1156637965 Q * bonbons Quit: Leaving
1156639550 M * derjohn2 ntrs_, if anyone told me what the chnages are. I only know about one scheduler fix.
1156639576 M * daniel_hozac hmm, there were scheduler fixes between rc29 and rc31?
1156639656 M * ntrs_ derjohn2, I thought you had a predetermined communication protocol with Bertl to update that page, but I guess that's not the case.
1156639736 M * derjohn2 ntrs_, i had and i will enforce it ;) You wont believe it, but sometimes even Bertl_oO and daniel_hozac dont remember what exactly the changes were 
1156639749 M * daniel_hozac that's what interdiffs are for! :)
1156639752 M * derjohn2 daniel_hozac, you told me so this eveninf !
1156639759 M * derjohn2 *evening!
1156639762 M * daniel_hozac derjohn2: that was rc28 to rc29.
1156639777 M * derjohn2 daniel_hozac, ah.
1156639815 M * derjohn2 daniel_hozac, whats the diff between rc29 and rc31 ?
1156640008 M * daniel_hozac rc31 is (IMHO): quota on XFS fixed, INFO_HIDE -> HIDE_VINFO (i don't know what the point of that one is, but there appear to be some minor changes with regard to what it does), stat syscalls added, IPv6 patch enum conflict resolved, the address removal fix announced on the mailing list, helper return code changes, persistent fixes, vc_[sg]et_vhi_name got a better return value.
1156640045 M * derjohn2 IPv6 is in? the bonbons patch??
1156640049 M * daniel_hozac no.
1156640110 M * derjohn2 helper return code fixes. does this affect util? do I need to upgrade?
1156640147 M * daniel_hozac nope, unless the helper returns a failure for the startup method... /me checks that
1156640232 M * daniel_hozac nah, seems to work fine.
1156640270 M * derjohn2 the changea are for rc30 and rc31 altogether?
1156640276 M * daniel_hozac Bertl_oO: confirmed the bug in delta-vroot-feat01, we'll have to check for vroot_get_real_bdev != NULL first.
1156640283 M * daniel_hozac there was no -rc30 for devel.
1156640288 M * daniel_hozac only for stable.
1156640300 M * derjohn2 changes there? already listed? 
1156640310 M * derjohn2 if so, i leave the rc30 out at all
1156640323 M * daniel_hozac http://daniel.hozac.com/vserver/2.0.2.deltas has the delta names ;)
1156640346 M * derjohn2 k
1156640362 M * daniel_hozac delta-net-fix02 is the address removal, delta-vci-bump01 bumps the kernel version number, persist* and helper* are the same as for devel.
