1188779100 Q * bonbons Quit: Leaving
1188780302 J * friendly12345 ~friendly@ppp59-167-75-40.lns1.mel6.internode.on.net
1188783719 J * DoberMann_ ~james@AToulouse-156-1-3-154.w81-250.abo.wanadoo.fr
1188783828 Q * DoberMann[ZZZzzz] Ping timeout: 480 seconds
1188784830 M * hparker Looking in /proc/virtual/name/limit, is VMM swap in use?
1188784883 M * Bertl no, VM is the virtual memory or address space allocated
1188784994 M * hparker Ok... i'm showing a hit on RSS, nothing on VM, RSS hard limit of 130000 and gcc blew up during compile :(
1188785013 M * Bertl sounds fine, hard limit == ram+swap
1188785038 M * Bertl i.e. you finally got out of memory _and_ swap (as far as the guest is concerned)
1188785045 M * hparker Oh, I thought it was ram only
1188785051 M * Bertl that is the soft limit
1188785071 M * hparker ah hah!!! Thanks!
1188785076 M * daniel_hozac well, RSS is by definition RAM only, no?
1188785101 M * Bertl yes, but that is not how the virtualization shows it
1188785109 M * daniel_hozac right.
1188785110 M * Bertl (VIRT_MEM that is)
1188785146 M * hparker so rss.soft is ram, rss.hard is ram+swap?
1188785159 M * Bertl so yeah, RSS _is_ memory pages only, but rss.soft and rss.hard and rss are different things :)
1188785176 M * hparker And the breakdown is? ;)
1188785182 M * Bertl hparker: yes, with recent patches that is okay
1188785204 M * Bertl hparker: you should see the result with mem inside the guest
1188785204 M * hparker What's rss?
1188785226 M * Bertl rss sets both, rss.soft and rss.hard (basically giving you all ram, no swap)
1188785318 A * hparker plays more
1188786536 J * virtuoso_ ~s0t0na@pppoe-107.58.110.89-adsl.spbnit.ru
1188786947 Q * virtuoso Ping timeout: 480 seconds
1188787339 M * hparker Bertl: What mem command are you speaking of? i'm not familiar with that
1188787381 M * Bertl http://linux-vserver.org/Capabilities_and_Flags
1188787405 M * Bertl (check the flags section)
1188787518 M * Bertl hparker: you can see what is enabled in /proc/virtual/<xid>/status
1188787680 M * hparker thanks
1188787684 M * Bertl np
1188790260 J * coderanger_ ~coderange@c-65-96-210-168.hsd1.ma.comcast.net
1188794245 J * ktwilight_ ~ktwilight@45.194-66-87.adsl-static.isp.belgacom.be
1188794493 Q * ktwilight Read error: Connection reset by peer
1188796374 M * Bertl okay, off to bed now ... have a good one everyone! cya!
1188796379 N * Bertl Bertl_zZ
1188798184 Q * nebuchadnezzar Quit: ERC Version 5.2 (IRC client for Emacs)
1188798221 J * nebuchadnezzar ~nebu@zion.asgardr.info
1188798522 J * pbryan ~pbryan@S010600095baae0ff.vf.shawcable.net
1188798547 M * pbryan Is anyone running Linux-VServer on Ubuntu Fiesty?
1188798583 N * DoberMann_ DoberMann
