1259800758 J * blues ~blues@agf55.neoplus.adsl.tpnet.pl 1259800877 Q * blues_ Ping timeout: 480 seconds 1259801285 Q * AmokPaule Quit: Nettalk6 - www.ntalk.de 1259801862 Q * dowdle Remote host closed the connection 1259802849 Q * FireEgl Quit: Leaving... 1259804743 J * jrklein_ ~jrklein@ppp-70-130-46-49.dsl.wchtks.swbell.net 1259806145 J * Aaron ~chatzilla@CPE001c10a88bc9-CM001947579cda.cpe.net.cable.rogers.com 1259806571 M * micah Bertl: in case you were interested the odd user-space problem that was causing that guest to not to be able to fork a /dev/pts/2, I have finally tracked it down 1259806587 M * micah Bertl: /etc/security/limits.conf has: * soft nproc 100 1259806605 M * micah and * hard nproc 150 1259806618 M * micah interesting because the guest doesn't have more than 42 procs running 1259806707 M * micah in any case, increasing the soft nproc limit to 148 (exactly), allows the ssh to fork 1259806727 M * micah and interestingly... the number of processes running on that system right now? 47 1259806744 M * micah seems like there is a digit off there :) 1259806827 M * micah i blame it on the kernel :) 1259807664 Q * Aaron Quit: ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451] 1259808563 M * Bertl micah: what kernel is that? 1259808611 M * micah Bertl: the broken one 1259808618 M * Bertl and what does /proc/virtual//* contain (could you upload the contents of those files)? 1259809013 M * micah Bertl: http://micah.riseup.net/pastes/2009-12-02T215329 1259809206 M * Bertl when you change the hostname in that guest, does it change the host name of the host too? 1259809359 M * micah bertl: i have not tried, you mean by doing: hostname foo in the guest? 1259809445 M * Bertl correct 1259809502 A * micah tries 1259809524 M * micah doesn't appear to change it in the host or the guest 1259809534 M * micah err, actually it does in the guest 1259809559 M * micah but not on the host 1259809562 M * Bertl but hostnam on the host still reports the host's hostname? 1259809567 M * micah yes 1259809596 M * Bertl okay, good, you are trying to logon as root? 1259809608 M * Bertl (when you get the nproc issue) 1259809822 M * micah no 1259809840 M * micah as a regular unprivileged user 1259809867 M * Bertl a common user on that system (host and guests)? 1259809895 M * Bertl i.e. could the sum over all processes on that system for that user reach 148? 1259809924 M * Bertl (assuming the user is identical when the uid matches) 1259809976 M * micah well common as in I just did: adduser micah, and no that user has zero processes running when I ssh in 1259810044 M * Bertl yes, but it probably got uid=500 or so, and maybe there are some processes running with that uid (on the host or in other guests)? 1259810095 M * Bertl i.e. could you check with vps auxwww how many processes are listed with that uid? 1259810148 M * micah sure 1259810167 M * micah its uid 1002 1259810180 M * micah and there are none currently 1259810278 M * Bertl hmm, okay, could you test with that guest if the behaviour still exists with a recent kernel? 1259810312 M * Bertl (although I do not see that this would be broken in 2.6.26) 1259811268 M * micah hm, in fact I can't get any limits to be applied on the newer kernel 1259811522 M * Bertl which means? 1259811566 M * Bertl when you logon to the guest (e.g. via ssh) and do 'cat /proc/self/limits' nothing shows up? 1259811577 M * Bertl (I mean no limits for that user) 1259811678 M * micah sorry, I was basing my testing on what I was seeing before... ie. the number of processes in the guest was 47, I set the soft nproc to 100 and then was able to fork via ssh. i set it down to 47 and stll was able 1259811707 M * micah but, if I cat as that user /proc/self/limits, I see: 1259811707 M * micah Max processes unlimited unlimited processes 1259811753 M * micah no 1259811760 M * micah i am mistaken. 1259811766 M * micah that unlimited was because I was logging in as root 1259811775 M * Bertl okay :) 1259811779 M * micah if I login as a regular user, I see the max number of processes I've specified in limits.conf 1259811804 M * Bertl okay, and if you lower that to e.g. 10? 1259811830 M * Bertl does the logon still work and the limit catch? 1259811865 M * micah yes 1259811893 M * micah well, what do you mean by 'and the limit catch'? 1259811924 M * Bertl if you try to spawn e.g. 9 'sleep 10' or so, it should stop you early 1259811949 M * Bertl i.e. does the limit work/apply? 