 M * Hollow hm.. then it should have been right
1139295119 M * jkl that could have happened
1139295134 M * Hollow anyway.. try to downgrade and see what happens
1139295137 M * jkl ok
1139295151 A * mugwump searches linux-vserver.org for qemu
1139295179 M * Bertl http://vserver.13thfloor.at/Stuff/QEMU/
1139295189 M * mugwump Bertl++
1139295197 M * Hollow jkl: the thing is.. pre8 is quite old.. normal baselayout in ~arch has pre15 already, so i'd like to update the ebuild first before doing any debugging on ~arch
1139295246 M * jkl ok it's downgraded, im going to try restarting it?
1139295260 M * Hollow yeah
1139295306 M * jkl hm, timeouts
1139295329 Q * Aiken_ Ping timeout: 480 seconds
1139295329 M * jkl there she goes
1139295336 M * jkl phew
1139295348 M * Hollow phew?
1139295360 M * jkl yeah, like a "sigh of relief" ?
1139295367 M * Hollow it's working?
1139295374 M * jkl well things started
1139295375 M * jkl hehe
1139295378 M * Hollow wth
1139295393 M * jkl everything is running as it should be
1139295402 M * Hollow but ic.. will update ~arch soon to pre15 and probably you can arrange a test copy of this guest
1139295446 M * jkl well, im buying a bigger HD
1139295460 M * jkl so i can run pre-production test servers 
1139295460 M * Bertl mugwump: you might have to adjust the scripts slightly ...
1139295467 M * mugwump hmm, yes, I see :)
1139295472 M * jkl i don't want to get spooked like this again!
1139295525 M * Hollow jkl: ok.. just keep a tar of the root around or so, then we can test things without tainting the real one
1139295544 M * jkl yeah, ill do that
1139295570 A * mugwump wonders why he only has one file in /dev/loop
1139295571 M * Hollow ok, will have to leave for school now, too late anyway
1139295584 M * jkl thanks for your help
1139295588 M * jkl you really bailed me out!
1139295596 M * Bertl mugwump: that's from pre-udev (or more precisely devfs times)
1139295615 M * Bertl mugwump: you probably have /dev/loop{0,1,2,3 ...}
1139295622 M * Hollow jkl: you're welcome!
1139295626 M * mugwump well, I have both for now
1139295628 M * Hollow which TZ are you in?
1139295642 M * mugwump aha, is that how you loop mount a partition.  sneaky
1139295726 M * jkl MST
1139295727 M * jkl USA
1139295759 M * jkl almost time to go to bed
1139295762 M * jkl :)
1139295790 M * Hollow ok, so we'll probably meet in (my) evening again ;)
1139295859 M * Hollow jkl: ok, i have commited 1.11.14-r1 to stable meanwhile... you can safely remove any baselayout entries from p.keywords now
1139296187 M * jkl ok cool
1139296189 M * jkl thanks
1139296278 M * jkl sys-apps/baselayout-vserver ~x86
1139296279 M * jkl heh
1139296283 M * jkl where did that come from
1139296284 M * jkl yeesh
1139296559 M * Hollow hm, guess maths is not _that_ important..
1139296569 M * Hollow at least not at 8am :D
1139297203 Q * eugenesan Quit: Trillian (http://www.ceruleanstudios.com
1139297777 M * mugwump yay!  my kernel is booting (1b)
1139297790 M * Bertl congrats!
1139297800 J * andrew_ ~andrew@tnlug.linux.org.tw
1139297859 M * mugwump hmm, it's not the fastest
1139297888 M * mugwump darn, what's it doing now...
1139297999 M * mugwump oh, inittab for /dev/ttyS0 :)
1139298177 M * mugwump hmm, still no getty spawned
1139298268 M * mugwump right, I guess having a /dev/ttyS0 file is important too
1139298308 M * mugwump heh.  serial8250: too much work for irq4
1139298856 J * meandtheshell ~markus@85-125-228-199.dynamic.xdsl-line.inode.at
1139299104 M * mugwump Bertl, do you have a recipe for starting qemu with gdbserver?
1139299793 M * mugwump hmm, why the change to proc_check_chroot?
1139299800 M * mugwump in fs/proc/base.c?
1139299813 M * mugwump (details: if (capable(CAP_CHCONTEXT)) return 0;)
1139300140 Q * Bertl Ping timeout: 480 seconds
1139300236 J * Doener doener@i5387FCBA.versanet.de
1139300321 M * daniel_hozac that was to fix chcontext --xid 1 netstat -pnlt, among other things, i think.
1139300354 M * Doener morning
1139300358 M * daniel_hozac morning
1139300726 J * SNy 9a2cb72906@bmx-chemnitz.de
1139302096 Q * shedi Quit: Leaving
1139302408 M * mugwump wahey!  I have a compiling 1c
1139302484 M * Doener 1c?
1139302493 M * mugwump see utsl.gen.nz/vserver/patch-plan.txt
1139302516 M * Doener ah! :)
1139302710 A * mugwump grrs at squillions of modprobe: FATAL: Could not load /lib/modules/2.6.16-rc2-g7af61406-dirty/modules.dep: No such file or directory
1139302803 M * Doener hrhr
1139302996 J * Bertl herbert@212.16.62.52
1139303014 M * Bertl oops seems I lost connectivity ...
1139303064 M * mugwump wahey!  1c boots and can see stuff in /proc/virtual and /proc/XX/vinfo
1139303086 M * Bertl cool!
1139303103 M * Doener morning Bertl 
1139303111 M * Bertl morning Doener!
1139303132 M * Doener Bertl: was the general idea of the page guide right?
1139303154 M * Doener or too little content to judge that?
1139303170 M * Bertl well, it definitely looked right to me ...
1139303209 M * Doener ok, then I'll continue on that in the evening
1139303241 M * Bertl good, will fix a few things in the next days and spend some time on documentation too
1139303316 M * Doener I'll also add a few pages that describe linux stuff itself which is important for vserver, like I did for the namespaces... ioctl vs. netlink interface (ifconfig vs. iproute) is high on that list ;)