1156640412 M * derjohn2 BTW: I told you, -rc30 _is_ a release ;)
1156640420 M * daniel_hozac as i said.
1156640437 M * derjohn2 but it was never on the topic 
1156640443 M * derjohn2 anyway
1156640447 M * daniel_hozac nope, because there wasn't a 2.1.1-rc30 ;)
1156640456 M * daniel_hozac and -rc31 followed pretty quickly anyway.
1156640460 N * Bertl_oO Bertl
1156640461 M * derjohn2 this was splitted in former times
1156640472 M * Bertl to clarif that: the rc30 was done for PLM
1156640481 M * derjohn2 the quota fix: its about quota inside a guest=
1156640482 M * derjohn2 ?
1156640488 M * daniel_hozac and on the host.
1156640502 M * Bertl the xfs one?
1156640504 M * daniel_hozac (i think?)
1156640506 M * daniel_hozac yeah.
1156640533 M * Bertl thing is, I didn't realize that xfs was _that_ different that it allows 'quota' without quota
1156640555 M * Bertl there is quota in the kernel, and xfs quota, which also works without 'the quota'
1156640570 M * daniel_hozac lol, that sounds really confusing.
1156640570 M * Bertl each of them define CONFIG_QUOTACTL
1156640583 M * derjohn2 Bertl, welcome back :) 
1156640596 M * Bertl thus I move the vroot/quota part to QUOTACTL
1156640604 M * derjohn2 yes, XFS has an own quota system. are you referring to that?
1156640615 M * Bertl allowing to have xfs quota without 'normal quota'
1156640675 M * derjohn2 who or what uses the stats syscalls?
1156640680 M * daniel_hozac vstatd.
1156640680 M * Bertl daniel_hozac: regarding the vroot-feat, yes, it is buggy, we'll move that test into a separate section before that gets into devel
1156640716 M * derjohn2 vstatd? Hollo_ tools? or where from?
1156640720 M * daniel_hozac yep.
1156640732 M * derjohn2 so its part of vcd ?
1156640743 M * daniel_hozac no, separate daemon, i think.
1156640748 M * daniel_hozac but part of the larger suite.
1156640763 M * Bertl the 'H-tools' :)
1156640772 M * derjohn2 the changes are all for stable and devil?
1156640782 M * Bertl lol, nice typo :)
1156640789 M * daniel_hozac haha.
1156640793 M * daniel_hozac which changes?
1156640799 M * daniel_hozac i listed them separately.
1156640809 M * derjohn2 ffoo? *scrollback*
1156640878 M * daniel_hozac Bertl: could you explain delta-hide-feat01?
1156640908 M * derjohn2 this ? INFO_HIDE -> HIDE_VINFO
1156640928 M * daniel_hozac yep, that's it.
1156641002 M * Bertl yes, it is a side product of one of my tests
1156641026 M * derjohn2 daniel_hozac, http://daniel.hozac.com/vserver/2.1.1.deltas -> 404
1156641037 M * Bertl it basically switches from INFO_HIDE to a more approriate HIDE_VINFO, cleans up obsolete checks, and moves one check 
1156641043 M * daniel_hozac derjohn2: i know, i didn't finish it :(
1156641118 M * derjohn2 daniel_hozac, Am I blind? delta-hide-feat01? is not listed in 2.0.2 changes. Is that devel only?
1156641124 M * daniel_hozac derjohn2: yes.
1156641126 M * Bertl we are not going to put the spectator stuff in 2.1.x yet, that is probably 2.3.x material (maybe late 2.1.x)