1188799197 J * sharkjaw ~gab@158.36.44.106
1188799682 J * meandtheshell ~markus@85.127.109.11
1188800841 M * Supaplex pbryan: debian etch here
1188801093 J * dna ~dna@150-241-dsl.kielnet.net
1188801800 J * DavidS ~david@p5481136B.dip0.t-ipconnect.de
1188801951 N * DoberMann DoberMann[PullA]
1188802155 Q * pusling Ping timeout: 480 seconds
1188802803 J * pusling pusling@88.212.70.38
1188803703 J * yarihm ~yarihm@84-74-20-183.dclient.hispeed.ch
1188804927 J * besonen_mobile_ ~besonen_m@71-220-228-198.eugn.qwest.net
1188805322 Q * besonen_mobile__ Ping timeout: 480 seconds
1188807842 J * kir_home ~kir@217.206.93.210
1188808370 J * esa` ~esa@ip-87-238-2-45.adsl.cheapnet.it
1188808382 Q * esa Ping timeout: 480 seconds
1188809142 Q * kir_home Quit: Ухожу я от вас
1188809289 J * Baby ~miry@195.37.62.208
1188809931 J * tobifix tobifix@IVV7SCHAZ.UNI-MUENSTER.DE
1188809939 M * tobifix hello
1188809949 J * dna_ ~dna@150-241-dsl.kielnet.net
1188810030 M * tobifix i have 2.6.22.2-vs2.2.0.3 running. when i add another ip to a vserver, i get the following error: Cannot find device "eth1"
1188810030 M * tobifix SIOCGIFFLAGS: No such device
1188810083 M * tobifix further more i recognized another strange problem. when i install a vserver, the ip disappears on my root host
1188810151 M * tobifix on another host with the 2.6.20.12-vs2.2.0 i have no problems
1188810263 J * Pazzo ~ugelt@reserved-225136.rol.raiffeisen.net
1188810358 Q * dna Ping timeout: 480 seconds
1188810724 J * pmenier ~pmenier@LNeuilly-152-22-72-5.w193-251.abo.wanadoo.fr
1188810953 J * bonbons ~bonbons@2001:960:7ab:0:20b:5dff:fec7:6b33
1188813843 N * Bertl_zZ Bertl
1188813847 M * Bertl morning folks!
1188813859 M * Bertl tobifix: do you have a device eth1?
1188813876 M * Bertl tobifix: disappears means? what does 'ip addr ls' on the Host system show?
1188814087 M * tobifix Bertl, ip addr show, shows me all ips
1188814099 M * tobifix but why aren't they show via ifconfig
1188814139 M * tobifix i also cannot use a device name like eth1
1188814144 M * tobifix just eth0:* works
1188814345 M * Bertl well, first, ifconfig is about 5 years old, and was obsoleted by 'ip addr ls'
1188814356 M * Bertl (actually iproute2)
1188814370 M * tobifix Bertl, ok ;)
1188814372 M * Bertl so it is not very surprising that ifconfig doesn't see everything
1188814407 M * tobifix thanks, but why can i add only interfaces with eth0:*`?
1188814410 M * Bertl of course, if you really _want_ to see stuff (on the host) with ifconfig, you can use the ancient aliases (eth0:x) instead of simple ips
1188814445 M * Bertl second, Linux-VServer does IP isolation, not network stack virtualization (for several reasons, one being performance)
