1120956217 J * calamari ~jeffryj@lilly.csoft.net
1120956229 M * Bertl welcome calamari!
1120956230 M * calamari hi
1120956297 M * calamari when I use disk space inside a vserver, df still shows the same usage number.  is this a bug?
1120956377 M * Bertl depends on your setup
1120956433 M * calamari is there another tool I can use to get an accurate reading?
1120956458 M * Bertl could you give an example?
1120956468 M * Bertl how you test ...
1120956500 M * calamari sure... I copy an rpm to the vserver, rpm -i to install it, then remove the rpm and test the new disk usage
1120956548 M * Bertl and?
1120956564 M * calamari the disk space used is the same as before I did anything
1120956595 M * Bertl maybe the rpm was already installed?
1120956639 M * Bertl please elaborate a little on your setup (kernel/tools/etc)
1120956661 M * calamari sure.. one sec I want to verify a couple things
1120956752 M * calamari it works correctly on a vanilla vserver machine
1120956765 M * calamari so perhaps this needs to be asked elsewhere.. sorry :/
1120956784 M * calamari it's a planetlab vserver setup that gives the weird results
1120956808 M * Bertl aha, well, IIRC they patched a few things
1120957678 M * Vudumen now i go to bed. Bertl if you have any problems with the machine privmsg me and i'll look into :)
1120957683 M * Vudumen good night alls
1120957726 M * Bertl okay, thanks, night!
1120958106 M * calamari bbl.. thanks guys
1120958107 Q * calamari Quit: Leaving
1120958379 J * ntrs ~ntrs@62.162.246.101
1120958409 M * Bertl welcome ntrs!
1120959098 Q * ntrs Quit: Leaving
1120960194 M * Bertl okay, off to bed now ... back tomorrow ...
1120960201 N * Bertl Bertl_zZ
1120960378 Q * yarihm Quit: Leaving
1120962297 T * * http://linux-vserver.org/ | latest stable 1.2.10, devel 1.9.5, 2.0-rc6, ng9.5  -- He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 
1120962297 T * Bertl -
1120962325 J * matti matti@linux.gentoo.pl
1120964932 J * Aiken_ ~james@tooax6-250.dialup.optusnet.com.au
1120965257 Q * Aiken Ping timeout: 480 seconds
1120967853 J * monrad ~monrad@213083190130.sonofon.dk
1120976862 Q * DaCa-fi Ping timeout: 480 seconds
1120977261 J * DaCa-fi ~danny@mail.limehouse.org
1120978433 J * AprilDL_ ~chatzilla@ip68-9-200-247.ri.ri.cox.net
1120978433 Q * AprilDL Read error: Connection reset by peer
1120978437 N * AprilDL_ AprilDL
1120979897 J * Doener` ~doener@p54875C6E.dip.t-dialin.net
1120980295 M * Hollow morning
1120980332 Q * Doener Ping timeout: 480 seconds
1120981656 M * DaPhreak lo Hollow 
1120983897 J * yarihm ~yarihm@217-162-114-239.dclient.hispeed.ch
1120984582 Q * jkl_ Ping timeout: 480 seconds
1120986235 Q * jonsmel_zZ Read error: Connection reset by peer
1120987230 Q * complexho Remote host closed the connection
1120987635 J * xcyborg_ ~service@ns.bitdefender.com
1120987636 Q * xcyborg Read error: Connection reset by peer
1120991299 J * jkl_ eric@c-67-165-222-86.hsd1.co.comcast.net
1120992738 J * Wenix ~jrc@0x55510d61.adsl.cybercity.dk
1120992914 M * Wenix Is there any known problems with SSH inside a vserver.. When I connect to SSH I get disconnected.. until ssh is restarted in the vserver
1120993010 Q * Aiken_ Ping timeout: 480 seconds
1120993708 M * Vudumen it's not a vserver issue
1120993713 M * Vudumen i had the same on a freebsd box
1120993760 M * Vudumen i had this after many (unsuccessful) tries to login to the box with invalid credentials
1120994133 M * Wenix Vudumen: I'm not trying to say its a vserver issue, I just hoped someone else had the same problem and might know the solution
1120994319 M * Vudumen well we filtered tcp/22 :)
1120994427 M * Wenix hmm, not the solution I were looking for :)
1120994480 M * Wenix well, I almost never restart a vserver, so its only a problem for me once every 6 months.. I'll google it a little later :)
1120994482 M * Vudumen yes it was not for mee but i don't have more time for this issu eand it's a proper workaround now for us
1120994515 M * Vudumen hmm we had this issue after some days
1120994620 M * Wenix For me the solution is: start the vserver (with broken sshd), then enter vserver and restart ssh, everything works perfectly from this point on..  anyway, I'm about to leave now anyway.. thanks for the input :)