1259811998 M * micah yep: -bash: fork: Resource temporarily unavailable 1259812043 M * Bertl okay, could you check for me with a different guest, but same uid, just to rule that out? 1259812150 M * Bertl it's supposed to be separated by user namespaces, but could be a bug :) 1259812242 M * micah on the newer (.31) kernel you mean? 1259812390 M * Bertl yes, well, if you like on both ... 1259812810 J * saulus_ ~saulus@c192110.adsl.hansenet.de 1259813022 M * Bertl the interesting thing is, mainline does not provide any way to show the nproc value (i.e. the current number of processes accounted to an user) 1259813193 Q * SauLus Read error: Connection reset by peer 1259813197 N * saulus_ SauLus 1259815608 Q * jrklein_ Quit: Computer has gone to sleep 1259815635 J * jrklein ~jrklein@ppp-70-130-46-49.dsl.wchtks.swbell.net 1259815655 Q * jrklein 1259816437 M * Bertl k, off to bed now .. drop me a note if you find something 1259816443 N * Bertl Bertl_zZ 1259817064 J * jrklein ~jrklein@ppp-70-130-46-49.dsl.wchtks.swbell.net 1259822428 J * balbir ~balbir@122.181.150.106 1259824120 Q * balbir Ping timeout: 480 seconds 1259825114 J * SDRam ~g7135@192.192.154.47 1259825338 Q * derjohn_mob Ping timeout: 480 seconds 1259825961 J * sharkjaw ~gab@90.149.121.45 1259826749 Q * SDRam 1259828676 J * thierryp ~thierry@zankai.inria.fr 1259828923 Q * thierryp Read error: Connection reset by peer 1259829573 J * taenzerme ~Adium@static-87-79-237-223.netcologne.de 1259830620 J * derjohn_mob ~aj@tmo-109-226.customers.d1-online.com 1259830966 J * FireEgl FireEgl@173-16-9-10.client.mchsi.com 1259831511 J * gnuk ~F404ror@pla93-3-82-240-11-251.fbx.proxad.net 1259832194 J * wibble wibble@vortex.ukshells.co.uk 1259832276 Q * Mr_Smoke Read error: Connection reset by peer 1259832323 J * Mr_Smoke ~smokey@layla.lecoyote.org 1259832442 J * scientes ~scientes@174-21-130-99.tukw.qwest.net 1259833903 Q * DoberMann[ZZZzzz] Read error: Connection reset by peer 1259833913 J * DoberMann[ZZZzzz] ~james@cap31-6-88-180-72-76.fbx.proxad.net 1259835272 J * AmokPaule ~amokpaule@brsg-4dbbc847.pool.mediaWays.net 1259835412 Q * gnuk Quit: NoFeature 1259835515 Q * FloodServ synthon.oftc.net services.oftc.net 1259835722 Q * nenolod Ping timeout: 480 seconds 1259835785 J * nenolod ~nenolod@petrie.dereferenced.org 1259837540 J * FloodServ services@services.oftc.net 1259839882 J * gnuk ~F404ror@pla93-3-82-240-11-251.fbx.proxad.net 1259840412 J * davidkarban ~david@80.250.18.198 1259841518 P * taenzerme 1259841520 J * taenzerme ~Adium@static-87-79-237-223.netcologne.de 1259843160 Q * gnuk Remote host closed the connection 1259843856 J * gnuk ~F404ror@pla93-3-82-240-11-251.fbx.proxad.net 1259847948 Q * sharkjaw Quit: Leaving 1259848118 J * fLoo ~fLoo@2001:6f8:119b:ffff::1 1259849578 N * Bertl_zZ Bertl 1259849583 M * Bertl morning folks! 1259850253 M * ghislain hello bertl 1259850839 Q * kir Quit: Leaving. 1259850894 Q * BWare Quit: *p00f* 1259850903 Q * ascii Ping timeout: 480 seconds 1259851161 Q * taenzerme Quit: Leaving. 1259851787 Q * mEDI_S Quit: mEDI_S 1259852071 J * mEDI_S ~medi@255.255.255.255.li 1259852099 J * m_o_d ~kane@host.ltv.pl 1259852111 M * ghislain i wanted to know the best solution to copy a vserver from an host to another knowing they use separate root partition and also tagging 1259852134 M * ghislain i was thinking 1/ keep the same xid 1259852220 M * ghislain 2/ mount the directory in /mnt/whatever, 3/ rsync over there, 4/ rsync the /etc/vservers directory of this guest, 5/ umount the /mnt/whatever start the guest, enter it , redo an rsync 1259852251 M * ghislain 6/ chxid -R the guest , seems a little complicated :) 1259852383 J * ascii ~ascii@host168-3-dynamic.51-79-r.retail.telecomitalia.it 1259852426 Q * scientes Ping timeout: 480 seconds 1259852438 J * taenzerme ~Adium@static-87-79-237-223.netcologne.de 1259852526 Q * badiane Ping timeout: 480 seconds 1259852655 M * ghislain i am looking at the rsync build method without understanding all about it right now, googling it right now 1259852887 M * Bertl well, the rsync build method will take care of all the things, except the tagging 1259852927 M * Bertl (except for the mounts) 1259852955 M * ghislain bertl: the last time i tried creating a vserver with build moved the mount point so if i mount the empty recipient guest directory i wonder if it will work 1259852962 M * Bertl copying the filesystem low-level will duplicate the