1306627331 Q * Piet Ping timeout: 480 seconds
1306632841 N * Piet_ Piet
1306634162 Q * ser Remote host closed the connection
1306634783 J * aj__ ~aj@p4FFD212B.dip.t-dialin.net
1306635160 Q * derjohn_foo Ping timeout: 480 seconds
1306635457 Q * arekm Ping timeout: 480 seconds
1306635524 J * ser ~ser@host1.tldp.ibiblio.org
1306636397 J * arekm arekm@carme.pld-linux.org
1306639232 M * Bertl_oO off to bed now ... have a good one everyone!
1306639237 N * Bertl_oO Bertl_zZ
1306643072 J * bsingh ~balbir@122.172.3.21
1306649715 Q * Romster Quit: Geeks shall inherit properties and methods of object earth.
1306654842 J * ncopa ~ncopa@3.203.202.84.customer.cdi.no
1306658235 J * bonbons ~bonbons@2001:960:7ab:0:d1f5:d8da:3e24:fcfd
1306658755 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com
1306659702 J * petzsch ~markus@dslb-092-078-115-160.pools.arcor-ip.net
1306664150 J * harobed ~harobed@arl57-1-82-231-110-14.fbx.proxad.net
1306667967 Q * petzsch Quit: Leaving.
1306668753 J * petzsch ~markus@dslb-092-078-115-160.pools.arcor-ip.net
1306668769 J * sannes ~ace@cm-84.209.81.224.getinternet.no
1306668979 Q * bsingh Ping timeout: 480 seconds
1306669500 J * bsingh ~balbir@122.172.42.170
1306669703 J * Net147 Net147@60-242-179-244.static.tpgi.com.au
1306669850 Q * Net147 
1306670382 Q * Marillion Ping timeout: 480 seconds
1306671585 Q * ncopa Quit: Leaving
1306671885 J * Marillion ~dirk@178.63.150.30
1306674753 Q * petzsch Quit: Leaving.
1306675677 Q * manana Ping timeout: 480 seconds
1306676179 J * VirMan c113a53b@ircip2.mibbit.com
1306676186 M * VirMan Hi :)
1306676190 J * manana ~mayday090@nat049-252-205-109.tvoe.tv
1306676225 Q * harobed Ping timeout: 480 seconds
1306676270 M * VirMan i got maybe silly question.. but whu on older servers when i entered guest and typing free -m i was seeing exactly how much memory i have, and in current (newer) release there is limit , but i see whole 32 gb ram avilable ?