1120994707 Q * Wenix Quit: Keep up the good work
1120995996 J * renihs ~renihs___@193.170.52.70
1120997800 Q * renihs Remote host closed the connection
1121001133 J * renihs ~renihs___@193.170.52.70
1121002266 Q * renihs Remote host closed the connection
1121005668 J * wullon ~wullon@d80-170-50-3.cust.tele2.fr
1121005758 P * wullon 
1121005759 J * wullon ~wullon@d80-170-50-3.cust.tele2.fr
1121008231 J * AprilDL__ ~chatzilla@ip68-9-200-247.ri.ri.cox.net
1121008372 Q * AprilDL Ping timeout: 480 seconds
1121008903 J * AprilDL___ ~chatzilla@ip68-9-200-247.ri.ri.cox.net
1121008905 N * AprilDL___ AprilDL
1121008910 Q * AprilDL__ Read error: Connection reset by peer
1121009303 J * romke ~romke@procyon.romke.net
1121009527 Q * _are_ Quit: bbl
1121010075 N * Bertl_zZ Bertl
1121010085 M * Bertl evening folks!
1121010094 P * wullon 
1121010279 M * romke Bertl: you did It again, (every time I compile, test and upgrade kerner-vserver-2.0-rc... next morning You publish next rc version :P)
1121010419 M * Bertl nah, it was tonight :)
1121010451 M * romke :P
1121010459 J * Judderman ~Judderman@82-38-225-14.cable.ubr03.wake.blueyonder.co.uk
1121010461 M * romke (Sat Jul  9 06:02:47 2005 == tonight?)
1121010560 M * Bertl 09-Jul-2005 01:14 (patch)
1121010565 P * Judderman 
1121010594 M * romke Bertl: btw, after upgrading from -rc4 to -rc5 my guest I use for building packages started to fail builds because of 'Cannot allocate memory' -- any clues (I know, upgrade to rc6, but beside that?)
1121010618 M * Bertl no, is fixed in rc6 :)
1121010688 M * romke Bertl: so it's rc5 issue?, ugh, so I must go back to rc4 'cause I'll never build rc6-kernel on that guest :|
1121011335 M * Bertl hmm, yeah, probably ... but building on the host would work
1121011393 M * Bertl funny that this issue wasn't reported earlier btw ... it's quite essential ... makes me wonder if the folks around _do_ test the -rc* at all ...
1121011765 Q * flock Ping timeout: 480 seconds
1121011880 J * doum doum@64.235.209.226
1121011926 M * doum hello folk
1121012017 M * Bertl welcome doum!
1121012110 M * Bertl okay, dinner time ... back later ...
1121012116 N * Bertl Bertl_oO
1121012386 N * doum Shuri
1121012556 Q * Shuri Remote host closed the connection
1121013083 M * gin_lit hi all
1121013598 N * Bertl_oO Bertl
1121013613 M * Bertl hey gin_lit!
1121013656 M * gin_lit I have a little question :)
1121013670 M * gin_lit may you answer to it?
1121013848 M * Bertl I guess so ... :)
1121013872 M * gin_lit How I can you disk limits for vps?
1121013879 M * gin_lit how I can configure it?
1121013935 M * Bertl http://linux-vserver.org/index.php?page=Quota+and+Disk+Limits
1121013948 M * Bertl http://vserver.13thfloor.at/Linux2.6/index.php?page=Per+Context+Disk+Limits
1121013958 M * gin_lit yes, I read it.
1121013964 M * Bertl (tools changed slightly, but the basic concept is the same)
1121013982 M * gin_lit what should I patch to make it work?
1121014003 M * Bertl depends on 2.6 you already have everything patched in with the linux-vserver patch
1121014017 M * Bertl on 2.4 you need to add the quota patch ...
1121014020 M * gin_lit i.e. I think 'mount' do not know about option tagctx
1121014036 M * Bertl try tagxid
1121014041 M * gin_lit hm
1121014060 M * gin_lit yes :) it works
1121014187 M * gin_lit it is one of things which I want to work 
1121014298 M * Bertl yeah, I remember, you are the "2-year virtuozzo-alternative" guy, right?
1121014308 M * gin_lit :) yes
1121014339 M * Bertl I guess you have to wait a little for the ngnet to catch up ...
1121014387 M * gin_lit actualy I do not need it anymore
1121014420 M * gin_lit I have found hide_netif option :)
1121014446 M * Bertl ah, good!
1121014470 M * gin_lit :)
1121014599 M * gin_lit one another this which I want is COW filesystem
1121014761 M * Bertl ah, well, have you got a good one? :)
1121014776 M * gin_lit which?
1121014786 M * Bertl yeah, that's the main question ...
1121014834 M * Bertl what we have right now is unification, which is actually very powerful and sufficient for almost all cases (if used properly)
1121014852 M * Bertl in the near future we will add cow-breaking of unified inodes
1121014867 M * Bertl (a patch for doing that already exists)
1121014908 M * gin_lit ah
1121014912 M * gin_lit when it be able for testing?
1121014926 M * Bertl except for that, a fully fledged COW filesystem is a) too heavy and b) too intrusive to use for the different filesystems ...