tagging 1259852965 M * ghislain bertl: anyway i will just test this now and see :) 1259853012 M * ghislain bertl i see with dd you mean ? but the size are different i am not too confortable with this :) will try the rsync 1259853029 M * Bertl dump/restore should do for ext2/3 1259853051 M * Bertl (similar might exist for other filesystems) 1259853083 M * jpic_ hi, anybody got syslog-ng 3 to work? 1259853097 M * jpic_ i have this error: Excess number of arguments 1259853126 M * Bertl sounds like an userspace problem, what arguments do you give? 1259853148 M * ghislain bertl thx i will have a look (feel like a rookie admin now lol ) 1259853178 M * jpic_ syslog-ng -s /etc/syslog-ng/syslog-ng.conf with this config: http://pastebin.com/m6992c59a 1259853193 M * jpic_ (which is the default) 1259853306 M * Bertl first, I'd remove the /proc/kmsg 1259853317 M * Bertl (entire section) doesn't make sense in a guest 1259853345 M * jpic_ done 1259853380 M * Bertl then, I'd use the proper option, i.e. -f instead of -s 1259853394 M * Bertl (-s is syntax check) 1259853417 M * jpic_ yes, right 1259853433 M * jpic_ WARNING: the match() filter without the use of the value() option is deprecated and hinders performance, please update your configuration; 1259853459 M * jpic_ and .. Error in configuration, unresolved source reference; source='kernsrc' 1259853474 M * Bertl yeah, remove that, and fix the config file 1259853488 M * Bertl i.e. the match() invocations 1259853524 M * jpic_ thanks, works! 1259853545 M * Bertl you're welcome! but JFTR: not Linux-VServer related :) 1259853556 M * jpic_ i though it would be actually ... 1259853558 M * jpic_ sorry 1259853563 M * Bertl np 1259853569 M * jpic_ if you really want some, my kernel started oopsing recently :) 1259853584 M * jpic_ since we have apache in a vserver 1259853586 M * Bertl depends on the kernel version :) 1259853637 M * jpic_ indeed, i'm not up to date, thanks again 1259853641 M * Bertl note that apache runs in a Linux-VServer guest for all my domains and the Linux-VServer site too :) 1259853648 M * jpic_ will upgrade from vs2.3.0.36.14-grsec2.1.14-20090709 to vs2.3.0.36.24-grsec2.1.14-20091116 1259853688 M * Bertl make that and let us know if the issue persists 1259853742 M * jpic_ sure, i'll do that sunday 1259853747 M * jpic_ (night) 1259853918 J * badiane ~badiane@cpe-72-229-37-2.nyc.res.rr.com 1259854829 M * ghislain bert: the rsync method remove all the /etc/init.d daemon also :( anyway this leave me just with the /etc/ to rsync 1259854842 M * daniel_hozac ghislain: don't specify a distribution. 1259855035 M * ghislain daniel_hozac: ahah that was the parameter i should not have used :) Thanks 1259855121 M * ghislain daniel_hozac: is there a way to rsync it again or will it move the existing dir out of his way whatever trick i try to use ? :) 1259855137 M * daniel_hozac just run rsync manually. 1259855179 M * Bertl (with the proper options) 1259855205 M * ghislain lol ok 1259855228 M * ghislain this is what i liked with the build method i was sure the options were right. 1259855243 M * ghislain thanks again 1259855562 M * Bertl I usually use -axHPSD --numeric-ids 1259855674 J * dowdle ~dowdle@scott.coe.montana.edu 1259855703 J * geb ~geb@199.4.82-79.rev.gaoland.net 1259855772 J * scientes ~scientes@174-21-130-99.tukw.qwest.net 1259855976 M * ghislain bertl: thanks, was 'in the man' lol 1259855983 Q * AmokPaule Quit: Nettalk6 - www.ntalk.de 1259856071 Q * taenzerme Quit: Leaving. 1259856088 J * taenzerme ~Adium@static-87-79-237-223.netcologne.de 1259857259 J * imcsk8 ~ichavero@189.135.88.160 1259859103 J * niki ~niki@0x5553169c.adsl.cybercity.dk 1259859226 Q * Psy0rz Ping timeout: 480 seconds 1259860176 Q * derjohn_mob Ping timeout: 480 seconds 1259860182 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1259861043 J * tbradshaw ~tbradshaw@ip70-179-167-127.fv.ks.cox.net 1259861075 M * tbradshaw hello. I've got a vserver that's backed full of zombie processes (and D's) that can't be killed, and I need to reboot the vserver 1259861082 M * tbradshaw however, vserver stop fails 1259861091 M * daniel_hozac what's in dmesg on the host? 1259861095 M * tbradshaw would anyone know how I could force a reboot? 1259861158 M * tbradshaw the signal to noise in dmesg is very high, do you have any particular messages that I should be looking for? 1259861176 M * tbradshaw or message prefixes or such? 1259861229 M * daniel_hozac what kind of "noise" do you have there? 