1188814467 M * Bertl thus, you can add whatever IPs you like to any _existing_ host interface
1188814508 M * tobifix aaah
1188814511 M * tobifix ;)
1188814574 M * Bertl so, if you have an eth1 (or you use a dummy0 for example) you can bind new ips to that too
1188814737 M * hparker lo Bertl... getting places here ;)
1188814767 M * Bertl :)
1188814807 M * hparker The link you gave me helped make the flower page make more sense
1188814851 M * Bertl yeah, well, everything is directly accessable via the wiki :)
1188814881 M * hparker Yeah... and i've read it... Most of it looked Greek at the time ;)
1188815097 Q * DavidS Remote host closed the connection
1188815116 J * DavidS ~david@p5481136B.dip0.t-ipconnect.de
1188815696 M * hparker What's the difference in SCHED_(HARD|PRIO|PAUSE)?
1188815741 M * Bertl SCHED_PRIO activates the priority adjustments, based on the Token Bucket state
1188815771 M * Bertl while SCHED_HARD activates hard scheduling (i.e. putting processes on hold when out of tokens)
1188815796 M * Bertl and SCHED_PAUSE actually puts the entire context on hold regardless of the Token Bucket
1188815803 M * hparker ahh
1188815816 M * hparker Makes that an easy decision ;)
1188816198 J * tuxmania ~bonbons@2001:960:7ab:0:20b:5dff:fec7:6b33
1188816202 Q * tuxmania Remote host closed the connection
1188816792 N * virtuoso_ virtuoso
1188817710 Q * Sn00py 
1188818017 P * friendly12345 
1188820050 Q * bzed Ping timeout: 480 seconds
1188820161 M * meandtheshell Hm ... Hollow once had a oneliner to check HDD health and so on ... does anybody remember?
1188820196 M * Hollow smartctl --all /dev/sdX
1188820196 M * Hollow ?
1188820292 M * meandtheshell ah ... let me check ... I guess it was that :)
1188820440 J * bzed ~bzed@devel.recluse.de
1188820652 M * meandtheshell hm ... that shows instant results ... I remember, back then it took some time (~2min or so ... some disk action took place during that time) - however, I'll check the man file
1188820682 M * Hollow yeah, you can run a couple of tests
1188820698 M * Hollow smartctl --test=long /dev/foo
1188820705 M * Hollow it will tell you how long it takes
1188820796 M * meandtheshell yep, just figured - /me puts this stuff into his notepad else, a couple of month from now I would have to ask #vserver again :)
1188820845 M * meandtheshell my disk made scary noises - that stuff scares my like heel man
1188820871 M * meandtheshell 5 years old HDDs ...
1188820968 J * [Akuma] ~Akuma_Fan@58.37.192.221
1188821009 M * Bertl welcome [Akuma]!
1188821026 M * [Akuma] hi
1188822372 Q * Aiken Remote host closed the connection
1188823208 J * Julius ~julius@p57B27564.dip.t-dialin.net
1188823218 M * Bertl wb Julius!
1188823226 M * Julius hi
1188823869 Q * [Akuma] Quit: Leaving
1188823932 Q * Julius Ping timeout: 480 seconds
1188824026 J * Julius ~julius@p57B27564.dip.t-dialin.net
1188825735 Q * sharkjaw Quit: Leaving
1188825940 M * Bertl time for a nap ... back later!
1188825945 N * Bertl Bertl_zZ
1188827794 M * glen_ huh. i can't get odccm/synce-gnome/pls working again
1188827817 M * glen_ in what order should the things be done? start hald, start dbus, start dbus-session, start odccm, start synce-gnome?
1188827827 M * glen_ at least it used to ask device password, but now it doesn't
1188827851 M * glen_ also i get this error when starting synce-gnome program:
1188827851 M * glen_  Introspect error: Process /usr/lib/notification-daemon exited with status 1
1188827873 M * glen_ sorry. wrong channel ;(
1188827876 J * renihs ~penguin@83-65-34-34.arsenal.xdsl-line.inode.at
1188827887 M * renihs morning :p
1188827972 M * renihs trying to patch vserver sources with aufs (needed for livecd) but somehow it bails out with too few arguments at http://pastebin.ca/679924