1121014950 M * gin_lit hm
1121014968 M * Bertl gin_lit: well, if you want to test it, I guess I can make it available pretty soon
1121014981 M * gin_lit :)
1121014995 M * gin_lit yes, I want to test it :)
1121015073 M * gin_lit ah
1121015100 M * gin_lit forgot to ask
1121015115 M * gin_lit how I can use separate uptime for each vps?
1121015127 M * gin_lit i.e. /proc/uptime
1121015134 M * Bertl virt_uptime
1121015149 M * gin_lit ?
1121015166 M * Bertl it's a flag like the hide_netif
1121015176 M * gin_lit hm
1121015272 M * gin_lit can I find somewhere a short description of all flags? :)
1121015302 M * Bertl sure, sec
1121015326 M * Bertl http://www.nongnu.org/util-vserver/doc/conf/configuration.html
1121015333 M * Bertl (this is the great flower page)
1121015362 M * Bertl it has some hyperlinks to the relevant C files (for flags/caps)
1121015370 M * Bertl http://savannah.nongnu.org/cgi-bin/viewcvs/util-vserver/util-vserver/lib/cflags-v13.c?rev=HEAD
1121015372 M * gin_lit yes, i saw this page :)
1121015376 M * Bertl (those are the flags)
1121015400 M * gin_lit but this page is too hard for read
1121015413 M * Bertl hmm, why?
1121015432 M * gin_lit css :)
1121015453 M * Bertl well, use a 'boring' style for example (there are several)
1121015476 M * gin_lit :) I just copied it to file
1121015980 M * gin_lit tagxid option cannot be used with remount :(
1121015989 M * Bertl yep, for good reasons
1121016028 M * gin_lit so, I cannot use / for vpses :(
1121016039 M * Bertl and you better do not ...
1121016079 M * gin_lit yes, but I want to test it on my home pc
1121016089 M * Bertl actually you _can_ use tagxid for /, but it is really not encouraged to do so ...
1121016105 M * Bertl (and you ahve to figure out how yourself :)
1121016927 M * gin_lit Bertl: thanks for everything, you are realy help me :)
1121016973 M * gin_lit and now I am going to my bed :)
1121016978 M * Bertl you're welcome!
1121017001 M * Bertl good night then!
1121017020 M * gin_lit thanks :)
1121017105 J * eXplasm2 explasm@p549F4295.dip.t-dialin.net
1121017505 Q * explasm__ Ping timeout: 480 seconds
1121022017 Q * yarihm Quit: Leaving
1121022459 M * maharaja hi bertl
1121022476 M * maharaja i've got some information regarding the crashes i encounter
1121022670 M * maharaja http://raoul.bhatia.at/~raoul/oops_latest.txt
1121022713 M * Bertl hey maja!
1121022750 M * Bertl hmm, interesting processes ...
1121022813 M * Bertl and you are absolutely sure that the hardware isn#t broken?
1121022846 M * maharaja as stated before, i'm having problesm with 2 different systems (except for the harddisks and the harddisk cables)
1121022869 M * maharaja both p4 thou, one with 1.5ghz (first p4 version) and the second one with 3.2 ghz (bought 2months ago)
1121022970 M * Bertl and as the name suggests, you use vanilla+vserver patches only?
1121022983 M * maharaja true
1121022995 M * Bertl what filesystems are used?
1121023001 M * maharaja ill copy the kernel's .config 
1121023007 M * maharaja ext3 and xfs
1121023016 M * Bertl any crypto loop?
1121023042 M * maharaja and the normal devpts/proc/sysfs/tmpfs stuff
1121023045 M * maharaja no
1121023069 M * maharaja only md-raid1 and lvm
1121023076 M * Bertl hmm, the only thing which catches my eye is the PREEMPT
1121023130 M * maharaja the config file: http://raoul.bhatia.at/~raoul/config-2.6.12.2-vs2.0-rc5-main
1121023328 M * Bertl hmm, P4, SMP, how many CPUs?
1121023342 J * Rushmoom ~me@80-219-253-246.dclient.hispeed.ch
1121023348 M * Bertl welcome Rushmoom!
1121023357 M * SiD3WiNDR p4 can't do smp!
1121023363 M * SiD3WiNDR :)
1121023382 M * Rushmoom Hi Bertl :) Is this an automagic welcome script, or are you just very friendly? ;-)
1121023401 M * maharaja single p4 3.2ghz with ht enabled, 1gb ram (2x 512mb), asus p4p800-mx motherboard
1121023418 M * SiD3WiNDR Rushmoom: afaik it's not automated :)
1121023432 M * Rushmoom SiD3WiNDR: Then it's friendly. I like that :)
1121023447 M * SiD3WiNDR Rushmoom: yea, same here :)
1121023467 M * SiD3WiNDR Bertl is a very friendly guy, but he wakes up and goes to sleep at the ultra strangest hours ;)
1121023513 M * Rushmoom SiD3WiNDR: So do I :-) (Gone to bed yesterday at 6:00 AM, got up today at around 17:00PM :))
1121023521 M * Bertl Rushmoom: very friendly!
1121023525 M * maharaja is it possible that the serial console "misses" some stuff? because i once got an oops without any call trace
1121023525 M * SiD3WiNDR oh, you fit perfectly together :p
1121023562 M * Bertl maharaja: yes, it's thinkable (if the serial rate is too high) but it looks more like your kernel corrupts it's structures
1121023609 M * maharaja Bertl: which might be because of ... ?
1121023614 M * Rushmoom Ok, I got question now: I seem to have messed up my VServer environment (Gentoo as host and guest). How do I properly reomve existing vservers?