1259861242 M * tbradshaw vserver stop informs me that "A timeout occured while waiting for the vserver to finish and it will be killed by sending a SIGKILL signal. The following process list..." 1259861261 M * tbradshaw and then "Vserver '/etc/vservers/servernode' still running unexpectedly; please investigate it manually..." and "RTNETLINK answers: Cannot assign requested address" 1259861337 M * tbradshaw I tried another (failed) stop 1259861340 M * tbradshaw no lines were added to dmesg 1259861366 M * daniel_hozac they wouldn't be. 1259861378 M * daniel_hozac your processes have already crashed in kernel space. 1259861418 M * tbradshaw indeed 1259861439 M * tbradshaw the most likely contender is a crash in fs/nfs/write.c that caused a dump 1259861455 M * daniel_hozac and, do you have that? 1259861476 M * tbradshaw ? 1259861485 M * daniel_hozac the stack trace. 1259861494 M * daniel_hozac along with the rest of the oops. 1259861533 M * tbradshaw I do, though, at the moment I'm not really interested in diagnosing why the error took place. (I can follow up with investigation after), I'm trying very hard to get the vserver restarted and back into service without having to reboot the host 1259861616 M * daniel_hozac well, crashes in kernel space genereally can't really be fixed properly without a reboot. 1259861621 J * derjohn_mob ~aj@d025076.adsl.hansenet.de 1259861629 M * daniel_hozac and without knowing what the exact error was, i can't say whether it's possible to work around it. 1259861641 M * tbradshaw interesting 1259861658 M * tbradshaw so a kernel space crash on a vserver is also "fatal" for the host? 1259861666 M * daniel_hozac it's a shared kernel. 1259861682 M * daniel_hozac anywhere you crash it it's fatal. 1259861706 A * tbradshaw nods 1259861761 M * tbradshaw Well, thank you very much for your help. I'm not yet familiar with vserver based virtualization, so this was very helpful 1259861794 M * Bertl it is more 'isolation' than virtualization 1259861806 M * Bertl think bsd jails on steroids :) 1259861814 M * tbradshaw sure! that makes sense 1259861837 M * tbradshaw thanks again, I'm going to send out the downtime notification and start the reboot of the host 1259861845 M * Bertl but, you could work around the host restart 1259861864 M * Bertl assuming that the nfs still works (which seems to be required for your guest) 1259861866 P * taenzerme 1259861916 M * tbradshaw I've got some application level constraints that the hung processes are blocking 1259861918 M * Bertl you could simply change the xid (context number) of your guest and start it again ... the other processes won't go away (those locked in the kernel) but if they do not interfere with the 'new' guest 1259861971 M * Bertl when they are 'just' in D state waiting for the dead nfs, you might be able to unmount that and thus unlock them 1259861993 M * Bertl when they are part of the kernel oop, you're out of luck 1259861996 M * tbradshaw unfortunately it looks like they are D state, waiting on Z ssh permissions 1259862002 M * tbradshaw and I think nfs died reading the key 1259862011 M * tbradshaw s/permission/processes 1259862076 M * tbradshaw we're still in beta, so the downtime isn't catastrophic 1259862090 M * Bertl what kernel/patch version is that, btw? 1259862119 M * tbradshaw the power button just got pressed, I'll have that for you shortly 1259862262 J * barismetin ~barismeti@tvwna-ip-c-158.princeton.org 1259862658 M * tbradshaw Bertl: Linux web 2.6.28.7-vs2.3.0.36.7 #1 SMP Sat Feb 28 15:13:34 EST 2009 x86_64 GNU/Linux 1259862683 M * Bertl I'd suggest to upgrade to 2.6.31.x soon (with a recent patch) 1259862783 A * Wonka wants 2.6.32 with vserver patches :) 1259862820 M * tbradshaw I'm not sure what's in Debian stable, and I'm not really in a position to initiate upgrades like that, but I'll pass the advice up the chain. :) 1259862821 M * Mr_Smoke Bertl: hows the stabilization going along ? :) 1259862849 M * Bertl Wonka: ah, 2.6.32 is out ... give me a minute :) 1259862860 M * Bertl Mr_Smoke: slowly 1259862887 M * Mr_Smoke I guess I *could* jump in and try it out ... 1259862907 M * Bertl won't hurt ... feedback is always welcome 1259862925 M * Bertl tbradshaw: in debian stable are only broken kernels for now :) 1259862967 M * Mr_Smoke Bertl: "won't hurt" except when/if I run across a problem. They tend to develop when in close range to my person :) 1259863009 M * Bertl one more reason to start testing right now ... 