1188827977 M * renihs should i just add a null argument?
1188827978 M * renihs :p
1188828124 M * renihs ah now it works
1188828127 M * renihs never mind
1188828142 M * renihs now i only need to patch in lzma support for squashfs :p
1188828145 M * daniel_hozac if you don't have an nd, i guess so.
1188828152 M * renihs nd?
1188828174 M * daniel_hozac note that it means that BME won't work for those filesystems.
1188828183 M * renihs hmm whats BME?
1188828195 M * renihs might be dumb question :p
1188828196 M * daniel_hozac bind mount extensions.
1188828199 M * renihs oh
1188828207 M * renihs that would be rather bad
1188828216 Q * tobifix 
1188828216 M * daniel_hozac i.e. mount -o bind,ro etc.
1188828217 M * renihs since it usually bindmount them :p
1188828220 M * renihs i
1188828227 M * renihs hmmm 
1188828244 M * renihs ahno i dont 
1188828250 M * renihs losetup -> then mount
1188828256 M * renihs thats not bindmount is it?
1188828263 M * daniel_hozac no.
1188828266 M * renihs mkay
1188828283 M * renihs hmm i had to drop unionfs, completely broken
1188828297 M * renihs but i added NULL stuff on unionfs too i think and bindmounts worked (i think)
1188829156 M * renihs cool, seems to work nicely now
1188829802 Q * badari2 Ping timeout: 480 seconds
1188829824 J * badari2 ~badari@bi01p1.co.us.ibm.com
1188830229 M * renihs daniel_hozac, mind taking a look at http://www.ips.at/bla/vm-installed.packages and tell me if i miss any specific vserver packages or stuff you dont want to miss on a vserver? :p
1188830490 Q * Julius Quit: Verlassend
1188830492 J * julius_ ~julius@p57B27564.dip.t-dialin.net
1188832428 J * Piet ~piet@tor.noreply.org
1188833243 J * balbir ~balbir@217.206.93.210
1188834357 Q * ensc Ping timeout: 480 seconds
1188835195 Q * pbryan Ping timeout: 480 seconds
1188835444 M * daniel_hozac renihs: what packages to install depends entirely on the purpose...
1188835688 J * besonen_mobile__ ~besonen_m@71-221-93-100.eugn.qwest.net
1188836089 Q * besonen_mobile_ Ping timeout: 480 seconds
1188836442 M * bonbons daniel_hozac: 2.2.0.3 seems to omit pid = 1 mapping for guest's init... ls /proc/1/ inside of guest does not exists, but there is /proc/x/ where X is the real pid of the guest's init
1188836458 M * daniel_hozac what kernel?
1188836518 M * daniel_hozac are the correct flags set?
1188836660 M * epicbjorn there's no /proc/1 here either, come to think about it
1188836784 J * jmcaricand ~jmcarican@d77-216-236-212.cust.tele2.fr
1188837194 M * daniel_hozac looks like a missing hunk in proc_pid_lookup.
1188837327 M * daniel_hozac and proc_pid_readdir.
1188837355 M * daniel_hozac if you list the real init's pid, it shows as pid 1, right?
1188837363 M * daniel_hozac i.e. in status and similar?
1188837674 M * bonbons a ps inside guest shows the right pid
1188837682 M * bonbons it's 2.6.22 kernel
1188837694 M * bonbons I would also assume missing hunk
1188837766 M * bonbons yes, in /proc/x/status the pid is mentionned as 1
1188838010 Q * balbir Ping timeout: 480 seconds
1188838074 M * bonbons looks like a hunk is also needed for PF_FILE sockets... (unless the hunk for proc would to the job there as well)
1188838132 M * daniel_hozac hmm?
1188838138 N * DoberMann[PullA] DoberMann
1188838201 M * bonbons I was playing with upstart (ubuntu's new init), it's client tool that communicates with upstart complains about wrong source for answers (which is real pid instead of fake init pid)
1188838266 M * daniel_hozac SCM_RIGHTS?
1188838271 M * bonbons that is via recvmsg() syscall, with cmsg_type=SCM_CREDENTIALS{pid=28983, uid=0, gid=0}} according to strace