1121023616 M * Bertl evening SiD3WiNDR, btw!
1121023617 M * maharaja you said defective hardware
1121023642 M * maharaja any other possible reason?
1121023642 M * Bertl maharaja: either that or one of the famous PREEMPT issues (i.e. xfs still not being PREEMPT safe)
1121023655 M * Rushmoom Is it safe to just delete /vservers/*, or will there be any config files left?
1121023660 M * maharaja so shall i remove the preempt stuff?
1121023671 M * maharaja Rushmoom: take a look into /etc/vservers
1121023692 M * maharaja and look at any symbolic links that might be there
1121023696 M * maharaja and use "ls -al"
1121023701 M * Bertl Rushmoom: it is safe, but there will be several config/run files
1121023720 M * Rushmoom maharaja: Ahh, thanks. So these configs and the actual vservers are the only two places where stuff is kept?
1121023751 M * Bertl /var/run/vserver will contain some files too
1121023774 M * maharaja Bertl: if you want to take a look at the dmesg output: http://raoul.bhatia.at/~raoul/dmesg-2.6.12.2-vs2.0-rc5-main
1121023798 M * Bertl ah, you are using a really funny mix ...
1121023815 M * maharaja Bertl: me or rushmoom?
1121023824 M * Bertl maharaja: you!
1121023857 M * maharaja Bertl: mhm, ok ... :)
1121023928 M * Bertl okay, you get the crash somewhat relable?
1121023933 M * Bertl *reliable even
1121023944 M * maharaja please rephrase your question
1121023963 M * Bertl I assume your system crashes now and then, right?
1121023981 M * maharaja yes
1121024005 M * Bertl is this when you a) do something special, b) when you wait some time, c) you don#t know/care
1121024006 M * maharaja most of the time, it crashes while doing fs io
1121024022 M * maharaja but i cannot reproduce this issue
1121024026 M * maharaja not willingly
1121024050 M * maharaja the system may crash while writing the http.conf file (inside a vserver on xfs) with vi
1121024062 M * Bertl okay, how long does it take to crash, inaverage?
1121024068 M * maharaja or copying 200mb of files
1121024073 M * Bertl s/in/in /
1121024078 M * maharaja other times i can do that without any problem
1121024079 M * maharaja +s
1121024082 M * maharaja one sec
1121024153 M * maharaja http://raoul.bhatia.at/~raoul/reboot-2.6.12.2-vs2.0-rc5-main
1121024196 M * maharaja the reboots at july 9th in the morning (02:00 - 06:31) where because of me trying to find the cause of our problem
1121024216 M * Bertl hmm, okay, and your primary concern is to a) stabilize the system or b) figure what is actually wrong ...
1121024237 M * maharaja a) first, then b)
1121024261 M * Bertl hmm .. well I guess you have to decide for one ... because the approaches are different
1121024287 M * maharaja a) is definetly my first choice
1121024292 M * Bertl for a) I'd suggest a bunch of changes which probably paper over the issue (i.e avoid it)
1121024321 M * Bertl for b) I'd suggest some methods to intentionally trigger the issue
1121024349 M * maharaja for the next days, i would like your suggestions for a)
1121024376 M * maharaja ill then try to free another p4 and am willing to help finding the cause
1121024390 M * Bertl okay, 1) disable highmem, enable the 2/2 split
1121024418 M * Bertl 2) disable the local apic and SMP support, switch the SMT off in the bios
1121024439 M * maharaja smt = hyperthrading, right?
1121024443 M * Bertl yep
1121024449 M * Bertl disable the PREEMPTION
1121024477 M * Bertl that was 3) and 4) disable the IRQ balancing code
1121024525 M * Bertl 5) disable ACPI and PM
1121024538 M * Rushmoom Another question: Is it safe to use the gentoo-initstyle (instead of plain), when I'm happy with it? Or is it *really* deprecated and shouldn't be used anymore?
1121024557 M * Bertl no, you can use whatever you prefer ...
1121024581 M * Rushmoom Bertl: Great, thanks. It's just that I made way better experiences with the gentoo-initstyle
1121024604 M * Bertl 6) disable CRYPTO LOOP
1121024609 M * DaPhreak Rushmoom: you aren't alone ... the plain stuff does sometimes not even start
1121024652 M * Bertl maharaja: which IDE chipset do you use for the disks?
1121024653 M * Rushmoom And my credo is "When I don't need a special feature of upgrade XY, then I don't need to upgrade"..
1121024662 M * Rushmoom DaPhreak: Same here :)
1121024675 M * Bertl Rushmoom: what version do you use right now (kernel patch)?
1121024686 M * maharaja Bertl: IDE interface: Intel Corp. 82801EB/ER (ICH5/ICH5R) Ultra ATA 100 Storage Controller (rev 02)
1121024711 M * Rushmoom Bertl: linux-2.6.11.11-vserver-2.0-rc4
1121024751 M * Rushmoom Bertl: Just compiling linux-2.6.11.12-vserver-2.0-rc4-r1 though
1121024872 M * Rushmoom I'm also not sure which are the current "stable" docs for Gentoo: The HOWTO or the installation part in the handbook
1121024881 M * maharaja Bertl: i changed the settings according to your suggestions
1121024901 M * Rushmoom IOW: http://www.gentoo.org/doc/en/vserver-howto.xml or http://gentoo.home.xnull.de/doc/en/vshandbook/index.xml?part=2&chap=1
1121024977 M * Bertl Rushmoom: what is rc4-r1?
1121025020 M * Rushmoom Bertl: The version of the vserver kernel patch, I thought. Isn't it?
1121025062 M * DaPhreak Bertl: its a sub-version of -rc4, i guess rc4 was with the bad xfs ..
1121025069 M * Bertl hmm, never had -r1, we are now at vs2.0-rc6
1121025082 M * Rushmoom Bertl: http://packages.gentoo.org/search/?sstring=vserver-sources
1121025096 M * Rushmoom Bertl: Maybe something Gentoo-specific...?
1121025103 M * DaPhreak Rushmoom: yeah
1121025114 M * Bertl ah, okay :)
1121025147 M * Bertl why just for x86/64 ?
1121025188 M * Rushmoom Bertl: I guess the userbase for "VServer on Gentoo" is just way to small to maintain several different architectures...