1259863017 M * Mr_Smoke Sure :) 1259863029 M * Mr_Smoke I need to move one last website off the old server 1259863031 M * Bertl encountered problems -> bug report -> bug fix :) 1259863044 M * Mr_Smoke then I can consider upgrading 1259863050 M * Mr_Smoke i know ;) 1259863220 Q * scientes Ping timeout: 480 seconds 1259863368 M * Wonka Bertl: since about 11 hours already! ;) 1259863442 M * Bertl why didn't you tell me earlier :) 1259863459 M * Mr_Smoke Heheh 1259863500 M * Wonka Bertl: haven't seen it earlier myself - had to work on my day job 1259864240 M * Bertl Wonka: http://vserver.13thfloor.at/ExperimentalT/patch-2.6.32-vs2.3.0.36.26.diff 1259864249 M * Bertl let me know how it goes ... 1259864414 J * Psy0rz ~psy0rz@lounge.datux.nl 1259864442 M * jpic_ hi, anybody got this one to compile? patch-2.6.31.6-vs2.3.0.36.24-grsec2.1.14-20091116.diff 1259864483 M * jpic_ error raised: kernel/sched.c: In function `tg_shares_up': 1259864484 M * jpic_ kernel/sched.c:1590: sorry, unimplemented: inlining failed in call to 'cfs_rq_throttled': function body not available 1259864495 M * daniel_hozac what compiler is that? 1259864511 M * jpic_ 3.6 ... 1259864518 M * jpic_ 3.4.6* 1259864532 M * jpic_ do we *have* to migrate? 1259864549 M * daniel_hozac of what? 1259864558 M * daniel_hozac that's not what gcc error messages look like. 1259864577 M * jpic_ gcc 3.4.6* 1259864643 M * Bertl daniel_hozac: well, aparently they look like this now :) 1259864665 M * daniel_hozac i want proof that it's actually gcc producing the error first :-) 1259864674 M * jpic_ how can i do that? 1259864697 M * daniel_hozac make V=1 1259864738 M * jpic_ http://dpaste.com/128579/ 1259864827 Q * gnuk Quit: NoFeature 1259864983 M * Bertl jpic_: what's in line 1590 and 1612 of kernel/sched.c? 1259865279 M * jpic_ http://pastebin.com/m4d02f6d5 1259865291 M * jpic_ i gave you more than 22 lines 1259865321 M * jpic_ 1612 is 23 on the paste 1259865905 Q * barismetin Quit: Leaving... 1259865908 J * barismetin ~barismeti@tvwna-ip-c-158.princeton.org 1259865946 M * ghislain dam i cannot find the kernel parameter to increase the nb of vroot 1259865974 M * daniel_hozac max_vroot 1259865975 M * Bertl jpic_: definitely a mainline/gcc issue .. but you could try with 2.6.32, maybe it was already worked around, otherwise, get a known working gcc/toolchain and you should be fine 1259866000 M * ghislain thanks daniel 1259866031 M * ghislain no trace of it on the site, time to do some wiki editing 1259866053 Q * tbradshaw Quit: tbradshaw 1259866066 M * ghislain oh, cannot edit the FAQ 1259866079 M * daniel_hozac are you logged in? :) 1259866205 M * ghislain humm i dont think so never had to be in fact ;) 1259866217 M * ghislain /vmlinuz-2.6.31.6-vs2.3.0.36.26aq root=/dev/md1 ro max_vroot=20 this should work 1259866226 M * daniel_hozac assuming it's builtin. 1259866272 J * AmokPaule ~amokpaule@brsg-4dbbc847.pool.mediaWays.net 1259866345 M * ghislain yes, i am logged , in next step edit the FAQ :) 1259866361 Q * AmokPaule 1259866504 M * ghislain cannot be changed at root time i beleive ? :) 1259866523 M * Wonka Bertl: eeeh... i did not plan to build and reboot into it now... 1259866536 M * Bertl coward! 1259866558 M * Bertl ghislain: what do you want to change? 1259866590 M * ghislain bertl: vroot device number :), i edited the FAQ for the kernel parameter now 1259866613 M * ghislain bertl: i mean maximum vroot device number 1259866644 M * Bertl 'root time' = 'run time'? 1259866695 M * ghislain bertl: yes, hum late here seems i am tired :) 1259866727 M * Bertl well, as with other modules, like loop, you can unload and reload with new module arguments 1259866776 Q * fzylogic Quit: fzylogic 1259866833 M * ghislain bertl: i love to built the module IN the kernel lol so out of luck anyway this was just to put it in the faq if possible 1259866858 M * ghislain unloading will mean stopping all guest using the vroot then change then restart them so . 1259866870 M * daniel_hozac nah. 1259866880 M * ghislain dam wrong again ? 1259866885 M * Bertl you might want to change the kernel default then (for the built in) 1259866888 M * daniel_hozac i'm pretty sure you could just reload it. 1259866930 M * ghislain ok 1259867056 Q * barismetin Quit: Leaving... 1259867205 M * ghislain modprobe -r vroot max_vroot=20 ? 