1139303342 M * Doener will also spend some time on reading up on the new mount stuff and extend the namespace page then
1139303347 M * Bertl yep, sounds good
1139303459 M * harry does anyone here know powerpc asm? (as in ibm's power5 cpu?)
1139303466 M * Bertl yup
1139303469 M * harry and aix debuggers etc...
1139303473 M * harry sec, phone
1139303482 M * Bertl aix no, but ppc yes
1139303513 M * harry kinky...
1139303526 A * harry has to go get something that has arrived (i don't know what, but it's somnething)
1139303530 M * harry excitiiiiiiiing :0
1139303530 M * harry ;)
1139303533 M * harry so... brb
1139303681 Q * Dr4g Quit: Leaving
1139304539 J * prae ~prae@ezoffice.mandriva.com
1139304591 M * Bertl welcome prae!
1139304681 M * prae Hi Bertl
1139304721 M * Bertl prae: recently I checked out mdk 2006.0 ...
1139304745 J * shedi ~siggi@tolvudeild-204.lhi.is
1139304745 M * Bertl prae: are many folks using mandriva for servers?
1139304855 M * prae yes, some companies
1139304881 M * Bertl I just wondered, because it required a few tricks to make it usable for servers
1139304973 M * Bertl prae: are you interested in feedback?
1139305297 M * prae Bertl: yes, of course
1139305304 M * prae (sorry for the lag  :)
1139305307 M * Bertl np
1139305324 M * Bertl well, the first issue I encountered was with x11 being required
1139305339 M * Bertl doesn't make sense to me to have x11 on a server
1139305353 M * mugwump okie, that's me ... seeya folks
1139305362 M * Doener bye mugwump 
1139305363 M * Bertl cya mugwump!
1139305400 M * prae Bertl: normally, x11 is available under mdv server
1139305439 M * Bertl prae: well, no problem with 'available' but it cannot be removed :)
1139305480 M * Bertl i.e. an mdk 2006.0 system has to have x11. period.
1139305511 M * prae oh! ok, sorry, I don't see the second answer ... well, so, hmmm, me too, but some customer need a Xserver under server, it's strange , I know
1139305540 M * prae normally, we can removed x11-* under installation
1139305555 M * Bertl yes, problem is, it is required by basesystem :)
1139305572 M * prae oh ? no joke ?!
1139305592 M * Bertl (not directly, but indirectly via that strange bootsplash setup thing)
1139305602 M * Bertl no, unfortunately not
1139305615 M * prae "libgtk-linux-fb-2.0.so.0"
1139305615 M * prae ?
1139305622 M * prae $ rpm -qR bootsplash  | grep -ris X
1139305622 M * prae rpmlib(PayloadFilesHavePrefix) <= 4.0-1
1139305622 M * prae libgtk-linux-fb-2.0.so.0
1139305641 M * FaUl someone here who is familar with bus-errors?
1139305657 M * prae it's strange, I install vserver-mdv without Xlibs  :|
1139305663 M * prae wait,  I check it
1139305673 M * FaUl anyone with an idea why this: " if (0 == (request = find_id(header->id)))" may cause a "bus-error"
1139305697 M * Bertl header not defined?
1139305797 M * FaUl shouldn't this give a segmentation fault then?
1139305807 M * Bertl depends
1139305815 M * FaUl on what?
1139305821 M * Bertl if it is an unaligned acces, it could cause a buserror too
1139305897 M * prae Bertl: no x11-* if I install vserver with basesystem  :-\
1139305910 M * prae Bertl: problem appears only in mdv 2006.0 inst' ?
1139305928 M * Bertl prae: okay, is there a simple way to see a dependancy tree?
1139305979 M * prae maybe a dependancy in "phantom" and old package
1139306011 M * prae Bertl: when you install mdv 2006.0 and uncheck xserver, somes X11 stuff are installed ?
1139306042 M * Bertl I tried to minimize an existing install, and it keeps installing x11
1139306050 M * Bertl (or depending on it)
1139306064 M * Bertl is there some 'special' server flag to set?
1139306120 M * FaUl mh, I wonder why they reimplement dns instead of using the usual libc-provided routines anyway
1139306140 M * prae Bertl: hmmm, normally, we install server with a kickstart or select "minimal install"
1139306159 M * prae Bertl: have a "fresh" mandriva installed into server ?
1139306162 M * prae +you
1139306192 M * Bertl well, I didn't install it myself, so maybe the problem lies there
1139306207 M * Bertl I just tried to remove packages I do not want/need
1139306243 M * Bertl (maybe I should try a new install first)
1139306276 M * prae ok, it's possible to send me a tarball with all /root/drakx/ and "rpm -qa" 's result ?
1139306303 M * prae (from minimal install with X11)
1139306363 M * prae second question, when you removed unwanted packages, it's after a minimal install or normal install ?
1139306461 M * Bertl will retry with a clean install first, maybe something got messed up ...
1139306483 M * Bertl the second thing I observed was that bonding devices are not handled properly ...
1139306540 M * Bertl i.e. if you have eth0, eth1 and bond them together to form bond0 and you want to use vlans on that new bonding device, the network scripts do not handle that ...
1139306814 M * prae :-\
1139306839 M * prae I see this with core mdv-dev
1139306907 M * FaUl Bertl: ok, they use recvfrom to recive an dns-answer and then using foo /* which is a struct that matches an dns-answer */ = buf;
1139306926 M * FaUl first field in header is a int_16_t id
1139306940 M * FaUl and if you try to access that id it dies with bus_error
1139306967 M * Bertl FaUl: use gdb to get more info on that
1139306992 M * FaUl Bertl: how? :-)
1139306997 M * Bertl FaUl: very likely some memory got messed up before
1139307011 M * Bertl FaUl: gdb your-program =
1139307016 M * Bertl s/=/?/
1139307048 M * prae Bertl: a question from mdv-dev, you use a static ipaddr ou dhcp ?
1139307052 M * FaUl yes, i did that, thats how i found out that it dies at the access of header->id :-)
1139307132 M * Bertl prae: tried both, but currently a static one .. detail: there is no address assigned to the bond0 device itself, but I assume the main issue is that the bond is not configured
1139307159 M * Bertl prae: writing configs for bond0 didn't help IIRC
1139307535 M * Bertl prae: ah, had to dig out my documentation there, the important detail seems to be the vconfig
1139307537 Q * meandtheshell Ping timeout: 480 seconds
1139307597 M * Bertl prae: /etc/sysconfig/network-scripts/ifup uses /sbin/vconfig, where vconfig installs it in /usr/sbin/vconfig
1139307629 M * Bertl prae: sorry for emphasizing on the bond stuff, I did just remember that wrong
1139307774 M * Bertl prae: here is what I actually fixed: http://vserver.13thfloor.at/Stuff/vconfig-fix.diff
1139307838 M * prae I report this to the coredev
1139307906 M * prae hmmm
1139307907 M * prae $ ls -lah /sbin/vconfig /usr/sbin/vconfig
1139307907 M * prae ls: /sbin/vconfig: No such file or directory
1139307907 M * prae -rwxr-xr-x  1 root root 7,9K fév 22  2005 /usr/sbin/vconfig*
1139307909 M * prae pouhahah  :)
1139307915 M * prae OMG
1139308114 M * prae Bertl: it appears under 2006.0 only ?
1139308276 M * Bertl well, seems that 8.2 and 9.0 had no vconfig/vlan support 
1139308286 M * Bertl haven't checked on 10.x yet
1139308321 M * Bertl 10.1 has the same issue
1139308444 M * Bertl prae: okay, and the third issue I encountered, and still have, seems to be that certain partitions cannot be unmounted on shutdown (in a clean way)
1139308499 M * Bertl I have a separate partition for /var and /var/log, and it seems that /var/log cannot be unmounted ...
1139308547 M * Bertl aside from those things, it seems to work quite well ...
1139308782 M * Bertl okay, off to bed now .. have to get some sleep ... back in the evening ...
1139308793 M * Bertl have fun, and cya!
1139308798 N * Bertl Bertl_zZ
1139308833 M * prae Bertl_zZ: I send your feedback to the dev; thx
1139312635 J * Viper0482 ~Viper0482@p5497614F.dip.t-dialin.net
1139313566 Q * mcp Read error: Connection reset by peer
1139313918 J * cryo ~say@212.86.233.146
1139314045 J * liquid3649_ ~Viper0482@p5497614F.dip.t-dialin.net
1139314116 Q * liquid3649_ Quit: 
1139314160 J * lilalinux ~plasma@80.69.35.186
1139314247 J * Aiken_ ~james@tooax6-050.dialup.optusnet.com.au
1139314435 J * sef_ ~sef@82.207.148.66
1139314480 Q * Viper0482 Ping timeout: 480 seconds
1139314534 M * sef_ hi all! any tips where I can find information regarding vserver + preemptible linux kernel? thanks....