1188838285 M * daniel_hozac right, okay.
1188838759 Q * pmenier Quit: pmenier
1188840611 Q * esa` Read error: Connection reset by peer
1188840632 J * esa ~esa@ip-87-238-2-45.adsl.cheapnet.it
1188840919 Q * esa Read error: Connection reset by peer
1188841628 J * esa ~esa@ip-87-238-2-45.adsl.cheapnet.it
1188841718 Q * esa Read error: Connection reset by peer
1188843296 J * Aiken ~james@ppp121-45-255-55.lns2.bne4.internode.on.net
1188843555 J * duckx ~Duck@tox.dyndns.org
1188843851 J * ensc ~irc-ensc@p54B4FBE4.dip.t-dialin.net
1188844008 M * waldi hmm, 2.2.0.3 on sparc looks broken
1188844019 M * waldi it crashs
1188844025 M * waldi with 2.6.22
1188844528 M * waldi hrm, no, it is generic ...
1188845995 M * bonbons waldi, what's the trigger for the crash?
1188846008 Q * jmcaricand Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/
1188846016 M * waldi bonbons: unknown
1188846042 M * waldi it just dies. no message, no break
1188846095 M * bonbons hm, not good. Here I just detected that init pid issue, had no crashes yet
1188846161 M * bonbons but only really needed features are compiled in
1188846417 M * ntrs__ Hi all. Is there an easy way to use tun/tap in a vserver? It is required for OpenVPN.
1188846530 M * julius_ openvpn is able to create tun devices
1188846619 J * fatgoose ~samuel@206-248-175-211.dsl.teksavvy.com
1188846692 M * epicbjorn main problem with vpns in vservers is the routing table
1188846706 M * epicbjorn the routing table is shared amongst the host and the guests
1188847427 Q * transacid Remote host closed the connection
1188847445 J * transacid ~transacid@transacid.de
1188847672 Q * coderanger_ Quit: coderanger_
1188847943 Q * Pazzo Quit: ...
1188850030 Q * DavidS Quit: Leaving.
1188850066 Q * julius_ Remote host closed the connection
1188852645 J * coderanger_ ~coderange@wireless-100.media.mit.edu
1188853392 Q * puck Read error: Connection reset by peer
1188853569 Q * coderanger_ Quit: coderanger_
1188854847 J * t|om ~tresk1234@dslb-084-062-001-220.pools.arcor-ip.net
1188854913 Q * fatgoose Quit: fatgoose
1188855231 Q * bonbons Ping timeout: 480 seconds
1188855325 M * t|om Hi all I have wired problem with my vserver (debian -sarge) and Tomcat5.5 withant
1188855352 M * t|om I installed sun-java-sdk from bakports
1188855364 M * t|om and tomcat5.5 from etch
1188855504 M * t|om now I watend to install something in tomcat
1188855620 Q * meandtheshell Quit: Leaving.
1188855640 M * t|om I set ANT_OPTS="-Xms384M -Xmx512M" and JAVA_OPTS="-Xms384M -Xmx512M" I get the following error at #ant -version
1188855668 M * t|om Error occurred during initialization of VM
1188855668 M * t|om Could not reserve enough space for object heap
1188855668 M * t|om Could not create the Java virtual machine.
1188855716 Q * Borg- charon.oftc.net galapagos.oftc.net
1188855716 Q * UukGoblin charon.oftc.net galapagos.oftc.net
1188855719 M * t|om and without ANT_OPTS I get teh ant version
1188855761 M * t|om now I install a java app. and my system can't uncompress it
1188855783 N * Bertl_zZ Bertl
1188855786 M * Bertl morning folks!
1188855799 M * t|om hi
1188855804 J * UukGoblin ~jaa@sr-fw1.router.uk.clara.net
1188855804 J * Borg- borg@aprogas.student.utwente.nl
1188855811 M * t|om I ever get this error :( /postgres/build.xml:232: Execute failed: java.io.IOException: java.io.IOException: Cannot allocate memory
1188855815 M * Bertl t|om: maybe your guest hasn't enough memory? maybe java is just broken by design?
1188855854 M * t|om >free   gives me Mem:       1048576     693488     355088
1188855855 M * Bertl did you set limits? did you check with /proc/virtual/<xid>/limits?
1188855866 M * t|om no I havn't
1188855874 M * t|om but I do
1188855903 M * Bertl (please use paste.linux-vserver.org for everything longer than 3 lines)
1188855926 M * t|om ah, stop I' havn't a vsrrver as host I'm only a vserver customer
1188855969 M * Bertl well, let's upload /proc/meminfo then and /proc/self/vinfo
1188855970 M * t|om my tomcat5.5 installation works fine
1188856063 M * t|om http://paste.debian.net/36169
1188856087 M * t|om the /proc/self/vinfo doesn't exist
1188856130 M * Bertl what does 'uname -a' give you?
1188856147 M * t|om Linux vs161178.vserver.de 2.6.9-023stab044.1-smp #1 SMP Sat May 5 12:04:40 MSD 2007 i686 GNU/Linux
1188856192 M * Hollow :)
1188856195 M * Bertl sorry, wrong channel, that is OVZ
1188856220 M * Bertl i.e. not Linux-VServer but a different product (maintained by SWSoft)
1188856236 M * t|om omg, 
1188856245 M * Bertl t|om: check out #openvz on freenode
1188856280 M * t|om ok, many thx, but you think this is the right place to search for the solution or is it another thing?
