1285546402 J * ntrs_ ~ntrs@77.29.87.199 1285546849 Q * ntrs Ping timeout: 480 seconds 1285560874 Q * ntrs_ Ping timeout: 480 seconds 1285561973 J * derjohn_mob aj@80.187.225.128 1285564518 J * ghislain ~AQUEOS@adsl2.aqueos.com 1285565383 J * ghislain1 ~AQUEOS@adsl2.aqueos.com 1285565601 Q * ghislain Ping timeout: 480 seconds 1285566050 Q * derjohn_mob Ping timeout: 480 seconds 1285566065 J * balbir_ ~balbir@122.248.161.59 1285566120 J * derjohn_mob ~aj@80.187.225.128 1285566628 J * `kbad ~kyle@apache2-cid.adonis.dreamhost.com 1285566771 M * `kbad what's the difference between 2.3.0.36.31 & 2.3.0.36.32? 1285567112 M * Bertl mainly mainline adaptations 1285567330 M * Bertl off to bed now ... have a good one everyone! 1285567336 N * Bertl Bertl_zZ 1285567475 J * ncopa ~ncopa@180.40.189.109.customer.cdi.no 1285567730 Q * derjohn_mob Ping timeout: 480 seconds 1285568138 J * ncopa_ ~ncopa@180.40.189.109.customer.cdi.no 1285568327 Q * ncopa Quit: Ex-Chat 1285568345 J * derjohn_mob aj@88.128.212.112 1285568360 N * ncopa_ ncopa 1285569482 Q * balbir_ Ping timeout: 480 seconds 1285570354 Q * fback Ping timeout: 480 seconds 1285570639 J * fback fback@red.fback.net 1285570704 J * balbir_ ~balbir@122.248.163.1 1285571428 Q * derjohn_mob Ping timeout: 480 seconds 1285571887 J * derjohn_mob aj@88.128.185.86 1285573022 J * thierryp ~thierry@zankai.inria.fr 1285573204 Q * balbir_ Ping timeout: 480 seconds 1285574120 Q * derjohn_mob Ping timeout: 480 seconds 1285574474 J * balbir_ ~balbir@122.248.161.59 1285574754 J * petzsch ~markus@p4FF451BF.dip.t-dialin.net 1285576296 J * manana ~mayday090@84.17.25.149 1285576483 J * imcsk8 ~ichavero@201.144.130.18 1285577128 Q * petzsch Read error: Connection reset by peer 1285577873 J * derjohn_mob ~aj@213.238.45.2 1285579206 J * BenG ~bengreen@cpc6-aztw22-2-0-cust100.aztw.cable.virginmedia.com 1285579276 Q * imcsk8 Quit: This computer has gone to sleep 1285579373 Q * BenG 1285579761 Q * _nono_ Read error: Connection reset by peer 1285579889 N * Bertl_zZ Bertl 1285579894 M * Bertl morning folks! 1285580321 M * karasz hello Bertl 1285581277 Q * balbir_ Ping timeout: 480 seconds 1285581521 Q * manana Ping timeout: 480 seconds 1285581715 J * petzsch ~markus@p4FF451BF.dip.t-dialin.net 1285581864 J * _nono_ ~gomes@libation.ircam.fr 1285581915 M * urbee hi 1285581917 M * urbee :) 1285581932 M * urbee i'd like to add a second partition to the guest, which is not located on the same hard drive 1285581935 M * urbee is that possible? 1285581957 M * urbee eg. guest on HDD1, hdd1 is out of space so id like to add a partition which is located on hdd2 1285582142 J * BenG ~bengreen@cpc6-aztw22-2-0-cust100.aztw.cable.virginmedia.com 1285582447 M * Bertl urbee: sure, just add it to the guest configs fsta 1285582450 M * Bertl *fstab 1285582540 M * urbee what opts? 1285582560 M * urbee /partition2 /partition2onguest ext3 noatime 0 1 1285582563 M * urbee would this be correct? 1285582596 M * urbee (i dont wanna add the whole partition, just one directory in it, since the partition is used for other things too) 1285582621 M * Bertl then you want a bind mount, not a filesystem mount 1285582633 M * Bertl filesystem mounts always use an entire device 1285582974 J * barismetin ~barismeti@zanzibar.inria.fr 1285583184 M * urbee okay.. 1285583194 M * fback daniel_hozac: I'm getting empty .deb files with your scripts. Where to start investigation? 