1139314559 Q * Aiken__ Ping timeout: 480 seconds
1139314560 M * sef_ googled and vserver wiki searched, still couldn't find any information about it
1139315137 Q * balbir Quit: Leaving
1139316843 P * sef_ Client Exiting
1139316922 N * Bertl_zZ Bertl
1139316937 M * Bertl hmm, couldn't really sleep ... back now
1139316992 J * mkhl MiLLeNNiuM@200-148-40-42.dsl.telesp.net.br
1139317297 M * FaUl Bertl: got the bug! __attribute__ ((aligned (16))); fixed the problem
1139317321 M * FaUl (though its more like an ugly workaround then a nice solution)
1139317550 M * Bertl well, alignment is important :)
1139317595 M * FaUl mh, is there some #define that the compiler sets automagically which let me figure out which architecture i am?
1139317599 M * FaUl on which
1139317626 M * Bertl yes, each arch has one ... see the syscall.h for examples
1139317644 M * FaUl and this is set by gcc autamagically?
1139317656 M * FaUl or defined in some special headers?
1139317667 M * Bertl automagically ...
1139317675 M * FaUl nice
1139317677 M * Bertl http://vserver.13thfloor.at/Stuff/SYSCALL/vserver.h
1139317781 M * FaUl IIRC alpha need alignment as well - do you know on what? 64bit? 32bit? 16bit?
1139317910 M * Bertl depends on the datatype
1139317924 M * FaUl lets uint_16_t
1139317972 M * Bertl i.e. character is char aligned, word 2 byte aligned, int 4 byte, long 8 byte
1139317981 M * FaUl ah, ok
1139318348 J * Wonka produziert@chaos.in-kiel.de
1139318351 M * Wonka morning
1139318354 M * FaUl moin Wonka 
1139318360 M * Wonka hi FaUl...
1139318386 M * Bertl welcome Wonka!
1139318413 M * Wonka ui, several known nicks here
1139318413 M * Wonka hi cehteh, cemil too
1139318614 M * Bertl so you immediately feel @home here :)
1139318641 M * Wonka :)
1139318669 M * Wonka i am writing this in an irssi on a linux-2.4.29-vs1.2.10...
1139318684 M * Wonka which is to be replaced with a newer version soon
1139318692 M * Wonka i'd like to use a 2.6
1139318701 M * Bertl ah, good :)
1139318712 M * Wonka anything known on what not to do?
1139318712 M * Bertl and how is 1.2.10 for you?
1139318728 M * Wonka in 1.2.10, there are several annoyances
1139318731 M * Bertl Wonka: well, yes, don't use debian stable (for now :)
1139318745 M * Wonka can't change my IP while running
1139318760 M * Wonka ping uses the hosts IP as source, not mine
1139318779 M * Wonka no IPv6
1139318800 M * Bertl the first two sound like config issues to me
1139318830 M * Bertl ipv6 will have to wait for ngnet
1139318883 M * Wonka :(
1139318898 Q * Doener Quit: Leaving
1139318908 M * Wonka what is this CONFIG_VSERVER_NGNET in 2.1.0.5.1?
1139318939 M * Bertl the first step to getting away from legacy networking
1139318955 M * Wonka no chance to get IPv6 anytime soon?
1139318974 M * Wonka hm, so leave it at N now
1139318989 M * Wonka this needs to run stable
1139319080 M * Bertl then I would suggest to choose the stable branch ...
1139319103 M * Wonka i don't really like using 2.6.14.4
1139319111 M * Wonka we are nearly at 2.6.16...
1139319210 M * Bertl one has nothing to do with the other ...
1139319256 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.15-vs2.0.1.2.diff
1139319352 M * Wonka hm, oops
1139319405 J * NetAsh ~foo_bar@193.219.160.108
1139319410 M * Wonka anyway, am building 2.1.0.5.1 for my notebook at the moment, to play around with that
1139319416 M * NetAsh hello
1139319430 M * Bertl welcome NetAsh!
1139319438 M * NetAsh simple question regards startin vservers on host start
1139319450 M * Bertl Wonka: no problem with that, it is very 'stable' anyway ...
1139319453 M * NetAsh ## putting the flavor into /etc/vservers/<vserver-id>/apps/init/mark.
1139319453 M * NetAsh #MARK=default
1139319492 M * Bertl hmm, yes?
1139319528 M * NetAsh exactly what shuld be writen into /etc/vserver/foo/apps/init/mark file?
1139319556 M * Bertl the name used for the startup script
1139319652 M * NetAsh I just unkomented MARK=default in init script, and in .../apps/init/mark placed default
1139319656 M * NetAsh it worked
1139319689 M * Bertl that's the idea ...
1139319705 M * Bertl if you need shutdown, make sure to pay attention to the first paragraph
1139319723 M * NetAsh about lock file?
1139319744 M * Bertl if you want to make two different categories for (e.g. runlevels) you can simply copy the vserver-default to something else, change the MARK/mark and use that
1139319828 M * NetAsh I just was confused with #MARK=default
1139319847 M * NetAsh and at first time I placed into init/mark file the string MARK=default
1139319885 M * NetAsh it looked weird, and faild
1139319885 M * NetAsh so I asked
1139319888 M * NetAsh I sugest rewrite the comment a litle
1139319899 M * NetAsh ./
1139319918 M * Bertl patches are always welcome!
1139320068 M * NetAsh -## putting the flavor into /etc/vservers/<vserver-id>/apps/init/mark.
1139320068 M * NetAsh -#MARK=default
1139320068 M * NetAsh +## putting the flavor name into /etc/vservers/<vserver-id>/apps/init/mark.
1139320068 M * NetAsh +## This script represents:
1139320068 M * NetAsh +#MARK=default
1139320076 M * NetAsh something like this ;)
1139320163 M * NetAsh btv Bertl how far 2.1 is from stable?
1139320217 M * Wonka and how far ist ngnet from testable?
1139320297 Q * Aiken_ Quit: Leaving
1139320321 M * sannes .. and when will it compile? ... ok ok j/k :P
1139320373 M * Bertl I'm basically waiting for ebiederm here, as I do not want to have two different versions
1139320385 M * Bertl (and twice the efford, of course)
1139320464 M * NetAsh the one with *_zZ ?:)
1139320501 M * NetAsh ./
1139320516 M * NetAsh ok and there I gues is seriuos question
1139320523 M * Wonka hrm
1139320535 M * NetAsh wahts about kernels powernow fietures & vserver
1139320548 M * NetAsh aka cpu scaling
1139320549 M * Wonka what about suspend2 and vserver...
1139320551 M * Wonka *sigh*
1139320557 M * Wonka *patch*
1139320671 M * Bertl hmm, well, the typical linux-vserver application does not require suspend or cpu scaling, but they should work as expected
1139320671 M * NetAsh Wonka is this some sort of ansver in nonhumanoid language, or just aditional points in my question?