1259867244 M * daniel_hozac -r removes it. 1259867262 M * ghislain dam 1259867264 M * ghislain lol 1259867270 M * ghislain completly messing thing 1259867272 M * daniel_hozac but you'd actually want to put options vroot max_vroot=20 in /etc/modprobe.conf and then just do modprobe vroot 1259867274 M * ghislain let me reread that 1259867308 J * barismetin ~barismeti@tvwna-ip-c-158.princeton.org 1259867361 J * fzylogic ~fzylogic@dsl081-243-128.sfo1.dsl.speakeasy.net 1259867471 Q * imcsk8 Quit: This computer has gone to sleep 1259867512 M * ghislain ok i put this in the faq then: If 8 vroot device is not enough for you, you can add more with the kernel parameter max_vroot (exemple for built in kernel vroot /vmlinuz-2.6.31.6-vs2.3.0.36.26aq root=/dev/md1 ro max_vroot=20 ). If vroot is a module you'd actually want to put for exemple "options vroot max_vroot=20" in /etc/modprobe.conf and then just do modprobe vroot 1259867582 M * ghislain thanks a lot for the info i will change my kernel settings to add this one as a module now 1259867756 M * Bertl example (not exemple) 1259868242 M * fzylogic was there ever a solution decided upon for that OOM killer bug I brought up yesterday? 1259868300 M * Bertl no decision yet, although I tend to the following: 1259868326 M * Bertl - report the situation via klog (as error/warning) 1259868343 M * Bertl - continue with the guest being over-limit 1259868402 M * Bertl I still have to create a proper test setup for further investigation 1259868420 M * fzylogic I think shutting the context down is better in terms of being able to recover more quickly. It's also closer to the situation you have with a panic'd non-virtualized host 1259868457 M * fzylogic I could see us implementing something to watch for downed guests and start them back up automatically 1259868458 M * Bertl maybe, but it would involve killing unkillable processes, which is a little tricky 1259868466 M * fzylogic easier than scanning a log for some error message 1259868532 M * fzylogic that could very well be true. I don't know what tests the kernel uses to determine if something's killable 1259868560 M * Bertl for example, processes hanging in I/O operations are unkillable 1259868926 M * Bertl but I can offer to send a vshelper request in that case (to userspace) which in turn, can try its best to shutdown/restart the guest, if you volunteer to test that :) 1259868957 M * fzylogic I'm sure we could do that 1259868974 M * fzylogic we're already having 4 hosts crash each day. couldn't be any worse, right? :) 1259869225 Q * jrklein Quit: Computer has gone to sleep 1259869311 M * jpic_ fzylogic: what kernel? 1259869328 M * fzylogic basically anything from 2.6.28+ 1259869335 M * Bertl fzylogic: okay, I'll prepare something for 2.6.32 1259869336 M * fzylogic specifically 2.6.31.5 and .6 right now 1259869345 M * fzylogic awesome, thanks Bertl 1259869366 M * micah Bertl: hey sorry I had to run last night and today has been really demanding with work. i hope i can do that test soon though 1259869392 M * Bertl fzylogic: please prepare a test setup, where we can go over limit with a guest on demand 1259869403 M * fzylogic already got one 1259869436 J * ghislain1 ~AQUEOS@LPuteaux-151-41-11-129.w217-128.abo.wanadoo.fr 1259869786 Q * ghislain Ping timeout: 480 seconds 1259870972 J * jrklein_ ~jrklein@156.26.14.52 1259871471 J * jrklein_1 ~jrklein@156.26.8.214 1259871647 Q * jrklein_ Ping timeout: 480 seconds 1259871932 Q * bonbons Quit: Leaving 1259872048 Q * derjohn_mob Ping timeout: 480 seconds 1259872139 Q * davidkarban Quit: Ex-Chat 1259872309 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1259872367 M * Bertl fzylogic: http://vserver.13thfloor.at/ExperimentalT/delta-oomhelper-feat01.diff 1259872509 M * fzylogic Thanks, I'll let you know when it's up and running 1259872833 M * Bertl you still need to extend the vshelper to handle this case properly 1259872936 M * fzylogic of course 1259873497 M * urbee humm 1259873506 M * urbee if my server is using alot of swap space 1259873510 M * urbee does that mean i'm out of RAM ? 1259873533 M * Bertl yes, but how do you tell it is using 'swap space'? 1259873552 M * urbee well 1259873554 M * urbee htop shows 1259873560 J * rush2end_ ~rush2end@118-160-164-51.dynamic.hinet.net 1259873562 M * urbee its using 400mb out of 1gb swap 1259873567 M * Bertl inside the guest, I presume? 