1306676320 N * Bertl_zZ Bertl
1306676324 M * Bertl morning folks!
1306676330 M * VirMan i set rss for 128000 (512M) and there is limit , but running top or sth shows whole 32g
1306676334 M * VirMan hi :)
1306676351 M * VirMan Bertl: it is afternoon here ;)
1306676388 M * Bertl and somewhere on the world it is night, imagine that!
1306676396 M * VirMan hehe i know :)
1306676405 M * VirMan nevermind ;)
1306676415 M * Bertl how did you set rss and what kernel/patch are we talking about?
1306676422 M * VirMan Where i shuld search for in above matter ? :)
1306676472 M * Bertl depends on your kernel/patch version :)
1306676475 M * VirMan 2.6.35-vs2.3.0.36.32 , and 0.30.216-pre2955
1306676488 M * Bertl okay, and how do you set the rss limit?
1306676498 M * VirMan in file rlimits/rss
1306676505 M * VirMan as always
1306676506 M * Bertl well, there is your problem
1306676517 M * Bertl newer kernels use cgroups to limit memory
1306676537 M * VirMan hm even if i set in compilation to use legacy memory limits?
1306676552 M * Bertl did you read the help text there?
1306676569 M * VirMan about old file?
1306676577 M * Bertl it is a dummy without any function, just to make older util-vserver happy
1306676586 M * VirMan aa...
1306676592 M * VirMan ok.
1306676600 M * VirMan i'll search for cgroupps
1306676609 M * VirMan Bertl: thanks
1306676617 M * Bertl there should be at least one wiki page with all the details
1306676624 M * Bertl bp
1306676626 M * Bertl np
1306676657 M * VirMan http://www.nongnu.org/util-vserver/doc/conf/configuration.html - i found that page
1306676687 M * VirMan but there is not much about contents of these files
1306676718 M * Bertl http://linux-vserver.org/util-vserver:Cgroups
1306676799 M * VirMan ok. sounds bit more complicated ;)
1306676806 M * VirMan but i guess it will help :)
1306676822 M * Bertl once you know how it works, it's rather simple
1306676846 M * Bertl although I'd suggest a 2.6.38.x kernel
1306676901 M * VirMan mhm - i don't have ebiuld for that - what util-vserver is ok for what you suggested?
1306676934 M * Bertl the latest snapshot is always the best choice, but the version you have should be fine for this purpose
1306676969 M * VirMan yup - i just want network , memory and process isolation - i do not need to even limit cpu.
1306679951 M * VirMan ok.
1306679962 M * VirMan still got problem : /dev/cgroup/vs6060/memory.memsw.limit_in_bytes: Permission denied
1306680000 M * Bertl did you enable the swap limits ?
1306680019 M * VirMan you got me - what is this?
1306680039 M * VirMan this occured during guest restart
1306680045 M * Bertl a kernel option, you seem to be using
1306680065 M * VirMan ok.. i used menuconfig. i guess it's not listed there
1306680115 M * Bertl I'm pretty sure it is, assuming your kernel supports it
1306680131 M * VirMan i checked all options
1306680142 M * VirMan excluding debug
1306680146 M * Bertl CONFIG_CGROUP_MEM_RES_CTLR_SWAP=y in 2.6.38
1306680204 M * VirMan ok it's not listed there.
1306680259 M * Bertl well, the mainline 2.6.35.12 kernel has it
1306680268 M * Bertl not sure what your kernel source actually is
1306680341 M * VirMan i got only CONFIG_CGROUP_MEM_RES and it was unset
1306680347 M * VirMan i set it now to y
1306680358 M * Bertl and you will get the option for the other
1306680431 M * VirMan so to have this option in kernel i need to use newer sources.. 
1306680457 M * Bertl no, you just need to enable CONFIG_CGROUP_MEM_RES
1306680494 M * Bertl CONFIG_CGROUP_MEM_RES_CTLR=y to be precise
1306680512 M * Bertl then you will get CONFIG_CGROUP_MEM_RES_CTLR_SWAP as option
1306680555 M * VirMan i put only memory.limit_in_bytes file and it worked.
1306680586 M * Bertl of course, if you do not want to use the swap limits, it's fine as well
1306680593 M * VirMan i hope this includes also memory alocated from swap ?
1306680618 M * VirMan i mean if ost will run out of main memory it an allocate swap
1306680627 M * VirMan and put guest tasks there?
1306680655 M * Bertl read the cgroups documentation specifically those for the memory controller to get all the gory details
1306680667 M * Bertl it's a mainline feature and Linux-VServer uses it as is
1306680677 M * VirMan ok :)
1306680700 M * VirMan documentation is not rich but i'll try to find :)
1306680764 M * VirMan ok - still i can see whole memory in guest
1306680777 M * VirMan even with cgroup 
1306680801 M * Bertl did you enable the memory controller?
1306680820 M * Bertl i.e. did you rebuild and install the kernel?
1306680842 M * sannes daniel_hozac: How do you want the vumask util to work? I suppose I should mirror what the other v* utils do .. but, if you have any preferences ..