1139320686 M * Wonka kernel/built-in.o: In function `suspend2_cleanup_console': undefined reference to `vx_rmap_pid'
1139320689 M * Wonka kernel/built-in.o: In function `suspend2_notify_userspace':netlink.c:(.text+0x35f7f): undefined reference to `vx_rmap_pid'
1139320692 M * Wonka kernel/built-in.o: In function `nl_set_nofreeze':netlink.c:(.text+0x3619c): undefined reference to `vx_rmap_pid'
1139320695 M * Wonka kernel/built-in.o: In function `suspend2_cleanup_usm': undefined reference to `vx_rmap_pid'
1139320708 M * Wonka NetAsh: it's not related to your question
1139320722 M * Bertl that is with an additional patch, yes?
1139320758 J * Roey ~katz@h-69-3-4-130.mclnva23.covad.net
1139320779 M * NetAsh Bertl servers eat power, and cpu scaling gives a lot of sence then server is idle
1139320799 M * Bertl welcome Roey!
1139320817 M * Bertl NetAsh: yes, agreed, but usually providers have no idle servers :)
1139320841 M * Bertl NetAsh: nevertheless what the vanilla kernel supports should work fine, no=
1139320842 M * Wonka Bertl: yes, with suspend2
1139320845 M * Bertl s/=/?
1139320865 M * NetAsh okk
1139320880 M * Bertl Wonka: you got a few warnings earlier (which you ignored), in those files, you have to add an include line
1139320910 M * Wonka i am at the moment looking for the files, in which to add #include-s
1139320934 M * Wonka ah, find_task_by_pid it is
1139320972 M * Bertl #include <linux/vs_cvirt.h>
1139320972 M * Bertl (that's what you probably want in your 'new' files)
1139321084 M * Bertl btw, interesting that suspend2 uses pid information ...
1139321096 N * ebiederm_zZ ebiederm
1139321104 M * Bertl morning ebiederm!
1139321108 M * ebiederm Morning.
1139321116 M * Wonka it needs to, it's communicating with a userspace process
1139321125 M * Wonka it's userui, to be specific
1139321128 M * Bertl Wonka: ah, okay, makes sense ...
1139321166 M * Bertl will require some changes once the vpid stuff is in mainline ...
1139321182 J * Milf ~Miranda@ipsio146.ipsi.fraunhofer.de
1139321189 M * Bertl ebiederm: so you can add suspend2 to the list :)
1139321192 M * Milf Good morning
1139321193 M * Bertl welcome Milf!
1139321208 M * Wonka "Miranda I'd Like to F...?"
1139321222 M * Milf Where can I find an XML file detailing the possible config options in 2.0 style config?
1139321225 M * ebiederm Bertl:  I think I can wait until suspend2 is merged can't I?
1139321244 M * Milf Wonka: I use that name longer than that movie exists or any porn sites
1139321253 M * Wonka Milf: ok...
1139321275 M * Bertl Milf: it's in your util-vserver package
1139321276 A * Wonka uses his name since before he knew the novel and before the movie existed...
1139321315 M * Bertl Milf: or online as html here: http://www.nongnu.org/util-vserver/doc/conf/configuration.html
1139321357 M * Milf Ok, so the great flower page _is_ the xml file?
1139321373 M * Bertl no, it is the result of using sgml on the xml :)
1139321389 M * Bertl well, or the other way round ...
1139321421 M * Wonka waahrgh! it's violently colored, and it blinks!
1139321433 M * Bertl Wonka: depends on the stylesheet
1139321476 M * Bertl Milf: util-vserver-0.30.210/doc/configuration.xml
1139321783 Q * mnemoc Ping timeout: 480 seconds
1139322025 M * Milf Thanks.
1139322063 M * Milf Wonka: I just found out lately that my browser lets me choose the stylesheet used in displaying
1139322084 M * Milf I've lost my horror of 2.0 configstyle and the flower page over that knowledge
1139322100 J * mnemoc ~amery@200.75.27.10
1139322104 M * Bertl :)
1139322138 J * Smutje_ ~Smutje@xdsl-84-44-246-39.netcologne.de
1139322243 M * Milf And now I finally, after one year, find myself having a HiWi who has time to write me a tool that lets me create the config from an XML-File.
1139322249 Q * Smutje Ping timeout: 480 seconds
1139322249 N * Smutje_ Smutje
1139322266 M * Milf I will set him to the task if no one else has done anything like it until now.
1139322596 M * Roey Bertl:  hi hi :) didja catch my link?
1139323511 Q * Loki|muh Ping timeout: 480 seconds
1139323650 N * nokoya nokoyaz
1139323717 J * Loki|muh loki@satanix.de
1139323722 M * Bertl night nokoyaz!
1139323734 M * Bertl Roey: link?
1139323746 N * nokoyaz nokoya
1139323749 M * nokoya night :D
1139324288 M * Roey Bertl:  yeah from last night
1139324289 M * Roey hmm
1139324290 M * Roey yeah.
1139324441 M * Bertl Roey: ah, the ascii art ...
1139324855 M * Roey =)
1139326796 Q * mkhl Ping timeout: 480 seconds
1139327366 J * Doener doener@i5387DE50.versanet.de
1139327524 Q * Milf Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org
1139328739 J * meandtheshell ~markus@85-125-228-199.dynamic.xdsl-line.inode.at
1139329203 J * stefani ~stefani@superquan.apl.washington.edu
1139329846 J * Dr4g Dr4g@82-40-43-86.cable.ubr06.uddi.blueyonder.co.uk
1139330360 Q * shedi Quit: Leaving
1139331049 Q * Dr4g Quit: Leaving
1139331264 J * Dr4g Dr4g@82-40-43-86.cable.ubr06.uddi.blueyonder.co.uk
1139332020 Q * jkl Quit: vserver restart
1139332384 J * jkl eric@c-67-172-156-116.hsd1.co.comcast.net
1139333379 Q * lilalinux Remote host closed the connection
1139333578 J * lilalinux ~plasma@80.69.35.186
1139333675 Q * lilalinux Remote host closed the connection
1139333843 J * Dr4g_ Dr4g@82-40-43-86.cable.ubr06.uddi.blueyonder.co.uk
1139333993 J * Viper0482 ~Viper0482@p54975EBF.dip.t-dialin.net
1139334056 Q * NetAsh Quit: 
1139334257 Q * prae Quit: Execute Order 69 !
1139334266 J * menomc ~amery@200.75.27.63
1139334274 M * Bertl hey menomc!
1139334300 Q * Dr4g Ping timeout: 480 seconds
1139334369 Q * mnemoc Ping timeout: 480 seconds
1139334369 N * menomc mnemoc
1139334392 M * Bertl Vudumen: ping!
1139334900 Q * arnaud Quit: Leaving
1139335092 M * Roey Bertl:  how about vserver with nx, do you know about that?
1139335106 M * Bertl nx?
1139335148 J * lilalinux ~plasma@h1-gw.of.net-lab.net
1139335161 M * Bertl there are a lot 2 letter acronyms, which nx do you refer to?
1139335182 M * Roey freenx
1139335185 M * Roey from NoMachine
1139335195 M * Roey which doens't have an amd64 client or server :( :(
1139335219 M * Bertl what about it?
1139336875 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1139337005 A * oliwel waves hello to the crowd