1121025219 M * DaPhreak Bertl: just because noone, who owns another arch (like ia64, hppa, sparc ..) has tested it yet
1121025230 M * Rushmoom Bertl: But it also doesn't mean that it doesn't work, it's just not tested yet...
1121025286 M * Bertl aha .. i.c.  well, I think the kernel _is_ tested on various archs ... and IIGTR, gentoo just provides 'sources' no?
1121025324 M * Rushmoom Bertl: Yes, AFAIK these are vanilla sources (kernel.org) + Gentoo-specific patches + VServer patches
1121025350 M * maharaja Bertl: shall i recompile the kernel with these changes and do some testing?
1121025356 M * Bertl DaPhreak: so who is maintaining those distros?
1121025366 M * Bertl s/distros/distro package/
1121025373 M * Rushmoom Bertl: Hollow i guess
1121025414 M * DaPhreak Bertl: http://www.gentoo.org/cgi-bin/viewcvs.cgi/sys-kernel/vserver-sources/vserver-sources-2.0_rc4-r1.ebuild?view=markup
1121025419 M * DaPhreak look into the header
1121026970 Q * AprilDL Read error: Connection reset by peer
1121026979 J * AprilDL___ ~chatzilla@ip68-9-200-247.ri.ri.cox.net
1121026980 N * AprilDL___ AprilDL
1121027303 M * maharaja *compiling new kernel*
1121028266 M * Medivh hi
1121028275 M * Medivh which util-vserver do i need for 2.0-rc6...?
1121028458 M * Bertl at least util-vserver 0.25 IIRC
1121028480 M * Bertl but I'd suggest to use 0.30.207 or later
1121029167 M * Rushmoom The Gentoo VServer Handbook says, that the host system needs a special baselayout (baselayout-vserver)
1121029187 M * Rushmoom But in the installation process, this baselayout is only unmasked, but never installed
1121029211 M * Rushmoom So if I follow the instructions precisely, I end up with the regular baselayout. Is this correct?
1121029231 M * Rushmoom (Handbook: http://gentoo.home.xnull.de/doc/en/vshandbook/index.xml?part=2&chap=1)
1121029390 M * Bertl (probably some gentoo folks can answer that :)
1121029423 M * Rushmoom Bertl: I hope so :-) (Hollow, are you there..? *g*) 
1121029455 M * Rushmoom Also, the vserver/x86 Profile stated in the docs does not exist. 
1121030181 M * Hollow Rushmoom: vserver-howto.xml is curernt
1121030200 M * Hollow and the vserver/x86 profile exists here too
1121030621 M * Rushmoom Hollow: Thanks for your reply! But is the vserver/x86 Profile only contained in your Stage3 tarball?
1121030633 M * Hollow ne, it's in portage cvs
1121030640 M * Hollow and on all mirrors respectivly
1121030656 M * Rushmoom Hollow: So it's just a matter of downloading it and setting the right symlink?
1121030677 M * Hollow Rushmoom: it should be in /usr/portage/profiles/vserver/x86 if your tree is up-to-date
1121030752 M * Rushmoom Hollow: Ah, right, thanks for the hint. I thought it should show up after a ls -F /etc/make.profile/
1121030834 M * Hollow  /etc/make.profile is a symlink to the profile you're using..
1121030842 M * Hollow e.g.:
1121030866 M * Hollow cd /etc; rm /etc/make.profile; ln -s ../usr/portage/profiles/vserver/x86 make.profile
1121030933 M * Rushmoom Thanks a lot for clearing this up for me :)
1121030942 M * Rushmoom But what about the baselayout?
1121030955 M * Rushmoom When I use your stage3-tarball, it's already contained, IIRC
1121030972 M * Rushmoom But when doing a stage2, I have to emerge it seperately, I guess?
1121031043 M * Hollow Rushmoom: every stage contains baselayout-vserver, but if you're using a stage2 "emerge system" will probably install it again
1121031088 M * Hollow Rushmoom: and btw: the handbook states to emerge baselayout.. you don't need to, but if you do, you have to "emerge baselayout-vserver"
1121031094 M * Hollow instead of the original baselayout
1121031096 M * Rushmoom Hollow: It did not, IIRC. Although I unmasked baselayout-vserver...
1121031108 M * Hollow baselayout-vserver is not masked...
1121031111 J * Aiken ~james@tooax8-067.dialup.optusnet.com.au
1121031131 M * Rushmoom Hollow: Yes, I saw that, and changed it accordingly from "emerge baselayout" to "emerge baselayout-vserver"
1121031135 M * Hollow if you're running stable the only thing you need is to keyword baselayout-vserver with ~x86
1121031137 M * Bertl evening Aiken!
1121031153 M * Bertl evenin Hollow, btw! :)
1121031163 M * Hollow evening Bertl ;)
1121031183 M * Rushmoom Hollow: Then it said, that baselayout is blocking baselayout-vserver, which makes perfect sense...