1285583228 M * fback daniel_hozac: there are binaries in src directory, so it's rather problem with package generation 1285584107 J * balbir_ ~balbir@122.172.27.208 1285584906 M * urbee vnamespace --enter -- mount --bind /host/path /vservers/guest/path 1285584933 M * urbee how to mount this permanently? 1285584940 M * urbee i mean what exactly to add in fstab 1285584941 M * urbee :p 1285584982 Q * petzsch Quit: Leaving. 1285585085 M * urbee /host/path /guest/path bind noatime 0 1 1285585086 M * urbee ? 1285585088 M * daniel_hozac fback: is there a log file? 1285585152 M * Bertl urbee: bind is the option, but otherwise yes, i.e. use none/auto as filesystem (see man fstab) 1285585459 Q * fback Remote host closed the connection 1285585465 J * fback fback@red.fback.net 1285585586 M * Bertl daniel_hozac: feature request for vspace/vnamespace: can we add a short usage line for -i ? 1285585626 M * Bertl I always keep forgetting the option name, and it isn't documented anywhere except for the source :) 1285585713 M * fback daniel_hozac: http://people.fback.net/fb/build.log 1285585772 M * fback daniel_hozac: and example contents http://paste.linux-vserver.org/17998 1285586367 J * ntrs ~ntrs@77.28.171.158 1285586505 J * petzsch ~markus@p4FF451BF.dip.t-dialin.net 1285587115 Q * petzsch Quit: Leaving. 1285587485 J * petzsch ~markus@p4FF451BF.dip.t-dialin.net 1285587726 M * urbee Bertl: so /host/path /guest/path none bind 0 1 1285588411 J * manana ~mayday090@84.17.25.149 1285588486 M * Bertl looks good, yes 1285589326 M * urbee /home2: skipping bad line in /etc/fstab: bind mount with nonzero fsck pass number 1285589360 M * fback urbee: 0 0 at the end, not 0 1 1285589437 Q * BenG Quit: I Leave 1285589647 J * ntrs_ ~ntrs@77.28.162.133 1285589730 M * urbee ok 1285589733 M * urbee mount shows it mounted 1285589738 M * urbee but df -h doesnt show it 1285589846 M * urbee ok 1285589854 M * urbee did 'auto' instead of 'none' 1285589855 M * urbee works now :p 1285590079 Q * ntrs Ping timeout: 480 seconds 1285590591 Q * manana Ping timeout: 480 seconds 1285590862 N * zbyniu_ zbyniu 1285592020 M * Bertl nap attack ... bbl 1285592025 N * Bertl Bertl_zZ 1285592941 Q * petzsch Quit: Leaving. 1285594507 J * petzsch ~markus@p4FF451BF.dip.t-dialin.net 1285595492 J * tusk_ ~AndChat@213.55.131.85 1285595495 M * tusk_ Hello 1285595518 M * hijacker hi 1285595585 Q * petzsch Quit: Leaving. 1285595593 M * tusk_ I'm testing latest 2.6.35.5 patch do i need to have newer vserver-utils than 0.30.216-pre2882 ?? 1285595632 M * daniel_hozac yes 1285595648 M * tusk_ Where can i ifnd them?? 1285595740 M * tusk_ Oh ok forgive my eyes i didnt see the "pre" link :) 1285596197 M * tusk_ Do i need to restart my vservers after utils upgrade? 1285596413 M * `kbad anyone else notice higher loads on 2.6.35.4? 