1139337015 M * Bertl hey oliwel!
1139337018 M * oliwel Hi Berlt
1139337023 M * oliwel Nice to meet you ;)
1139337040 M * Bertl quite unusual, isn't it :)
1139337193 A * Hollow join waving
1139337195 M * Hollow +s
1139337260 A * oliwel waves again ;)
1139337285 M * oliwel Hollow: any plans / timelines for new kernel patches in portage?
1139337330 M * Hollow yep, when 2.6.16 is out
1139337337 M * oliwel All: possbile ugly bug....unregister_netdevice: waiting for vlan3 to become free. Usage count = 1
1139337346 M * oliwel Hollow: so no patchsets against .15 ?
1139337355 M * Hollow no
1139337363 M * oliwel ok
1139337379 M * Hollow at least no new ones... 2.1.0.4 is for 2.6.15
1139337418 M * oliwel yes but bertl recomedns 2.1.0.8 :)
1139337452 M * Hollow but there is no patch for 2.6.15
1139337458 M * oliwel hmm ok....
1139337459 M * Hollow 2.1.0.8 is for 2.6.16-rc1
1139337463 M * oliwel ah ok
1139337475 M * oliwel dont matter....
1139337476 M * Hollow but i can't use an rc kernel, due to genpatches
1139337486 M * oliwel sure
1139337492 M * oliwel So - any ideas on the "unregister_netdevice: waiting for vlan3 to become free. Usage count = 1" problem 
1139337497 M * Hollow no :)
1139337510 M * oliwel it appears since I have installed tun device and started openvpn (in the root context)
1139337524 M * oliwel Problem is that the whole system crashes
1139337528 M * Bertl oliwel: should be fixed in later kernels
1139337538 M * oliwel dont know if it is vserver related...
1139337566 M * oliwel uname -a
1139337568 M * oliwel Linux ankeix 2.6.15-vs2.1.0.3-gentoo 
1139337575 M * Bertl http://vserver.13thfloor.at/Devel/PAT-2.1.1/delta-indev-fix01.diff
1139337604 M * oliwel so its a known issue =
1139337639 M * Bertl known and fixed since januray the 9th
1139337645 M * oliwel ok :)
1139337654 M * Bertl not present in earlier mainline kernels
1139337673 M * Bertl it was a change in mainline which triggered this bug
1139337690 M * oliwel Hollow: ok..is this already fixed in 2.1.0.4 gentoo
1139337722 M * Hollow dunno..
1139337731 M * Hollow rebooting.. brb
1139337736 M * oliwel LOL
1139337740 Q * Hollow Quit: SIGTERM
1139337753 M * Bertl oliwel: could be, as 2.1.0.4 was released a day after that
1139337793 M * oliwel I will check  downloading...
1139337889 J * Hollow ~hollow@home.xnull.de
1139337899 M * oliwel you know any way to reboot such a hanging device ? Otherwise a trip to the Rechenzentrum is on schedule tomorrow...
1139338003 M * oliwel FYI: gentoo linux-2.6.15-vserver-2.1.0.4 contains the fix
1139338072 M * Bertl oliwel: well, you could certainly do a 'reboot -f'
1139338085 M * oliwel hmm
1139338097 M * oliwel how big is the chance =)
1139338103 M * Bertl of what?
1139338110 M * oliwel that the system comes back
1139338132 M * oliwel currently the system runs at least 80%....
1139338133 M * Bertl depends .. if you ensure that all guests are stopped, and the filesystems are mounted ro/synced
1139338147 M * oliwel stop guests is no problem
1139338164 M * oliwel but I have hanging processes due to the blocking netdevice
1139338167 M * Bertl the filesystem is the only thing you should worry about when doing reboot -f
1139338182 M * Bertl it will _immediately_ reset the machine
1139338196 M * oliwel hmm ok
1139338198 M * oliwel see...
1139338216 M * oliwel will give it a try tomorrow - I am 5 ins away from the maschine then ;)
1139338234 M * Bertl k, good luck ...
1139338248 M * oliwel so - no blocking process will prevent the boot ?
1139338263 M * Bertl look, it's like this:
1139338274 M * Bertl - userspace does reboot -f
1139338287 M * Bertl - this calls into the kernel (via sys_reboot())
1139338301 M * Bertl - given the permissions are there, this does reset the machine
1139338317 M * Bertl - no userspace is asked, nothing unmounted, nothing synced
1139338319 M * oliwel ahh ok - see...never been so deep in kernel issues ,)
1139338322 M * oliwel ok
1139338334 M * Bertl very similar to pressing the reset button actually
1139338334 M * oliwel So I must give it a try to sync the disks
1139338338 M * oliwel ok
1139338389 M * Bertl if your /proc contains a sysrq-trigger, you could also do some sync and emergency remount
1139338420 M * Bertl (but you might trigger other stuff this way, which in turn might hang the machine)
1139338479 M * Bertl so IMHO stopping processes, and syncing disks + reboot -f is your best chance ...
1139338501 M * Bertl if you _can_ remount the filesystem ro, then do that ...
1139338594 M * oliwel unfortunatly I cant...
1139338600 M * oliwel unable to ro the rootfs
1139338741 M * oliwel I think I will give it a try ;)
1139338761 M * oliwel Other chance tomorrwo is to switch of the power - is not really better
1139338778 M * Bertl no, that's basically the same :)
1139338831 M * Wonka sync;sync;sync;reboot -f
1139338843 M * oliwel so - "reboot -f" issued - cross your fingers....will see in a minute if it comes back
1139338873 M * Bertl it will hcheck the rootfs at least
1139338881 M * oliwel heah
1139338883 M * oliwel jeah
1139338886 M * Wonka yes...
1139338891 M * oliwel hopefully it has no errors ;)
1139338918 M * oliwel I must go to the Housing center this week to swap some parts - so its not really more work
1139338950 M * oliwel HEY it pings :)
1139338958 Q * weasel Quit: hardware stuff
1139338968 M * oliwel its alive :)
1139339773 J * weasel weasel@asteria.debian.or.at
1139339775 J * prae ~benjamin@sherpadown.net
1139339984 J * shedi ~siggi@inferno.lhi.is
1139340128 J * Pazzo ~Pazzo@host130-250.pool8172.interbusiness.it
1139340286 M * Bertl welcome Pazzo!
1139340305 M * Pazzo Hi Bertl!
1139340320 M * Pazzo everything all right?
1139340335 M * Bertl yup, everything fine .. and you?
1139340354 M * Pazzo absolutely, thank you
1139340437 M * Pazzo Vservers are running without trouble and atm I'm busy with lots of VoIP stuff so I'm here really seldom :-)
1139340461 M * Bertl well, good to hear that everything works as expected
1139340462 M * Pazzo How is ngngngnet going?
1139340482 M * Bertl currently in some sync with mainline state ...
1139340484 M * Dr4g_ anyone here wanna sell me a 19" LCD/TFT screen ?
1139340560 M * Pazzo Dr4g_: don't think so - we are all geeks, using serial cables and monochrome CRTs ;-)
1139340577 M * Bertl lol
1139340612 M * Dr4g_ CRT's are bad for your eyes :P
1139340781 M * mugwump so is too bright a contrast
1139340785 M * mugwump moinmoin all
1139340819 M * Pazzo Bertl: I'm interested in changes from 2.0 to 2.1 and infos on how ngnet is doing things now - is there a chance to find such things on the wiki?
1139340830 M * Pazzo moin mugwump!
1139340857 M * oliwel Bertl: Hollow thanks so far - gonna go - spend some time in RL ;)
1139340861 Q * oliwel Quit: Chatzilla 0.9.69.1 [Firefox 1.5/2005111116]
1139340871 M * Bertl Pazzo: the feature matrix should give you a few hints on differences
1139341085 M * Pazzo Bertl:  "Bind Mount Extensions (for ro --bind mounts)" <- does that mean that I can use bind mounts inside a vserver?