1156641179 M * derjohn2 Bertl, ack! release 2.1.1 first and ++ the version ...
1156641207 M * daniel_hozac yeah, and it doesn't really work either, right?
1156641209 M * Bertl  yes, we should get the 2.02 out this weekend
1156641213 M * derjohn2 what is delta-net-fix02 ?
1156641214 M * daniel_hozac i mean, xid 1 would be unmigratable.
1156641221 M * derjohn2 Bertl, coo.. and 2.1.1 =
1156641222 M * derjohn2 ?
1156641226 M * daniel_hozac derjohn2: "the address removal fix announced on the mailing list
1156641227 M * daniel_hozac "
1156641241 M * Bertl have to clean that up, daniel_hozac has already started with the includes
1156641251 M * derjohn2 what is a vci? and what a vci bump?
1156641265 M * daniel_hozac "delta-vci-bump01 bumps the kernel version number"
1156641277 M * Bertl so maybe if we are lucky we get a good rc for devel mid next week
1156641294 M * Bertl vci is the Vserver Configuration Information
1156641302 M * derjohn2 aaahh
1156641309 M * daniel_hozac i can never remember that... heh.
1156641329 M * derjohn2 does it provide a VS versionb number hat normal ppl lile me understand=
1156641332 M * derjohn2 ?
1156641339 M * Bertl well, it's easy, I make it up every time anew :)
1156641356 M * Bertl *reorder*
1156641365 M * daniel_hozac lol.
1156641382 M * Bertl derjohn2: zou got a testme.sh at hand?
1156641408 M * derjohn2 well, in /usr/share/doc/examples ;) yes
1156641443 M * Bertl VCI: 0002:0001 273 03000016 (TbLgnP) 
1156641469 M * Bertl shows the version 2.00.01 and the flags (vci CONFIG options)
1156641470 M * derjohn2 VCI: 0002:0102 273 031101f6 (TbLgnPD)
1156641492 M * derjohn2 Bertl, OMG! and that's in the proc now?
1156641507 M * Bertl see, you have 2.01.02 with much more options selected
1156641525 M * Bertl that is in proc and available via syscall commands for a long time now :)
1156641531 A * derjohn2 became a user made of glass just now
1156641549 M * Bertl yeah, we know all about your cat now too :)
1156641559 M * derjohn2 Bertl, why want ther be written a "VS Version: 2.1.1" ?
1156641565 M * Bertl half bit compression and such
1156641605 M * derjohn2 now that I know it, it fine for me ... but the rest of the world?
1156641659 M * daniel_hozac i guess most people already know what vserver version they're running :)
1156641686 M * derjohn2 daniel_hozac, we often have ppl here that say "my colleague did the kernel build", nor?
1156641689 M * daniel_hozac (plus it's not very specific. that could be _any_ rc...)
1156641694 M * Bertl derjohn2: it's basically a way for tools to figure the kernel interface
1156641716 M * Bertl in an ideal world, it would only be relevant to libvserver :)
1156641761 M * derjohn2 daniel_hozac, could be 2.1.1-rc33 or whatever. I agree that you wouldnt see what patchea are added, but thats always a prob.