1285596958 J * user__ ~user@c-015.redwoods.edu 1285597262 Q * user_ Ping timeout: 480 seconds 1285597600 P * tusk_ Leaving 1285598058 J * dowdle ~dowdle@scott.coe.montana.edu 1285600008 Q * ncopa Quit: Ex-Chat 1285600383 Q * thierryp Remote host closed the connection 1285601998 J * petzsch ~markus@p4FF451BF.dip.t-dialin.net 1285602928 Q * barismetin Remote host closed the connection 1285603081 M * mnemoc Bertl_zZ: can you think on any reason why a `mount -nt proc none proc` in a vserver could block a machine? (no panic, just whole-world-freezing) 1285603691 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1285604686 Q * derjohn_mob Ping timeout: 480 seconds 1285604979 M * mnemoc Bertl_zZ: 2.6.35.5-vs2.3.0.36.32-dist 0.30.216-pre2883 no limits set 1285605132 M * daniel_hozac mnemoc: is it permanently blocked, or does it unblock eventually? 1285605158 M * mnemoc in unblocks 1285605159 M * mnemoc it* 1285605178 M * mnemoc the blocking goes from 2-3s to hours 1285605213 M * mnemoc sometimes i get bored and powercycle it 1285605251 M * mnemoc this time i powercycled it after 30m blocked 1285605336 M * mnemoc my usage is very abusive, i mount/umount after every package is built.... i'm editing the build kit to just leave the stuff mounted, but it shouldn't block anyway :) 1285605377 M * mnemoc it this case it was a very trivial test of a helper with the machine idling 1285605489 M * daniel_hozac any SELinux etc. enabled? 1285605493 M * mnemoc nope 1285605505 M * mnemoc let me confirm 1285605517 M * mnemoc # grep SELINUX /boot/kconfig_2.6.35.5-vs2.3.0.36.32-dist 1285605517 M * mnemoc # CONFIG_SECURITY_SELINUX is not set 1285605522 M * mnemoc confirmed 1285605546 M * daniel_hozac anything else CONFIG_SECURITY_ set? 1285605569 M * mnemoc yuck, yes 1285605598 M * daniel_hozac which one? 1285605617 M * mnemoc http://paste.linux-vserver.org/18003 1285605680 M * mnemoc it's generated, so any suggested change is highly welcomed 1285605777 Q * nkukard Ping timeout: 480 seconds 1285605787 M * mnemoc that machine is only used for building (and it's failing to do so) so i can do any test you want or give you ssh access 1285605965 M * mnemoc i'll disable smack and tomoyo 1285606110 J * nkukard ~nkukard@41-133-236-167.dsl.mweb.co.za 1285606157 M * mnemoc fresh after the restarting the same mounts takes 26ms, in following calls it goes degrading 1285606200 J * derjohn_mob aj@88.128.141.214 1285606242 J * morfoh ~morfoh@shell.opensde.net 1285606248 M * daniel_hozac really? 1285606251 M * daniel_hozac it degrades? 1285606259 M * daniel_hozac how do you unmount them? 1285606269 M * morfoh so mnemoc ... here is your "lazy cow" 1285606283 M * morfoh hi all 1285606317 M * mnemoc daniel_hozac: simple mount $sandbox/proc 1285606348 M * mnemoc hi morfoh 1285606374 M * morfoh mnemoc: ja ja ... nice to see you ;) 1285606390 M * daniel_hozac but it is getting unmounted? 1285606400 M * mnemoc not always :< 1285606419 M * mnemoc sometimes it gets "busy" even if no process is using it 1285606462 M * mnemoc makes any sense to you? 1285606480 J * imcsk8 ~ichavero@201.174.32.227 1285606634 M * daniel_hozac no... 1285606688 M * mnemoc i feared so 1285606704 N * Bertl_zZ Bertl 1285606706 J * lauanana ~lauanana@212-198-248-35.rev.numericable.fr 1285606709 M * Bertl back now ... 1285606726 M * Bertl mnemoc: could you test your 'workload' on a mainline kernel? 1285606728 M * morfoh mnemoc: did you told about the vxW kernel log message ? 