1121031188 M * Hollow yep
1121031214 M * Aiken hello
1121031220 M * Rushmoom Hollow: So I did "emerge -C baselayout && emerge baselayout-vserver", which massively broke things :(
1121031273 M * Hollow huh? why could you emerge -C baselayout if it's not installed?
1121031287 M * Hollow which stage did you use?
1121031295 M * Rushmoom Hollow: It *WAS* installed... Stage2
1121031306 M * Hollow vserver stage2 or generic stage2?
1121031311 M * Rushmoom Hollow: generic
1121031314 M * Hollow bad
1121031318 M * Rushmoom Oh :)
1121031338 M * Hollow that's why i'm providing vserver stages... because you seriously fuck up baselayout without ;)
1121031371 M * Rushmoom Hollow: Hehe, that also was my conclusion after these hassles ;-)
1121031373 M * Hollow the stages are available at http://dev.gentoo.org/~hollow/vserver/
1121031386 M * Rushmoom Hollow: BTW, would you mind also providing MD5 sums for the Tarballs?
1121031420 M * Rushmoom Hollow: Not because of a lack of trust, of course, only for checking the integrity of the archives...
1121031424 M * Hollow Rushmoom: sure, np... there will be new stages for 2005.1 soon anyway
1121031444 M * Rushmoom Hollow: No hurry :)
1121031456 M * Hollow there are still some minor bugs in the vserver stages which will be fixed too
1121031563 M * Rushmoom Hollow: So, to sum up, you would recommend vserver-howto.xml + Your stage3 tarball + keyword vserver-baselayout
1121031574 M * Rushmoom Hollow: Is that about it?
1121031585 M * Hollow yeppa.. though my stages already keyword baselayout-vserver iirc ;)
1121031622 M * Rushmoom Hollow: Ah, ok :) Adapting the HOWTO to initstyle=gentoo should cause any trouble, should it?
1121031664 M * Hollow baselayout-vserver uses plain initstyle
1121031693 M * Rushmoom Oh, the initstyle depends on the used baselayout?
1121031753 M * Rushmoom So you shouldn't mix baselayout-vserver + gentoo-initstyle?
1121031766 M * Hollow yes and no.. ;)
1121031777 M * Hollow gentoo init style should work too
1121031811 M * Hollow but it's an ugly hack imo
1121031843 M * Rushmoom Gentoo-initstyle in general, or in combination with baselayout-vserver
1121031884 M * Rushmoom I just had major trouble with the plain initstyle, and don't see any need (for me) to abandon the gentoo-style...
1121031891 M * Hollow sec..
1121031911 M * Rushmoom np
1121032077 M * Rushmoom Thanks for the quick reaction on the MD5 sums BTW ;-)
1121032091 M * Hollow Rushmoom: heh, np
1121032119 M * Hollow ok.. so a little explanation about the initstyles
1121032165 M * Hollow the vserver script calls a progam inside the context and chroot to boot up the server, as a normal machine would, but this can be done in multiple ways:
1121032191 M * Hollow the default is using init, and let it handle the system boot through inittab
1121032210 M * Hollow sysv directly calls sysv init rc scripts without init
1121032237 M * Hollow the gentoo init style however just called /sbin/rc default, to use the gentoo specific rc program to boot the server
1121032268 M * Hollow this however means, that every command issued by the baselayout before the default runlevel, are simply skipped
1121032323 M * Hollow so i adapted the baselayout to be able to boot with init and inittab but without all errors which existed before (mainly caused by mount etc which are not allowed inside vservers)
1121032376 M * Rushmoom Ah, this explains why I was getting "local filesystems failed to mount" and stuff when using the wrong baselayout...