1188856315 M * Bertl no, probably that's the right place, but I fear you won't get too much help there for two reasons
1188856344 M * Bertl first, the developers are seldom around, second there isn't much they can do without root access to your host
1188856381 M * Bertl s/your host/the host of your guest/
1188856423 M * t|om ok, and my host said he doesn't help me in this problem with java :( 
1188856424 M * Supaplex humm I wonder if uclibc guests will work.
1188856477 M * Bertl Supaplex: why wouldn't they?
1188856479 M * t|om but YOu think the solution is in the vserver envoronemnt not in my installtion, that's importent for my future search
1188856506 M * Bertl hard to say, with Linux-VServer I would have opted for too tight limits on the guest
1188856527 M * Bertl java is memory hungry and reserves memory just for the fun of doing it
1188856546 M * Supaplex Bertl: dunno. I'll have to toss an openwrt image inside one. (for failover dhcp/dns/etc) incase I need to do maintenence on my wireless router.
1188856585 M * matti Hi Bertl :)
1188856591 M * Bertl Supaplex: well, userspace should be just fine inside a guest (regardless of the userspace)
1188856601 M * t|om Bertl: Yes thats my impression, too
1188856605 M * Supaplex I suspected as much
1188856628 M * Bertl t|om: but maybe at some point, you decide to go for Linux-VServer, then we can discuss java settings and guest limits :)
1188856641 M * t|om I think my host wants, that I buy his java package, but I don't want to ;)
1188856680 M * t|om So I should change to a more open provider ,)
1188856713 M * t|om In the future I never buy al swsoft solution!
1188856746 M * t|om Just problems with this plesk things
1188856757 M * t|om but, thanks for your help
1188856767 M * t|om I try the other channel
1188856840 M * Bertl you're welcome! 
1188856918 M * t|om :)
1188856950 M * t|om But my hands are tied to htis now ,)
1188857107 M * Bertl good luck!
1188857212 Q * dna_ Quit: Verlassend
1188857230 M * t|om thx :)
1188857232 M * t|om cu
1188857239 P * t|om 
1188859318 M * Supaplex what's the 411 on iptables support? (for guests)
1188859375 M * Supaplex I wonder how I'm going to setup fail-over dhcp to. humm that'll be nasty for sure.
1188859379 M * Bertl there is a proof of concept for a policy daemon solution (called virtualtables)
1188859394 M * Bertl failover dhcp?
1188859417 M * Supaplex yes, incase my router goes offline. it's a feature of dhcpd3
1188859441 M * Supaplex it's not so much fail-over as it is load-balancing, with a heartbeat of sorts to take over.
1188859456 M * Bertl so what does it need iptables for?
1188859477 M * Supaplex I need it to handle the 10./ network ipmasq and routing
1188859494 M * Supaplex right now my openwrt box does all that. but I want some more redundancy
1188859513 M * Bertl well, then the 'failover' guest startup should do that
1188859522 M * Supaplex oh?
1188859546 M * Supaplex and it can shutdown on recovery?
1188859568 M * Bertl you probably want to run something like heartbeat on the host
1188859577 M * Bertl but yes, a guest can shutdown itself too
1188859592 M * Supaplex and give it self up. cool
1188859597 A * Supaplex jots down some notes
1188859763 Q * yarihm Quit: This computer has gone to sleep
1188859816 M * Supaplex uh woops. mtab -> /proc/mounts
1188859823 M * Supaplex no wonder vim freaked. :P
1188860689 M * Supaplex fakerunlevel: open("/var/run/utmp"): No such file or directory
1188860728 M * Supaplex what's that suppose to mean?  /var/run is there. I even tried touching var/run/utmp, and it tosses chroot-sh: unlink("var/run/utmp"): No such file or directory