1285606759 M * morfoh hi Bertl 1285606765 M * mnemoc morfoh: nope, haven't 1285606775 M * mnemoc Bertl: on chroot it doesn't degrade 1285606785 M * morfoh because the problems appear when we actually enter the sandbox with chroot 1285606809 Q * lauanana 1285606820 M * morfoh vxW: [mount,20790:#10003|10003|10003] denied 12 access to proc:ffff8801e8c00030[#0,1] 1285606839 M * morfoh this appears in the kernel logs 1285606852 M * mnemoc morfoh: i haven't experienced any issue when building from chroot ever 1285606882 M * Bertl morfoh: when does this message get logged (i.e. when you do what exactly)? 1285606909 M * morfoh mnemoc: initially the problem appeared exactly when we go into "chroot mode" 1285606915 M * mnemoc i haven't correlated it yet... it's one mount -tn proc, and 3 mount --bind 1285606916 M * morfoh Bertl: mounting proc 1285606923 M * mnemoc ok 1285606945 M * morfoh mnemoc: iirc I uncommented the proc mount 1285606956 M * morfoh and that message disappeared 1285606962 M * mnemoc ok 1285606978 J * fzylogic ~fzylogic@dsl081-243-128.sfo1.dsl.speakeasy.net 1285606985 M * Bertl but it still gets slower? 1285607007 M * mnemoc morfoh: you have been hunting this much longer than i, you should have come to ask months ago 1285607032 M * morfoh mnemoc: you know ... I am kind of shy ;) 1285607043 M * morfoh mnemoc: but yeah ... you are right ;) 1285607047 M * mnemoc morfoh: answe Bertl's question, please 1285607107 M * morfoh Bertl: it seems to get slower with the amount of mount/umount cycles 1285607165 M * Bertl just to clarify, you do a chroot _inside_ a guest, then (bind) mount proc there, eventually unmount it and that quite a number of times, yes? 1285607194 M * morfoh yes 1285607196 M * mnemoc Bertl: we mount inside a guest before chroot-ing 1285607204 M * morfoh ack 1285607214 M * morfoh chroot'ing comes after the bind mounts 1285607216 M * mnemoc exit chroot, and umount 1285607221 M * mnemoc and loop 1285607265 M * morfoh mnemoc: I have to move home now 1285607267 M * morfoh so bbl 1285607305 M * Bertl okay, could you make a snapshot of /proc/mounts (inside the guest's namespaces) and /proc/virtual/status every N mounts or so? 1285607370 M * Bertl and is a reboot required or does killing the guest solve the issue as well? 1285607406 M * morfoh Bertl: killing the guest seems to "solve" that issue 1285607419 M * morfoh but mnemoc likes to reboot the whole machine ;) 1285607421 M * mnemoc does /proc/virtual/status snapshoting need to be every N mounts? ... i would need to ssh from the guest to the host :p 1285607474 M * Bertl mnemoc: nah, it would be fine if you do it like every hour or so 1285607481 M * mnemoc ok, cron then 1285607491 M * Bertl I just want to see how it changes over times 1285607501 M * Bertl (if it changes at all) 1285607590 M * Bertl to me it sounds like the mounts aggregate over time, most likely because they happen at the same mount point, so a pile of stackered mounts keep hanging around, which need to be traversed every time you do any operation on them 1285607630 M * Bertl I presume this actually is a mainline issue, related to namespaces and mount propagation 1285607676 M * mnemoc ic 1285607700 M * Bertl (so it probably will show up on an unpatched kernel when using a similar namespace setup as Linux-VServer does, which we need to check in some way) 1285608310 M * mnemoc one funny oddness one can see in /proc/mounts (http://paste.linux-vserver.org/18004) is that it duplicates two of them weirdly following a symlink 1285608362 M * mnemoc TOOLCHAIN/loop/build -> ../.. 