1139341139 M * Pazzo I/O scheduler? CoW? XID propagation? (-> for quotas???) - sounds great!
1139341419 A * Pazzo will be back in ~30 mins
1139341689 J * bonbons ~bonbons@83.222.38.150
1139341860 Q * gerrit Ping timeout: 480 seconds
1139343548 J * gerrit ~gerrit@bi01p1.co.us.ibm.com
1139343565 Q * Bertl Ping timeout: 480 seconds
1139343873 J * Aiken ~james@tooax6-021.dialup.optusnet.com.au
1139344136 Q * Greek0 Read error: Connection reset by peer
1139344178 J * Greek0 ~greek0@85.255.145.201
1139344398 J * SuperLag ~aaron@38.99.66.175
1139344758 J * Bertl herbert@212.16.62.52
1139344951 M * SuperLag Hollow: the Gentoo docs make reference to only one vserver partition.  How would you limit the size of each vservers storage space? 
1139344978 M * Bertl for example with vdlimit
1139345010 M * Hollow :) hey SuperLag 
1139345017 M * SuperLag Hollow: howdy
1139345413 Q * SuperLag Quit: brb
1139345703 J * SuperLag ~aaron@38.99.66.175
1139345709 M * SuperLag weird
1139345715 M * SuperLag Linux zoom 2.6.14-vs2.0.1-gentoo #1 Tue Feb 7 14:46:29 CST 2006 x86_64 AMD Athlon(tm) 64 Processor 3200+ AuthenticAMD GNU/Linux
1139345955 M * Bertl SuperLag: what's weird?
1139346099 M * SuperLag Bertl: tried to join #vserver initially, and it took me to ##unavailable
1139346464 M * derjohn when I talk about "memory limits" (in devel) ... RSS ist the real memory in "pages" ? is getpagesize() meanwhile implemented in the utils? 
1139346582 J * rene- ~rene@dsl-201-128-115-222.prod-infinitum.com.mx
1139346598 M * SuperLag rene-: what part of .mx are you from?
1139346636 M * rene- hi
1139346643 M * rene- i am from mexico city
1139346647 M * SuperLag hehe
1139346651 M * rene- i have just installed a debian based vserver
1139346654 M * SuperLag a Chilango
1139346658 M * rene- where are you from?
1139346679 M * SuperLag I'm from the US, but I speak Spanish as a second language.
1139346683 M * rene- have you been to mx/chilangolandia?
1139346705 M * SuperLag Oh yeah.  Previous employer sent me there and Toluca 4 times total.
1139346720 M * SuperLag You familiar with Grupo ACIR?
1139346724 M * rene- i do
1139346733 M * rene- they re in radio
1139346752 M * SuperLag I worked on the Novell servers for those radio stations.
1139346773 M * rene- oh cool. i once went to puebla to set up asterisk for them, it was only a demo
1139346791 M * SuperLag They use an automation program for their stations, and I used to work for the company that developed that automation software.
1139346817 M * rene- i did see that automation, was windows based, wasnt it?
1139346822 M * SuperLag yep
1139346827 M * SuperLag Unfortunately
1139346830 M * rene- heh
1139346840 M * rene- my vserver is running, i can enter and exit
1139346843 M * rene- its all cool
1139346856 M * rene- ive noticed that while i can ssh to the virtual server, i cant scp to it
1139346877 M * rene- and that i dont have pretty much any app in my vserver,
1139346888 M * rene- see i want to compile stuff but make is not there and then cc is not there
1139346901 M * rene- i dont have apt-get in the vserver
1139346912 M * rene- so the question is: how to add software to the vserver?
1139346912 M * derjohn rene-, scp uses ssh ... 
1139346922 M * derjohn rene-, vserver foo enter
1139346940 M * derjohn rene-, how did you install th guest 
1139346980 M * rene- i did something like vserver vserver1 build   -n vserver1   --hostname vserver1.mydomain.com   --interface eth0:192.168.1.10/24   -m debootstrap -- -d sarge
1139347004 M * derjohn rene-, ahm, debootstrap
1139347005 M * rene- i had to create a file named uri buried within the vserver configuration to the location of some bootstrap file from debian
1139347029 M * derjohn rene-, after that you surely did "vserver foo enter" and then "base-config" ?
1139347037 M * rene- cool, is it fixable? how should i be creating it
1139347043 M * rene- i did never issued a base-config
1139347062 M * derjohn rene-, then do so .. you guest is unconfigured at the time ;)
1139347069 M * rene- the guide i was following makes no mention to it at all
1139347073 M * rene- which is weird
1139347081 M * rene- so that goes inside the vserver right
1139347085 M * derjohn rene-, and by the way: i bet you entered the HOST with ssh, not the guest ?
1139347111 M * rene- no i did try ssh to the guest, it has a different ip and hostname
1139347117 M * rene- and i did got the right hostname3
1139347126 M * derjohn rene-, because we are evil^W dicks^W not responsible for debootstrap :)
1139347135 M * derjohn rene-, thats normal
1139347138 M * SuperLag Hollow: Is it safe to assume I can use this box for Perl stuff still, for Gentoo, and not disturb any of the guests?
1139347139 M * derjohn rene-, change it now
1139347176 M * derjohn rene-, but feel invited to add that little detail to the wiki , ok ?
1139347178 M * SuperLag derjohn: what do most of you guys run vserver on? Debian? Gentoo? something else entirely?
1139347184 M * rene- so, do i create a new vserver or do i run base-config inside my existing
1139347192 M * rene- allr ight
1139347193 M * rene- will do
1139347211 M * rene- i got a debian config prompt
1139347359 M * derjohn SuperLag, well,  depends .... we have fedora, mandrake, crux, gentoom debian ... well I am on Debian but will try foreign guests (I swear!) :)
1139347369 Q * lilalinux Remote host closed the connection
1139347403 M * SuperLag haha
1139347499 M * derjohn rene-, and now look at the hosts /etc/ssh/sshd_config and bind the hosts ssh daemon to exactly one port and one IP. Then (!) apt-get install ssh within the guest, OK ?
1139347519 M * rene- thanks
1139347527 M * derjohn eh ... de nada ;)
1139347538 M * derjohn rene-, fill the wiki an we are happy :)
1139347665 M * SuperLag derjohn: I have two NICs on this server.  I want to use this one for myself, and use the other for the guests.  I know I have to set up iptables to NAT to private addresses on the other NIC.  Any good docs on how to set that up?  For as much as I've used Linux, all my firewall stuff has been hardware-based, and I've never used iptables.
1139347718 M * derjohn SuperLag, well obviously you will find docs on netfilter.org 
1139347741 M * daniel_hozac SuperLag: google iptables ;)
1139347750 M * derjohn SuperLag, http://www.tgunkel.de/it/software/doc/linux_server.de#h3-Vserver_Masquerading_SNAT
1139347779 A * derjohn quits bot mode now :)
1139347812 M * SuperLag daniel_hozac: yeah, I know, I know...
1139348154 Q * Aiken Read error: Connection reset by peer
1139348174 P * stefani I'm Parting (the water)
1139348237 J * Aiken ~james@tooax6-021.dialup.optusnet.com.au
1139348758 M * derjohn rene-, I extended my FAQ for you: http://linux-vserver.org/some_hints_from_john Please check if the last Q/A is understandabwel for a beginner. thx.