1306680870 M * VirMan ok i checked only that CONFIG_CGROUP_MEM_RES option to y
1306680888 M * VirMan manually in config file - not using menuconfig
1306680897 M * VirMan i'll rebuild it and reinstall
1306680958 M * VirMan ok - it asked me for additional option :)
1306680993 M * VirMan i thought that those kernels have default build in cgroups (because older stopped working)
1306681031 M * Bertl you have to talk to the person maintaining that old kernel
1306681072 M * VirMan i would like to have newer ebuilds in portage tree - but well there is no perfect world :)
1306681094 M * Bertl have you told the maintainer yet?
1306681121 M * VirMan hollow is not listed here - i told him few times
1306681135 M * VirMan who is in charge of kernels and util-vservers for gentoo ?
1306681138 M * ser hmm, an .ova file (Open Virtualization Format) - how to eat it with linux-vserver?
1306681257 M * VirMan ser : have no idea.
1306681276 M * Bertl ser: no idea what that would be :)
1306681276 M * ser :)
1306681323 M * VirMan Bertl: Do you know who i should talk to about gentoo portage tree?
1306681354 M * sannes  bugs.gentoo.org ?
1306681362 M * Bertl I presume there is some kind of maintainer listed somewhre, no?
1306681453 M * VirMan i'll try to find him - i hoped there is someone here to take care about gentoo :)
1306681855 M * VirMan should i suggest vs2.3.0.37-rc15 kernel ? and what util vserver version to add ?
1306681936 M * Bertl yep, 2.6.38.x with latest patch would be fine
1306681969 M * VirMan Bertl: and util-vserver in version ...?
1306682439 M * VirMan ok submitted with latest versions
1306682574 M * VirMan well - still after rebuild and restart it is not working - i see all memory
1306682799 Q * bsingh Read error: Connection reset by peer
1306683099 M * daniel_hozac sannes: doesn't need it's own binary, should just be options to vattribute.
1306683700 J * er ~sapan@c-68-36-169-19.hsd1.nj.comcast.net
1306683882 M * Bertl VirMan: check that cgroups are configured and show the appropriate limits
1306683917 M * Bertl also, I'd suggest to check with a recent kernel as well, as I'm not sure the implementation in 2.6.35 was complete