1285608399 M * mnemoc but anyway, i'll collect the data first 1285608504 M * mnemoc uh, no. the symlink aint there. but it's puzzled 1285609815 M * mnemoc Bertl: you were right 1285609838 M * mnemoc made a simple test which leave stackered mounts hanging around.... 1285609876 M * mnemoc http://paste.linux-vserver.org/18005 1285609944 M * mnemoc but bind mount is not supposed to be recursive... 1285609975 M * mnemoc why does it put my test_b/my_a inside test_b/loop?? 1285610099 M * mnemoc i'll fix my script to not trigger this issue 1285610428 M * mnemoc Bertl: http://paste.linux-vserver.org/18006 .... after a few hundred iterations *boom* 1285610691 M * mnemoc Bertl: the same test in the host doesn't "duplicate" the mount, no busy resource and an lingering mount 1285610766 M * mnemoc is there still value in /proc/virtual/status snapshots? 1285610813 M * Bertl yes 1285611249 J * ntrs__ ~ntrs@77.28.165.151 1285611320 M * mnemoc using different locations (test$i) first iteration took 2ms, after 600 iterations it gos in 300ms, 700 interations 400ms, ... 1285611336 M * mnemoc no change at /proc/virtual/status 1285611355 M * mnemoc getting slooooooow.... 1285611361 M * mnemoc i=800, 0.5s 1285611377 M * mnemoc /proc/virtual/status remains intact 1285611520 M * mnemoc 1000 loops over, first took 2ms, last 739ms 1285611544 M * mnemoc http://paste.linux-vserver.org/18007 <--- /proc/virtual/status 1285611557 M * mnemoc the 92 was 90 some times 1285611561 M * mnemoc but mostly the same 1285611571 Q * derjohn_mob Ping timeout: 480 seconds 1285611573 M * mnemoc i mean, all the rest didn't change 1285611682 Q * ntrs_ Ping timeout: 480 seconds 1285611881 J * britneypire ~britneypi@212-198-248-35.rev.numericable.fr 1285611894 M * mnemoc Bertl: restarting the guest solves the problem, but /proc/virtual/status changes. http://paste.linux-vserver.org/18008 1285611957 Q * britneypire 1285612008 M * Bertl does it constantly increase over time or on guest restart? 1285612013 M * mnemoc yes 1285612020 M * Bertl which one? 1285612059 M * mnemoc restarting the guest first test takes 3ms, i=1300 297ms 1285612125 M * mnemoc i=1500, 383ms 1285612143 M * mnemoc it's not constant, it's geometric.. 1285612150 M * mnemoc or worse 1285612284 M * bzed harry: before I build a new kernel - do you have any plan to release a new grsec patch soon? :) 1285612285 M * mnemoc 1836 574ms 1285612289 J * derjohn_mob ~aj@d046106.adsl.hansenet.de 1285612414 M * mnemoc Bertl: http://nopaste.opensde.net/?ZjQ4Mm <--- the hammering test. i=2000, 675ms 1285612434 M * mnemoc Bertl: can you reproduce it? 1285612487 M * mnemoc http://paste.linux-vserver.org/18009 <--- added proc/virtual/status after 2000 tests 1285612699 M * Bertl mnemoc: would be interesting to recreate the namespace setup with a mainline kernel (maybe talk to daniel_hozac) 1285612735 M * mnemoc i have no clue how to recreate it. daniel_hozac ? :) 1285612765 M * daniel_hozac vnamespace -n 1285612796 M * mnemoc Bertl: but, is it my greedily generated generic kernel's fault, or you could reproduce it there too running http://nopaste.opensde.net/?ZjQ4Mm ? 1285612836 M * daniel_hozac i wouldn't be surprised if it's from the shared bind mount thing. 