1139348813 M * rene- that was the doc i was reading
1139348817 M * rene- thanks
1139348828 M * derjohn nope, I just added one Q at the end :)
1139348838 M * derjohn rene-, the ssh thing (this Q comes often here)
1139348901 M * rene- i see
1139349103 Q * Viper0482 Quit: bin raus, 
1139349162 M * rene- derjohn i need kernel sources... i of course what was left after compiling the linux vserver kernel but  that is the host context, i tried with apt but it complained about no hard disk space available... how do i resize my vserver?
1139349228 Q * bonbons Quit: Leaving
1139349242 M * derjohn rene-, slow please, again. You want to complile a new kernel on the host?
1139349278 M * rene- no, i want to compile asterisk
1139349289 M * rene- a component zaptel requeres kernel sources
1139349299 M * rene- s/requeres/requires
1139349316 M * derjohn rene, do you compile within a guest?
1139349328 M * rene- i am in guest context
1139349331 M * rene- yes
1139349339 M * daniel_hozac you'll need to do that in the host.
1139349344 M * derjohn rene-, ok, the most probably the the kernel headers are missing
1139349345 M * daniel_hozac it builds kernel module(s), IIRC.
1139349350 M * rene- they are
1139349362 M * rene- i dont think that i have enough space in the guest
1139349384 M * rene- i am not sure i have them in the host.. i have the /usr/src/linux***
1139349384 M * derjohn rene-, with debian kernels you can apt-get install kernel/linux..-headers--version.foo
1139349395 M * derjohn rene-, you may copy them
1139349404 M * derjohn rene-, did you limit your guests space?
1139349411 M * rene- derjohn: i tried that (apt) but it complained about no space
1139349414 M * rene- no i didnt
1139349417 M * derjohn rene-, by default there is no limit
1139349427 M * derjohn rene-, then the hdd or the partiotion is full??
1139349427 M * rene- let me try with apt again
1139349458 M * rene- dammit might be
1139349468 M * rene- the partition is full
1139349494 M * derjohn rene-, we cannot extend that virtually at the time ... :)
1139349497 M * rene- can i create a new vserver say in home
1139349508 M * rene- in /home
1139349514 M * derjohn rene-, well .. yes .. and no ....
1139349533 M * derjohn rene-, as you are a beginner , I would resize the partition
1139349563 M * derjohn rene-, there are things like barrier flags, xid and such ... which you could miss my not going the std way
1139349568 M * derjohn rene-, YMMV :)
1139349570 M * derjohn but ....
1139349645 M * derjohn mount --bind /home/biggerserver /var/lib/vserver/damnfull would maybe help you. but keep my warning from above in mind.
1139349766 M * derjohn rene-, maybe apt-get clean gives you some space free
1139349784 M * derjohn rene-, i dunno how much asterisk needs to compile
1139349928 M * derjohn rene-, or: mount --bind /usr/src /var/lib/vservers/usr/src and such ..
1139349951 M * rene- derjohn
1139349954 M * rene- thats good
1139349957 M * derjohn k
1139349961 M * rene- i will try that last one
1139349993 M * derjohn rene-, and remount the kernel headers dir :) .. or: comopile on the host and move the binary?
1139350003 M * derjohn rene-, YMMV ...
1139350048 Q * matti Remote host closed the connection
1139350151 M * rene- i will try harder tonight, i think as a beginner i should repartition
1139350193 M * derjohn maybe, yes. but there is no real need to have the vservers on a own part.
1139350226 M * daniel_hozac it's strongly recommended, however.
1139350241 M * daniel_hozac if you ever want to use tagxid, such as for disk limits.
1139350334 M * derjohn daniel_hozac, BTW ... then I set a limit with vdlimit ... is this accounted dynamically? I remember a strange script on the wiki which gives the "df -h" the right vaules
1139350381 M * daniel_hozac hmm?
1139350390 M * daniel_hozac if you set the disk limit in the configuration, everything is handled for you.
1139350396 M * daniel_hozac (as of 0.30.210)
1139350420 M * mugwump ebiederm: you around?
1139350495 M * derjohn daniel_hozac, I just visited the wiki again .. the page is completely different now. ah, I see the legacy page ... kk
1139350502 M * SuperLag crap.
1139350508 M * derjohn SuperLag, sure
1139350514 Q * weasel Remote host closed the connection
1139350515 M * SuperLag I should have been here for this whole conversation.
1139350519 M * daniel_hozac yes, i rewrote it when enrico committed my configuration patch ;)
1139350521 J * weasel weasel@asteria.debian.or.at
1139350522 M * SuperLag this is exactly what I'm wondering about.
1139350539 M * derjohn SuperLag, the are logs ob this channel (realtime) :)
1139350546 M * SuperLag except I don't need to resize anything.  I have a 160GB drive at my disposal
1139350561 M * derjohn SuperLag, dispose it to me :)
1139350587 M * SuperLag well, I'm trying to figure out the best way to chop this all up.
1139350596 M * SuperLag I think 5 x 30GB
1139350616 A * SuperLag wonders how much he can reasonably charge for a 30GB vserver
1139350646 M * derjohn SuperLag, ?re you sure about that you are doing?
1139350661 M * derjohn SuperLag, what is the max disk I/O of your subsystem?
1139350673 M * derjohn SuperLag, TPS vs. Transferrate?
1139350692 M * derjohn SuperLag, I assuem 300G disk to be sata. 
1139350701 M * SuperLag it is
1139350712 M * mugwump sata with NCQ?
1139350715 M * SuperLag yes
1139350723 M * derjohn SuperLag, 50 * 30 = 1500 GB ?
1139350724 M * SuperLag first generation, not second
1139350736 M * SuperLag derjohn: haha. read that again.
1139350739 M * mugwump nothing special about SATA 2
1139350760 M * SuperLag derjohn: I didn't say 50 * 30.  I'm not THAT crazy.
1139350760 M * mugwump it just means you support staggered spin-up and NCQ, and the extra connector AIUI
1139350761 M * derjohn SuperLag, if those 50 users really need 30 GB each (of 10 ... or 5 ...)
1139350774 M * SuperLag derjohn: I said *5* time 30
1139350784 M * derjohn SuperLag, I would bet you get a .... hmmm ... super lag !
1139350807 M * SuperLag derjohn: even with as few as 4-5 other users?
1139350824 M * SuperLag derjohn: the goal eventually is to set up a different machine for this purpose
1139350829 J * ntrs ~ntrs@68-188-50-87.dhcp.stls.mo.charter.com
1139350831 M * daniel_hozac SuperLag: 5 guests per host? you must hate oversubscribing.
1139350833 M * derjohn SuperLag, lol .. i read 300 GB disks not 30 ... are there 30 GB sata disk around ???
1139350866 P * meandtheshell 
1139350869 M * SuperLag daniel_hozac: I want to make enough to pay for my colo fees, and a modest profit.
1139350885 M * SuperLag daniel_hozac: and have it still be worthwhile for subscribers.
1139350916 M * derjohn daniel_hozac, what ratio would you recommend?
1139350919 M * SuperLag derjohn: let's try this again.  It's an AMD64 machine with a single CPU.  2GB of RAM, just two drives.  160GB each.