1121032384 M * Hollow yep...
1121032421 M * Bertl Hollow: got a minute for me?
1121032454 M * Hollow Bertl: sure
1121032481 M * Bertl DaPhreak, IIRC, pointed me to http://packages.gentoo.org/search/?sstring=vserver-sources
1121032498 M * Bertl and I saw that only x86/64 is 'available'
1121032530 M * Bertl my question: would it be hard to add 'support' for more archs?
1121032549 M * Hollow nope... i'll file an arch bug to add the keywords
1121032573 M * Bertl I mean not all archs are tested yet, but most of them are
1121032620 M * Bertl alpha is supported with rc6, hppa, ppc, s390, sparc should work since 2.0
1121032629 M * Hollow Bertl: in theory all archs supported yb gentoo can be keyworded
1121032652 M * Hollow which tests need to be done to ensure vserver is working on that arch?
1121032671 M * Bertl first level tests are testme.sh and testfs.sh
1121032743 M * Bertl second level is probably a succesful guest startup ...
1121032757 M * Bertl (we still have no complete test/regression suite)
1121032789 M * Hollow ok, so test{me,fs}.sh should be enough for keywording it ~ARCH
1121032823 M * Bertl I guess so, where testfs.sh is not suited for fully automate testing yet
1121032830 M * Bertl *automated
1121032996 M * Bertl (but it will be soon)
1121033130 M * Hollow where can i find these scripts?
1121033151 M * Rushmoom http://vserver.13thfloor.at/Stuff/SCRIPT/testme.sh
1121033161 M * Hollow thx
1121033163 M * Rushmoom http://vserver.13thfloor.at/Stuff/SCRIPT/testfs.sh
1121033167 M * Rushmoom I guess ;-)
1121033172 M * Bertl yep
1121033196 M * Bertl testme.sh should be called with -L
1121033339 M * Aiken Bertl there is still the alpha problem of dietlibc + the tools
1121033366 M * Bertl hmm, somebody reported percpu issues with modules?
1121033367 M * Aiken I had subscribed to the dietlibc mailing list and no one replied to my message to that list
1121033386 M * Bertl Aiken: hmm, yes the chbind issues, right?
1121033396 M * Bertl (with the bad alignment)
1121033436 M * Aiken the order of varibales in the sigaction struct
1121033452 M * Bertl hmm, sigaction too?
1121033462 M * Bertl ah, yes, I remember now!
1121033471 M * Bertl we are talking of different things ...
1121033505 M * Bertl yes, the sigaction struct was wrong in 'mainline' no idea if it got fixed yet
1121033565 M * Bertl but I'm see those on my test machine ...
1121033571 M * Bertl chbind(12239): unaligned trap at 0000000120003b24: 000002000000007a 2c 0
1121033577 M * Bertl chbind(12239): unaligned trap at 0000000120001af8: 0000020000000072 2d 31
1121033583 M * Aiken I have dietlibc from cvs and it has not been changed
1121033595 M * Aiken an unpatched ping does that as well
1121033613 M * Bertl maybe an email to fefe would be a good idea?
1121033680 M * Hollow Bertl: can you give me an output of a successfull testfs.h?
1121033851 M * Aiken surpose it could be
1121033877 M * Aiken how do you trigger the chbind analigned trap error?
1121033945 M * Bertl Hollow: here an example for ext2 http://vserver.13thfloor.at/Stuff/testfs.sh_out.txt
1121033956 M * Bertl Aiken: I just run my testme.sh
1121033999 M * Aiken just sh testme.sh ?
1121034006 M * Aiken I get a clean run
1121034006 M * Bertl yep
1121034018 M * Hollow http://bugs.gentoo.org/show_bug.cgi?id=98602
1121034032 M * Bertl Aiken: interesting, did you look in dmesg?
1121034080 M * Aiken no unaligned trap errors shown with dmesg
1121034092 Q * jkl_ Ping timeout: 480 seconds
1121034157 M * Bertl Aiken: well, I get three of them on every run :)
1121034162 M * Bertl no, actually 4 :)
1121034272 M * Aiken just grabbed the latest testme.sh, no errors
1121034304 M * Bertl guess it depends on the tools ... and how they are compiled
1121034356 M * Aiken tools .30.207, dietlibc .29, gcc 3.3.6
1121034391 M * Bertl let me check something, and I'll upload my vserver-info - SYSINFO in a few minutes
1121034576 M * Aiken the utils I am using reference 2.6.11.11-vs2.0-rc4
1121034616 M * Bertl as kernelheaders, I presume?
1121034642 M * Aiken yes
1121034664 M * Aiken http://pastebin.com/310875
1121034714 M * Bertl 428/kernel vs #
1121034715 M * Bertl 428/glibc
1121034724 M * Bertl seems to be the only relevant difference ...
1121034843 M * Bertl http://pastebin.com/310876
1121034855 M * Aiken my configure line could have been something like      ./configure --prefix=/usr --sysconfdir=/etc
1121034940 M * Aiken how did you get 428/kernel ?
1121034958 M * Bertl well, it somehow happened :)
1121035062 M * Rushmoom Hollow, may I bother you once again with a short question?
1121035083 M * Hollow Rushmoom: sure
1121035110 M * Rushmoom I'm just doing a stage3 install (for a vserver host) with your stage3 tarball
1121035133 M * Rushmoom But the contained baselayout is definitely baselayout instead of baselayout-vserver
1121035171 M * Rushmoom And baselayout-vserver is masked
1121035227 M * Rushmoom So shuold I just do echo "sys-apps/baselayout-vserver ~x86" >> /etc/portage/package.keywords
1121035240 M * Rushmoom And then "emerge baselayout-vserver"?
1121035331 M * Rushmoom Because, the HOWTO doesn't mention anything that has to be unmasked...
1121035349 M * Bertl Hollow: ah, before I forget, the testfs.sh requires a simple script called ewrite
1121035410 M * Rushmoom Hollow: Oh, damn, after unmasking it I see, baselayout-vserver-1.11.12-r4 actually IS contained in the stage. Sorry!