1306684284 M * sannes daniel_hozac: will do
1306684322 M * er hi Bertl, i think i might have found our COW problem from the other day:
1306684330 M * er #define S_IXUNLINK      1024    /* Immutable Invert on unlink */
1306684345 M * er lib/vserver.h:#define   VC_IATTR_IUNLINK                0x00020000u
1306684375 M * er hi daniel_hozac if you are around
1306684375 M * Bertl okay?
1306684396 M * Bertl an what is the problem?
1306684443 M * er i am assuming there is a mismatch between the IUNLINK flags used in the kernel and in userspace
1306684464 M * daniel_hozac S_IXUNLINK and VC_IATTR_IUNLINK are for two separate interfaces.
1306684604 M * daniel_hozac what exactly is it that doesn't work for you?
1306684623 M * er daniel_hozac: COW stopped working for us since we upgraded to 2.6.32
1306684657 M * daniel_hozac COW when you do...?
1306684673 M * Bertl er: btw, what filesystem?
1306684674 M * daniel_hozac because COW has about 6 different ways it can happen in a recent kernel, IIRC.
1306684677 M * er write to a hard linked file in a guest
1306684717 M * er what I noticed was that vserver's definition of IS_COW in recent kernels had changed from
1306684736 M * er IS_IUNLINK(inode) && IS_IMMUTABLE(inode)
1306684748 M * er to IS_IXUNLINK(inodes) && IS_IMMUTABLE(inode)
1306684774 M * er where the value of I[X]UNLINK has changed from 2048 to 1024
1306684799 M * er so I was looking for that code in util-vserver - trying to find the version of the flag that it uses
1306684821 M * er Bertl: ext3
1306684842 M * daniel_hozac it doesn't.
1306684845 M * daniel_hozac the kernel handles that.
1306684860 M * daniel_hozac see __vc_set_iattr.
1306684910 J * petzsch ~markus@dslb-092-078-115-160.pools.arcor-ip.net
1306685075 M * er i see
1306685083 M * er well, that makes sense :)
1306687660 M * Bertl er: did you run testfs.sh yet?
1306687739 M * er no, doing now
1306687907 M * er xattr-related test passed fine on ext2/ext3
1306687928 M * daniel_hozac -x -y -z?
1306687980 M * er i used the commandline from the wiki: ./testfs.sh -t -x -y -z -D /dev/loop0 -M /mnt/tmp
1306688008 M * Bertl then I'd say your CoW link breaking is just fine :)
1306688149 J * bsingh ~balbir@122.166.160.99
1306688156 M * er or there's something squirrelly enough to pass the tests but fail real life
1306688189 M * Bertl my money would be on having the wrong attributes after all
1306688222 M * Bertl but I'd suggest to do the testfs tests by hand and see what happens
1306688237 M * Bertl if that works just fine, check what differs to your tests
1306688286 M * er testfs tries to actually break links?
1306688296 M * er (i.e., is that one of the tests?)
1306688321 J * harobed ~harobed@arl57-1-82-231-110-14.fbx.proxad.net
1306688424 M * Bertl do_xattr_cowbreak()?
1306688507 M * Bertl and do_xattr_iunlink() does similar tests
1306688685 Q * aj__ Ping timeout: 480 seconds
1306688728 M * er ok, thanks. i'll check the diff b/w my failed test and those
1306688815 M * Bertl what we don't actually test yet is breaking by append, but a patch to add this test to testfs.sh would be welcome
1306688875 M * Bertl note that you need to be careful that the append actually happens inside a context, i.e. typical bash >> doesn't work here