1188860842 M * Supaplex I'm still running etch (haven't upgraded yet)
1188860902 M * Bertl I see var/run/utmp vs /var/run/utmp
1188860930 M * Supaplex and fakerunlevel vs chroot-sh.
1188861143 M * Supaplex http://rafb.net/p/At2C8j46.html
1188861261 M * Bertl no idea, try to start an init?
1188861447 M * Supaplex same results with an empty inittab. still investigating.
1188861657 M * Supaplex + exec /usr/sbin/vnamespace --new -- /usr/sbin/vserver ----nonamespace tephra start
1188861661 M * Supaplex fakerunlevel: open("/var/run/utmp"): No such file or directory
1188861761 M * Bertl well, I would strace fakerunlevel
1188861788 M * Supaplex does vnamespace run that?
1188861805 M * Supaplex I know strace, but the invocation here leaves me wondering
1188861819 M * Bertl no idea about fakerunlevel
1188861842 M * Supaplex those two lines were a snippet from bash -x `which vserver` tephra start
1188862012 M * daniel_hozac vserver --debug tephra start
1188862035 M * daniel_hozac Bertl: did you see bonbons initpid problems?
1188862107 M * Bertl no, what's up there?
1188862135 M * daniel_hozac guest with plain initstyle gives ENOENT for ls /proc/1, and works fine for /proc/<real pid>
1188862162 M * Bertl and what does ls /proc show?
1188862171 M * daniel_hozac not 1, just the real pid.
1188862188 M * daniel_hozac this is in both 2.2.0.3 and 2.3.0.17 on 2.6.22.
1188862192 M * Bertl well, looks like init pid virtualization turned off to me?
1188862206 M * daniel_hozac yeah, but it's not :)
1188862212 M * Supaplex http://rafb.net/p/Xgg16B57.html
1188862217 M * daniel_hozac i.e. cat /proc/<pid>/status shows the pid as 1.
1188862461 M * daniel_hozac http://people.linux-vserver.org/~dhozac/p/k/delta-initpid-fix02.diff is a step in the right direction, but it has the "if /proc/1 has been accessed in a guest, the spectator will show that init twice and skip the real one" bug.
1188862539 M * daniel_hozac eh, -proc_flush_task hunks.
1188862593 M * Bertl currently I cannot test (my test system is not in an usable state), but what does the vcmd example show if you use /bin/bash instead of cat?
1188862613 M * Bertl note: we _know_ that 2.6.22 cannot properly virtualize init
1188862635 M * Bertl (because of the way mainline changed the procfs)
1188862650 M * daniel_hozac so, this is expected?
1188862665 M * Bertl but the virtualization should be sufficient to work for init and pstree and such
1188862693 M * daniel_hozac sure, ps still believes init is pid 1.
1188862726 M * Bertl the problem there is, that IMHO there is no simple way to make it work properly, because you would have to revalidate pid=1 the first time, but not the second
1188862729 M * daniel_hozac but anything using /proc/1 will fail.
1188862764 M * Bertl if you find a solution, I don't have a problem integrating your patches :)
1188862778 M * Bertl I played around with it for a few days without good solution
1188862898 M * daniel_hozac okay...
1188862954 M * daniel_hozac why can't we do what we've done in the past?
1188863013 M * Bertl because we have no way to tell which one is the right and which one the wrong one
1188863044 M * Bertl if we could cause revalidate to trigger on pid=1 && rpid!=init_pid
1188863057 M * Bertl then it would work just fine
1188863066 M * daniel_hozac so revalidate isn't even called?
1188863071 M * Bertl yep
1188863080 M * daniel_hozac ah...
1188863090 M * Bertl /proc/1 has an inode, everything is happy. period.
1188863143 M * Bertl I am waiting for the pid spaces to work around that
1188863166 M * Bertl (well, they have to) and it should be simple there to implement the blend through init
1188863174 M * daniel_hozac yeah.
1188863182 M * daniel_hozac that seems to be taking forever though :)
1188863351 M * Bertl yeah, the 'interest' is now lower than ever (kernel summit) and probably will fade for some time afterwards :)
1188863480 Q * Piet Ping timeout: 480 seconds