1121035412 M * Bertl (I will remove that dependancy in the next release though)
1121035438 M * Rushmoom But it's not unmasked by default in the stage-archive...
1121035737 M * Aiken /usr/include/asm/unistd.h:#define __NR_vserver                  428
1121035741 M * Aiken might be why it says glibc for me
1121035786 M * Bertl could be, what glibc do you use?
1121035799 M * Bertl (or to be precise, kernel headers)
1121035827 M * Aiken 2.3.5 with the raw headers from 2.6.8.1
1121035859 M * Bertl hmm, so unmodified kernel headres?
1121035866 M * Bertl *headers 
1121035870 M * Aiken yes
1121035880 M * Bertl ah, well, don't do that :)
1121035896 M * Bertl but yeah, that probably explains it
1121035981 M * Hollow Rushmoom: sorry... babes need care, you know.. ;)
1121035986 M * Hollow everything ok now?
1121036014 M * Rushmoom Yes. Yes. ;-)
1121036029 M * Hollow fine :)
1121036030 M * Bertl Hollow: babes?
1121036040 M * Hollow Bertl: yep, even i have one ;)
1121036058 M * Bertl babes is plural, no?
1121036062 M * Rushmoom Hollow: And your not talking about hardware here, right? ;-))
1121036070 M * Hollow Bertl: was just generally speaking ;)
1121036088 M * Hollow Rushmoom: heh
1121036100 M * Hollow Rushmoom: there are truly two kind of babes, yeah :P
1121036119 M * Aiken Bertl means changing something that has worked for me for many years :)
1121036139 M * Aiken aren't the sanitized header a recent thing?
1121036154 M * Rushmoom A friend of mine often got quoted for saying "My girlfriend is sick", which usually meant "My linux is broken"... >:-)
1121036161 M * Hollow heh
1121036214 M * Bertl Aiken: well, I know, I know, if I don't want to mess with any *libc I also do it ... but, the kernel folks are right in saying that kernel headers are for the libc only ... and not for userspace (yet)
1121036219 M * Rushmoom Hollow: Anyway, tell her nice greetings from Switzerland ;-)
1121036228 M * Hollow i'll do ;)
1121036345 M * Hollow Bertl: apropos libc... i'm still lost with the errno fsck up with shared objects...
1121036420 M * Aiken I surpose it is easy enough to grab a copy of sanitized headers and put them in place and rebuild the tools to see what happens
1121036451 M * Hollow don't the tools ship the headers themselves?
1121036491 M * Bertl Hollow: we can investigate this, if you insist on using glibc's errno handling
1121036524 M * Hollow best would be that it supports both diet and glibc...
1121036526 M * Hollow imo
1121036555 M * Bertl yes, agreed, and it does, given that you get some defines right
1121036577 M * Bertl that's why I didn't hard code the errno handling in the first place
1121036598 M * Hollow which defines would that be?
1121036688 M * Bertl #define __syscall_error(e) do { errno = (e); } while(0)
1121036704 J * renihs ~renihs___@193.170.52.70
1121036715 M * Hollow this is the current define in syscall.h right?
1121036718 M * Bertl if you define that to handle pic/get_errno_location stuff, then it will work quite fine I guess
1121036733 M * Bertl you just define it _before_ you include the syscall.h
1121036740 M * Bertl welcome renihs!
1121036742 M * Hollow k...
1121036784 M * Bertl Hollow: you can try if that fixes your issues by defining a global variable called sys_errno and using that instead
1121036793 M * renihs hi Bertl!
1121036803 M * Bertl Hollow: (the global variable must be part of the library)
1121036839 M * Hollow Bertl: i.e. __syscall_error(e) do { sys_errno = (e); } while(0)
1121036873 M * Bertl yep, with a static sys_errno (static in library sense)
1121036911 M * Bertl and as I previously suggested, don#t interleave the vserver macro with 'normal' code
1121036925 M * Bertl gcc get's that wrong too often (especially on x86)
1121036988 M * Hollow e.g.?
1121037000 M * Bertl make a wrapper for each syscall command
1121037021 M * Bertl which takes the required arguments, and does the call
1121037029 M * Bertl then returns the result, nothing else
1121037074 M * Hollow that's how we do it.. ;)
1121037083 M * Bertl okay, then it should be fine :)
1121037101 A * Bertl notes to have a look at the code soon :/
1121037251 M * Hollow Bertl: souldn't be hard..
1121037258 M * Hollow it's not that much code yet
1121037817 Q * renihs Read error: Connection reset by peer
1121037909 J * renihs ~renihs___@193.170.52.70
1121038133 M * Hollow ha!
1121038159 M * Hollow Bertl: i just copied the vserver() function from util-vserver using the standard syscall() function... will this cause problems?
1121038359 M * maharaja bertl: new kernel oops message (i recompiled the kernel as you asked)
1121038360 M * maharaja http://raoul.bhatia.at/~raoul/kernel_panic_new.txt
1121038374 M * maharaja cannot say though, if that was because i forgot to disable ht in the bios
1121038405 Q * AprilDL Quit: Chatzilla 0.9.67 [Mozilla rv:1.8b2/20050226]
1121038609 M * Hollow off to bed... back tomorrow
1121038621 M * maharaja me too
1121038628 M * Bertl okay, night then!
1121038880 M * Rushmoom Yep, good night!
1121039171 M * Aiken using the linux-libc-headers that lfs uses (that matches what my glibc was compiled with) the tools don't compile
1121039252 M * Aiken it dies when kernel/dlimit.h includes linux/spinlock.h
1121039496 M * Bertl funny, heh ...
1121039514 M * Bertl well, actually it should the headers from dietlibc, no?
1121039583 M * Bertl but if the 'kernel headers' work for you, I guess this is fine too
1121039593 M * Aiken I would assume so but it is still including the normal headers
1121039623 Q * eXplasm2 Remote host closed the connection
1121039716 M * Aiken watching the build, most is with diet but some parts of the build call gcc directly
1121039795 M * Aiken http://pastebin.com/310923
1121039812 M * Aiken that is how the build dies