1259873570 M * urbee no 1259873572 M * urbee host machine 1259873610 M * Bertl ah, server means host? 1259873616 M * urbee yes, sorry 1259873663 M * Bertl well, yes, then you're out of memory. :) 1259873670 Q * rush2end Ping timeout: 480 seconds 1259873689 M * urbee thx 1259874255 Q * jrklein_1 Quit: Computer has gone to sleep 1259875321 Q * barismetin Remote host closed the connection 1259875354 J * barismetin ~barismeti@tvwna-ip-c-158.princeton.org 1259876735 J * jrklein_ ~jrklein@ppp-70-130-46-49.dsl.wchtks.swbell.net 1259877168 Q * bonbons Quit: Leaving 1259877234 J * jrklein__ ~jrklein@ppp-70-130-46-49.dsl.wchtks.swbell.net 1259877235 Q * jrklein_ Read error: Connection reset by peer 1259877417 M * ghislain1 bertl: well in french this is exemple ;) 1259877618 J * jrklein_1 ~jrklein@ppp-70-130-46-49.dsl.wchtks.swbell.net 1259877618 Q * jrklein__ Read error: Connection reset by peer 1259877767 M * jrklein_1 sorry about going online/offline a few times there. keep losing connectivity while trying to configure a 6to4 tunnel. grr! 1259877994 Q * rush2end_ Remote host closed the connection 1259878102 Q * jrklein_1 Quit: Goodbye! 1259878701 Q * niki Remote host closed the connection 1259878889 M * ghislain1 daniel_hozack: bind mount works in the fstab but we can see the real path: 1259878889 M * ghislain1 /vservers/guest01/customerdata/var on /var type none (bind) 1259878921 M * ghislain1 daniel_hozac: any way to prevent that to appear ? 1259878932 M * Bertl don't bind mount from other guests then, bind mount from a common location on the host 1259878960 M * ghislain1 bertl: i bind mount a rep of the guest into another guest path 1259878966 M * ghislain1 i mean the same guest 1259879029 M * Bertl then do it in the inner namespace 1259879047 M * ghislain1 you mean in the mtab not in the fstab ? 1259879074 M * Bertl the mtab is just for fun (and funny tools like mount :) 1259879107 M * Bertl what does /proc/self/mounts show inside the guest? 1259879130 M * ghislain1 right now i use the util's fstab :), let me see 1259879189 M * ghislain1 GUEST:ctrl01:/%(root)> more /proc/self/mounts 1259879189 M * ghislain1 rootfs / rootfs rw 0 0 1259879189 M * ghislain1 /dev/root / ext3 rw,tag,relatime,errors=continue,data=writeback 0 0 1259879189 M * ghislain1 none /dev/pts devpts rw,relatime,gid=5,mode=620 0 0 1259879189 M * ghislain1 none /tmp tmpfs rw,nosuid,nodev,noexec,noatime,size=32768k 0 0 1259879190 M * ghislain1 /dev/vserversdata/ctrl01 /donneesclient ext3 rw,nodev,noatime,errors=continue,data=writeback,usrquota,grpquota 0 0 1259879190 M * ghislain1 /dev/vserversdata/ctrl01 /var ext3 rw,nodev,relatime,errors=continue,data=writeback,usrquota,grpquota 0 0 1259879192 M * ghislain1 /dev/vserversdata/ctrl01 /home ext3 rw,nodev,relatime,errors=continue,data=writeback,usrquota,grpquota 0 0 1259879192 M * ghislain1 proc /proc proc rw,relatime 0 0 1259879192 M * Bertl (please use paste.linux-vserver.org for everything longer than 3 lines) 1259879194 M * ghislain1 oups sorry 1259879228 M * ghislain1 http://paste.linux-vserver.org/14069 that was the url but the copy didnt worked out 1259879287 M * Bertl and what are the mounts containing the host path? 1259879335 M * ghislain1 bertl: dam they disapeared ?? they were here a moment before jsut restarted the guest and .. 1259879370 M * ghislain1 the weird thing changed 1259879381 M * ghislain1 now i got two time the /donneesclient :) 1259879382 M * ghislain1 http://paste.linux-vserver.org/14070 1259879404 M * ghislain1 i guess i still do not grasp the vroot thing 1259879421 M * Bertl what kernel/patch/util-vserver do you use? 1259879451 M * ghislain1 Kernel: 2.6.31.6-vs2.3.0.36.26aq 1259879452 M * ghislain1 util-vserver: 0.30.216-pre2864 1259879535 M * Bertl okay, and what's in your guest configs fstab/mtab? 1259879582 M * ghislain1 a mess: http://paste.linux-vserver.org/14071 1259879643 M * ghislain1 if i remove the /dev/vserversdata/ctrl01 from fstab the bind mount fails 1259879663 M * ghislain1 if i put it back i got two time the main device appear in the mount 1259879699 M * Bertl well, that's at least a funny config, what's the purpose? 1259879726 M * ghislain1 basicaly i have one partition with quota where i want to put /home and /var 1259879747 M * ghislain1 the root is shared on the host for all vserver to make hashification possible 1259879749 M * Bertl okay, any problem to put the rest of that partition inside the guest? 1259879776 M * ghislain1 what do you mean by that ? 