1306689300 M * er that's the test that has been failing for me.
1306689308 M * er (echo "" >> file in guest)
1306689329 M * er but point taken about >> not passing into guest context
1306689920 M * er it's not limited to append. regular write failed too. 
1306689945 M * er -bash-3.2# ls -i /etc/sudoers 
1306689945 M * er 21546906 /etc/sudoers
1306689945 M * er -bash-3.2# perl -e 'open FIL,">/etc/sudoers";close FIL;'
1306689946 M * er -bash-3.2# ls -i /etc/sudoers 
1306689946 M * er 21546906 /etc/sudoers
1306689948 M * Bertl so please extend the testfs.sh and upload a patch
1306689964 M * Bertl I'll cross check here in my setup then
1306690035 M * er i'll look into why test.sh works but manual testing fails
1306690057 M * er and if the difference can be covered by a test, then i'll go ahead and extend it
1306690151 M * er (i.e., our problem does not seem limited to append)
1306692065 Q * Piet Ping timeout: 480 seconds
1306692561 J * Piet ~Piet__@1RDAAAEN0.tor-irc.dnsbl.oftc.net
1306692601 M * daniel_hozac did the file have more than one link?
1306692613 M * daniel_hozac because a COW-marked file with just one link will just have the attributes removed.
1306692729 J * aj__ aj@80.187.245.244
1306694682 M * Bertl possible, but I don't think that changed 'recently'
1306694967 Q * VirMan Quit: http://www.mibbit.com ajax IRC Client
1306695370 Q * Piet Quit: Piet
1306695536 J * Piet ~Piet__@1RDAAAEPG.tor-irc.dnsbl.oftc.net
1306696178 M * daniel_hozac well, i can reproduce the issue here.
1306696202 M * er daniel_hozac: the attributes seem fine. 
1306696205 M * er [root@kupl1 ~]# showattr /vservers/purdue_2/etc/sudoers 
1306696205 M * er -----UICX /vservers/purdue_2/etc/sudoers
1306696205 M * er [root@kupl1 ~]# 
1306696221 M * Bertl daniel_hozac: with testfs.sh?
1306696241 M * daniel_hozac testfs "works", maybe we don't test this.
1306696285 M * Bertl that's what I was suggesting ... so it would be a good idea to add a proper check
1306696431 M * daniel_hozac http://paste.linux-vserver.org/19984
1306696475 M * Bertl okay, and we cannot recreate it with bash?
1306696493 M * Bertl (because I just saw perl and python :)
1306696521 M * daniel_hozac sure, plain bash works too.
1306696626 M * daniel_hozac it seems to be caused by FMODE_WRITE in may_open.
1306696645 M * daniel_hozac opening with O_RDWR does break the link
1306696658 M * daniel_hozac (O_RDWR == FMODE_WRITE)
1306696815 J * SwenTjuln_ ~SwenTjuln@77-111-2-36.ipv4.tusmobil.si
1306696885 Q * SwenTjuln Ping timeout: 480 seconds
1306696885 N * SwenTjuln_ SwenTjuln
1306696904 M * Bertl interesting ...
1306696923 M * Bertl does the intend go wrong or do we just not handle that case (yet)?
1306697071 M * daniel_hozac seems like it might have just been a change in mainline
1306697076 M * daniel_hozac that we failed to catch
1306697119 M * Bertl there are more changes ahead in 2.6.39
1306697136 M * Bertl CoW link breaking is one reason why we don't have a patch yet
1306697140 Q * aj__ Ping timeout: 480 seconds
1306697144 M * daniel_hozac okay
1306697161 M * Bertl nevertheless we should fix that in 2.6.38
1306697171 M * Bertl as it will be _the_ long time kernel IMHO
1306697202 M * daniel_hozac i think changing the may_open test to (flag & O_ACCMODE) != O_RDONLY should work.
1306697216 M * Bertl makes sense
1306697223 M * Bertl er: could you test that?
1306697552 M * er confirmed
1306697605 M * er opening a file r/w and closing it breaks the link
1306697655 M * Bertl I meant, could you test the fix? :)
1306697695 M * er oh, i didn't see that blurb. doing now.
1306697939 M * er or actually, will do in a bit, need to step out for a while.
1306698487 Q * petzsch Quit: Leaving.
1306698882 J * derjohn_mob aj@88.128.47.238
1306699511 Q * sannes Remote host closed the connection
1306700967 Q * bonbons Quit: Leaving
1306701381 M * er Bertl, daniel_hozac: the fix worked
1306701413 M * er thanks very much for the help.
1306701499 M * Bertl thanks for reporting the problem!
1306701771 M * er always happy to crash land in your vicinity :-) 
1306702139 Q * derjohn_mob Ping timeout: 480 seconds
1306703606 Q * harobed Ping timeout: 480 seconds
1306704218 Q * manana Remote host closed the connection
1306704781 M * Bertl daniel_hozac: http://vserver.13thfloor.at/ExperimentalT/delta-cow-fix23.diff
1306704796 M * daniel_hozac yeah, looks fine.
1306704877 M * Bertl okay, do we know which kernels are affected besides 2.6.38?
1306704899 M * daniel_hozac at least all of them from 2.6.32.
1306704928 M * Bertl so 2.6.32-2.6.37 for sure
1306704942 M * daniel_hozac yeah.
1306704973 M * Bertl okay, I guess we can let the 2.6.34 branch die
1306705015 M * Bertl or do you know anybody using 2.6.34?
1306705064 M * daniel_hozac no, i think 2.6.34 is abandoned.
1306705157 M * Bertl IMHO 2.6.36 is dead as well, but I think some distro is using that, no?
1306705175 M * fback debian maybe? ;>
1306705194 M * Bertl nah, they are using 2.6.32 or 2.6.26 :)
1306709391 J * derjohn_mob ~aj@d047107.adsl.hansenet.de
1306711691 N * ensc Guest2692
1306711701 J * ensc ~irc-ensc@p5DF2F248.dip.t-dialin.net
1306712110 Q * Guest2692 Ping timeout: 480 seconds