1285612895 M * mnemoc it is, the second mount adds 2 entries to /proc/mounts. (but not in ctx 0) ... i'll try using vnamespace 1285612997 M * mnemoc and if i umount in strict reverse order this doesn't happen 1285613020 M * daniel_hozac you might want to try using umount -l to work around your problems. 1285613066 M * mnemoc thanks, i'll add that option to the code 1285613089 M * daniel_hozac i'll try reproducing the issue here when i get out of this meeting. 1285613119 M * mnemoc vnamespace -n sh ./test.sh did the 2000 iterations all taking 2ms 1285613135 M * daniel_hozac it's a bit trickier than that. 1285613146 M * mnemoc :) 1285613197 M * mnemoc i'll be here the next 3h, and i'll fix my build kit now to avoid this problem, so no rush anymore :) 1285613236 M * daniel_hozac basically, you need vnamespace -n, in that run mount -o rbind /guest/root /guest/root, mount --rshared /guest/root, and then pivot_root to /guest/root 1285613264 M * daniel_hozac creating yet another namespace in there, before the pivot_root. 1285613361 M * mnemoc and then run my test.sh? 1285613365 J * sandra_f ~sandra_f@212-198-248-35.rev.numericable.fr 1285613484 Q * sandra_f 1285613592 M * daniel_hozac right 1285614314 M * mnemoc daniel_hozac: doesn't happen. it's stable at 32ms per test 1285614361 M * mnemoc ignore that last ocmment, my script was wrong 1285614774 Q * arekm_ Quit: leaving 1285615270 J * arekm arekm@carme.pld-linux.org 1285615793 J * manana ~mayday090@84.17.25.149 1285616134 J * petzsch1 ~markus@p4FF44FF5.dip.t-dialin.net 1285616524 Q * petzsch Ping timeout: 480 seconds 1285617254 M * mnemoc daniel_hozac: after your meeting. this is the script i'm using to test inside the namespace http://nopaste.opensde.net/?NzhkM2 pivot_root fails because of "Invalid argument" but the test gave the same errors than inside the guest and took 4m39.655s (sys: 3m19.196s) while in plain ctx-0 takes 17s and gives none 1285617487 J * petzsch ~markus@p4FF45325.dip.t-dialin.net 1285617818 Q * petzsch1 Ping timeout: 480 seconds 1285618666 J * lafille ~lafille@212-198-248-35.rev.numericable.fr 1285618781 Q * lafille 1285619081 Q * bonbons Read error: Connection reset by peer 1285619733 Q * michal Ping timeout: 480 seconds 1285619860 J * michal ~michal@www.rsbac.org 1285620082 Q * ntrs__ Ping timeout: 480 seconds 1285620625 J * orzel ~orzel@sylphide-consulting.com 1285620652 M * orzel Hello. I have some "page allocation error" although i have enough ram and the swap is not used much... here's an exemple http://freehackers.pastebin.com/HyGpfxpj 1285620665 M * orzel i somehow think this could be related to vserver... but i'm not sure 1285620679 M * Bertl off to bed now ... have a good one everyone! 1285620684 N * Bertl Bertl_zZ 1285620709 M * orzel it started less than 30minutes ago, and i already have 15 of those 1285620714 M * orzel Bertl_zZ: good night :) 1285622483 Q * manana Remote host closed the connection 1285623808 J * manana ~mayday090@84.17.25.149 1285624125 Q * petzsch Quit: Leaving. 1285624580 J * petzsch ~markus@p4FF45325.dip.t-dialin.net 1285625104 Q * manana Remote host closed the connection 1285626187 Q * ghislain1 Quit: Leaving. 1285626262 J * ghislain ~AQUEOS@adsl2.aqueos.com 1285626328 Q * ghislain 1285627578 Q * urbee Ping timeout: 480 seconds 1285627884 J * urbee urbee@93-103-199-233.dynamic.dsl.t-2.net 1285628471 Q * dowdle Remote host closed the connection 1285629757 Q * petzsch Quit: Leaving.