1259879799 M * ghislain1 the partition is dedicated to this guest only 1259879799 M * Bertl well, I'd simply mount the partition into the guest on e.g. /opt 1259879818 M * Bertl then do the bind mounts to /home and /var 1259879823 M * ghislain1 in the guest fstab you mean ? 1259879831 M * Bertl guest fstab is ignored 1259879876 M * Bertl forget that, it's what you are already doing, it's just confusing because of the unusual name 1259879894 M * ghislain1 ah ok you had lost me here for a second :) 1259879909 J * cuba33ci ~cuba33ci@118-168-237-103.dynamic.hinet.net 1259879910 M * ghislain1 the issue seems to be the bind mount and the vroot 1259879943 M * Bertl the vroot doesn't really matter, it's just for the quota tools 1259879968 M * ghislain1 perhaps because i create the device too late in pre-start ? 1259879978 M * ghislain1 should i do it in initialize ? 1259879985 M * Bertl but you probably need to put proper mount entries into mtab for /var and /home, I don't think they are smart enough to figure out bind mounts 1259880000 M * Bertl (the quota tools) 1259880125 M * Bertl so, taking one step back ... what was the problem again? 1259880200 M * ghislain1 if i put the vroot device only in mtab the bind mount fails, if i put the /dev/vserversdata/ctrl01 in the fstab and the vroot in mtab i get the two that appear in the guest even if they are just one 1259880222 M * ghislain1 but then the bind mount works 1259880288 M * ghislain1 and if i put all in one partition i cannot use vashify (so i guess as hard lkink do not cross partition do they ?) 1259880316 M * Bertl nope, hard links only work on the same partition 1259880333 M * Bertl further, mtab is just for the tools, nothing there affects what is mounted or not 1259880343 M * ghislain1 yes this is why i use a /var /home one and the rest on one shared partition 1259880356 M * ghislain1 ok 1259880427 M * ghislain1 i think my issue is that the vroot device need to be created before the mount so i should move pre-start to initialize to see 1259880446 M * Bertl no, vroot device can be create any time 1259880462 M * Bertl you can even set that up after the guest is completely started up 1259880464 M * ghislain1 even if the fstab use bind mount on them ? 1259880475 M * Bertl you do not bind mount a device, only a path 1259880497 M * ghislain1 yes but that path exist only inside the mounted vroot partition 1259880552 M * ghislain1 bah initialize do not change anything :) (just tested) 1259880565 M * Bertl once again, vroot is a proxy, you cannot mount it :) 1259880586 M * Bertl not as device and not as bind mount either 1259880667 M * ghislain1 so a vroot can or cannot be tied t a partition mounted on the util's fstab of the guest ? or should this partition mounted on the host 1259880691 M * ghislain1 sorry yes it can 1259880698 M * Bertl look, you know what /dev/sda1 is? 1259880712 M * ghislain1 yes 1259880720 M * Bertl okay, tell me :) 1259880727 M * ghislain1 first partition on sda disk 1259880748 M * Bertl nope, it's very likely a device node, which allows you to access a block device 1259880772 M * Bertl and out of tradition, it's the first scsi disk, first partition 1259880786 M * ghislain1 ok 1259880789 Q * FloodServ Service unloaded 1259880812 M * Bertl now, the vroot device node is similar, just that it doesn't reflect a device, it just proxies quota ioctl to a configured device 1259880820 J * derjohn_mob ~aj@d025076.adsl.hansenet.de 1259880831 J * FloodServ services@services.oftc.net 1259880831 M * Bertl which device, that you can change anytime on the host 1259880859 M * Bertl you still need to mount e.g. /dev/sda1 not the vroot proxying to /dev/sda1 1259880893 M * Bertl but, and that's the important part, you do not put /dev/sda1 into the guest, you use the proxy instead 1259880952 M * ghislain1 ok 1259880976 M * ghislain1 i see now thanks 1259881011 M * Bertl so, do all the setup without even thinking about the vroot 1259881040 M * Bertl once that works as you like it, put the vroot in place of the device (inside the guest), so that the quota tools are happy 1259881111 M * ghislain1 i think i got it 1259881207 A * ghislain1 vi various scripts and fstab 1259881590 M * ghislain1 ok it allmost works as i wanted it to work now :) thanks to you the only part left is just cosmetic 1259881630 M * ghislain1 but i should be able to work on that 1259884210 Q * ghislain1 Quit: Leaving. 1259884732 Q * barismetin Remote host closed the connection