1139350936 M * derjohn SuperLag, hey, that my setup ... with sempron64 ...
1139350951 M * SuperLag daniel_hozac: and I was thinking... 4 vservers.  30GB each.  
1139350955 M * derjohn SuperLag, i lied. I have 2 * 80 GB
1139350981 M * daniel_hozac SuperLag: that sort of machine could easily pull off 10 busy guests.
1139350989 M * SuperLag daniel_hozac: really?
1139351026 M * SuperLag daniel_hozac: if you were going to do something like that, how much storage space would you provide, and how much would you charge?
1139351084 M * daniel_hozac well, limiting storage to <available space> / <guests> seems like the most logical choice.
1139351108 M * daniel_hozac charging, well, i have no idea. i have never paid nor been paid.
1139351117 N * Bertl Bertl_oO
1139351160 M * SuperLag I'm trying to eliminate an expense of $99/mo. for my colocation fees. :)  and since I have (more than likely) more hardware than I usually use, I figured I'd put some of it to work
1139351298 M * derjohn SuperLag, in .de there a vituozzo based vserver for about 0 Euro around (base fee, traffic extra and such. 1000 users on 16GB/quad xeon). A physcial colo starts from about 30-40 euro/month. The the _lower_ boundary is somewhere between. 
1139351321 M * mugwump 0 Euro?  Free?
1139351346 M * derjohn well, there may be hidden costs, but let me look up ...
1139351408 M * derjohn mugwump, i dont spam here. I sent to the URL privately
1139351511 M * mugwump appreciated.  be sure to add it to the Linux-VServer hosting page on the wiki, though
1139351541 M * daniel_hozac <derjohn> ... vituozzo... ;)
1139351604 M * derjohn daniel_hozac, did i mis-spell it ?
1139351614 M * derjohn daniel_hozac, I said: CHEAP ...
1139351617 M * derjohn :)
1139351639 M * daniel_hozac well, i was more referring to mugwump's comment.
1139351656 M * derjohn daniel_hozac, eh , yes :)
1139351686 M * derjohn daniel_hozac, but linux-vserver still has no webinterface .. maybe a cool project for the linuxtag ... 
1139351695 M * daniel_hozac derjohn: openvps.org
1139351733 M * daniel_hozac umm, sorry, misfire in the brain.
1139351740 M * mugwump yes, and my horse has no petrol tank either
1139351798 M * daniel_hozac lol
1139351925 M * derjohn i just see: only the first three monath are free. then it's 10 euro/month.
1139352278 M * derjohn daniel_hozac, i wasn't aware of openvps.... do you run it?
1139352543 M * daniel_hozac no.
1139352571 M * derjohn are there screenshots around?
1139352654 M * daniel_hozac i don't know, aren't there any on their website?
1139352687 M * derjohn daniel_hozac, no it pretty sparse ... but well linux-vserver.org is also not "pretty"
1139353277 M * mugwump hmm, recruiting spam from VMware in my inbox
1139353401 Q * Roey Quit: Leaving
1139353694 J * bwana ~bwana@71-82-125-40.dhcp.roch.mn.charter.com
1139353801 Q * prae Quit: Pwet
1139353954 Q * weasel Write error: connection closed
1139353959 J * weasel weasel@asteria.debian.or.at
1139354953 M * rene- phew
1139354963 M * rene- i have one more day to get this straight
1139354985 M * harry rene-: what/
1139354985 M * harry ?
1139354990 M * rene- so i will reinstall my debian host environment from scratch all files in one partition
1139354994 M * rene- and try again with the vps
1139355003 M * rene- well i ran into a partition too short
1139355042 M * rene- my vserver was getting choked
1139355057 A * harry will need to do a lot of research tomororw
1139355059 M * harry tomorrow
1139355068 M * harry vserver vs xen vs vmware
1139355080 M * harry all free, all advantages/disadvantages
1139355108 M * rene- this vserve thing is great, i want to do virtual asterisk. and this is what it is better documented
1139355151 M * harry rene-: i know it's great.. i like it very much
1139355156 M * harry played with it some time
1139355166 M * harry even got it running on a monstermachine now
1139355173 M * harry even made a grsec + vserver patch
1139355186 M * harry but now it's time to make a decission on what we'll use in production
1139355272 M * daniel_hozac harry: xen+vserver? ;)
1139355295 M * rene- what does grsec does?
1139355299 M * harry daniel_hozac: problem with xen is: kernel patches are "immergable" with grsec or other security patches
1139355308 M * harry you need to patch the syscall traps
1139355318 M * harry or whatever you like to call it
1139355325 M * harry rene-: grsecurity.net
1139355331 M * harry pax, rbac, ...
1139355344 M * harry security stuff, randomization, nonexec stack/heap, ...
1139355360 M * daniel_hozac all of which RH's kernels have, along with xen ;)
1139355373 M * daniel_hozac (not grsec though)
1139355376 M * harry fucking with brk locations etc
1139355435 M * harry does lots of nice stuff that make buffer overflow exploiting somewhat harder/impossible
1139355439 M * harry no... just harder :)
1139355482 M * rene- damm:  pretty heavy stuff
1139355504 M * harry rene-: pretty neat, yeah
1139355520 M * harry so is vserver, and thats why i merged patches
1139355525 M * harry secure virtualisation soft
1139355526 M * harry ;)
1139355784 M * derjohn  mount -t xfs --options remount,noatime,rw,tagxid /dev -> mount: /local not mounted already, or bad option ... are there any issues with xfs and tagxid (on dm) ?
1139355854 M * bwana so whats the deal with xfs anyway
1139355895 M * bwana what benifits does it offer over ext3
1139355971 M * harry faster for large files like movies etc
1139355980 M * harry aggressive caching
1139355987 M * harry but ... very fast :)
1139356092 M * derjohn harry, may part size, speedy lookups (even though reiser is a little better here...)
1139356097 M * derjohn max. part soze
1139356100 M * derjohn size
1139356151 M * harry ah...
1139356158 A * harry off to bed now... it's allready too late
1139356158 M * harry sry
1139356208 M * derjohn harry, and stable ....
1139356405 M * daniel_hozac derjohn: why are you passing remount if it's not already mounted?
1139356406 Q * rene- Read error: Connection reset by peer
1139356420 M * daniel_hozac and, remount+tagxid doesn't work.
1139356446 M * daniel_hozac as in, it is disallowed by the kernel.
1139356463 M * derjohn daniel_hozac, you wont believe it: it is mounted! I just wanted to enable quota /limtiing
1139356465 M * derjohn argh
1139356467 M * derjohn duh
1139356493 M * derjohn well, then the error line should reflect that.... 
1139356503 M * daniel_hozac IIRC you'll have dmesg lines about it :)
1139356542 M * daniel_hozac and bad option seems very much appropriate.
1139356581 M * derjohn XFS: dm-0: tagxid not permitted on remount.
1139356588 A * derjohn slaps himself.
1139356622 M * derjohn but anyway: why does this error not go to the stdout? well I know it know ... ;)
1139356649 M * daniel_hozac because that would be very inappropriate.
1139356679 M * daniel_hozac outputting things is userspace's job.
1139356707 M * derjohn daniel_hozac, well, i used the userspace 'mount' utility. so that would be it's part?
1139356730 M * daniel_hozac and mount did output an error ;)
1139356773 M * derjohn yes, but it could have said "exit 1" ... that would be the same kind information
1139356775 M * derjohn :)