1156641813 M * derjohn2 Bertl, in an ideal world many beautiful nude women would massage me now ;)
1156641836 M * derjohn2 But anyway: I'll try to document that.
1156641837 M * Bertl you pevert! *G*
1156641857 M * Bertl sorry, just reminded me of LSL
1156642061 M * derjohn2 (delta-helper-feat04 = the address removal fix ?
1156642081 M * Bertl hmm?
1156642104 M * derjohn2 delta-helper-feat04 .. i got that from daniel's http://daniel.hozac.com/vserver/2.0.2.deltas
1156642125 M * derjohn2 just playing puzzle to get some http://linux-vserver.org/ChangeLogStableExperimental finished
1156642151 M * Bertl ah, okay, the net-fix is the address removal
1156642166 M * derjohn2 k
1156642188 M * derjohn2 this went into devel rc31 too, I assume
1156642231 M * Bertl the helper4 fixes a bug with a missing unhash, and adds possible return code checks
1156642325 M * derjohn2 so  there are (always?) more changes than only the sum of all deltas?
1156642348 M * Bertl hmm, I hope not
1156642374 M * Bertl actually the _sum_ of all deltas would be too much change :)
1156642381 M * derjohn2 http://daniel.hozac.com/vserver/2.0.2.deltas <-- which one is the xfs fix?
1156642404 M * Bertl the both xfsquot*
1156642434 M * derjohn2 ah, not listed in daniel_hozac page. is the quota-in-guest stuff only devel?
1156642466 M * Bertl all quota related 'fixes' for now are devel only
1156642500 M * Bertl and the vroot feature is not in devel at all
1156642524 M * derjohn2 hm? stable has feature more than stable?
1156642570 M * Bertl parse error
1156642620 M * derjohn2 please? 'vroot feature is not in devel at al' -> so its  in stable?
1156642636 M * Bertl no, not stable, not devel, not even experimental :)
1156642680 M * derjohn2 vroot is only a delta by now?
1156642707 M * Bertl yes, it was (as many others) just a test patch (in this case for nayco) to test my theory
1156642710 M * FaUl whats vroot? ;-)
1156642713 M * derjohn2 delta-vroot-feat01 -> i drop that from changelog of rc31 stable.
1156642723 M * Bertl yes please :)
1156642749 M * derjohn2 FaUl, thats a very mystical  and powerful thing :)
1156642771 M * Bertl FaUl: it's a quotactl proxy
1156642782 M * derjohn2 Bertl, and dont complain about many changes in the wiki ;)
1156642800 M * Bertl ah, I slowly get used to them :)
1156642842 M * FaUl Bertl: what exactly does it do? wasn't quota supported since - err - 1.somtething? 
1156642855 M * daniel_hozac yes, through vroot :)
1156642858 M * Bertl FaUl: yes, as the vroot is :)
1156642872 J * ntrs__ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com
1156642878 Q * ntrs_ Read error: Connection reset by peer
1156642886 M * FaUl so whats new on the vroot-stuff?
1156642919 M * Bertl that it now gets camelon-like features ...
1156642941 M * FaUl it may make itselv invisible?
1156642960 M * daniel_hozac no, more like take on the shape of its parent. ;)
1156643275 M * derjohn2 n8 folks ... I hope I got the changes correctly in log.
1156643287 M * ntrs__ I am having a problem starting a virtualized service on the host.
1156643299 M * derjohn2 *tired* will re-read tomorrow
1156643299 M * daniel_hozac hmm?
1156643304 M * Bertl derjohn2: good night, tx!
1156643308 M * daniel_hozac good night derjohn2!
1156643316 M * Bertl ntrs__: let's hear ...
1156643324 M * ntrs__ messages says: "kernel: dynamic contexts disabled."
1156643343 M * ntrs__ when I try to do "/etc/init.d/v_3dm start"
1156643351 M * Bertl okay, and one of your guests doesn't have a static context oid
1156643354 M * Bertl *id
1156643366 M * daniel_hozac ntrs__: that's a pretty descriptive error, IMHO.
1156643367 M * Bertl i.e. add a fixed id to the config, and it will start fine
1156643371 M * ntrs__ what guests
1156643375 M * ntrs__ I am doing this on the hsot
1156643377 M * ntrs__ host
1156643401 M * daniel_hozac sysvwrapper? i thought that was removed from alpha util-vserver already.
1156643402 M * Bertl hmm, okay, so what does v_3dm do?
1156643409 M * ntrs__ #!/bin/sh
1156643409 M * ntrs__ # chkconfig: - 85 15
1156643409 M * ntrs__ # description: Wrapper to start httpd bound to a single IP
1156643409 M * ntrs__ USR_LIB_VSERVER=/usr/lib/util-vserver
1156643409 M * ntrs__ exec $USR_LIB_VSERVER/vsysvwrapper 3dm2 $*
1156643417 M * ntrs__ that's pretty much it
1156643432 N * ntrs__ ntrs
1156643449 M * Bertl probably the '3' at the beginning confuses it or so?
1156643465 M * Bertl haven't uses vsyswrapper for ages now
1156643471 M * ntrs 3 at the beginning of 3dm2?
1156643489 M * ntrs Well, I have to use it because the service only needs to listen on the host IP
1156643492 M * Bertl it's just a wild guess
1156643499 M * daniel_hozac probably vsysvwrapper doesn't support static contexts.
1156643504 M * daniel_hozac like, at all.
1156643515 M * Bertl what about doing chbind --ip -- ...
1156643520 M * daniel_hozac it's legacy stuff.
1156643525 M * ntrs well, it used to work before this upgrade.
1156643536 M * daniel_hozac ntrs: well, you disabled dynamic contexts in the kernel ;)
1156643542 M * Bertl yes, zou probably disabled the dynamic contexts
1156643547 M * Bertl *you
1156643552 M * ntrs Ok, what is the proper "new" way of starting a service to listen only on the host IP?
1156643571 M * daniel_hozac chbind --nid X --ip Y -- ...
1156643589 M * ntrs Right I disabled the dynamic context because they were marked as obsolete and the description said that they will most likely be removed completely in the next version
1156643590 M * Bertl but probably the wrapper should be updated
1156643623 M * daniel_hozac that'd require the dynamic-contexts-in-userspace stuff though.
1156643624 M * Bertl (or replaced by something better :)
1156643627 M * daniel_hozac which isn't there yet.
1156643668 M * ntrs Ok, can someone give me an example on how to run this service now?
1156643697 M * Bertl just make the v_3dm script contain the line
1156643701 M * daniel_hozac exec chbind --nid 8 --ip 127.0.0.1/8 -- 3dm2 $*
1156643711 M * Bertl for example
1156643718 M * ntrs in the script?
1156643722 M * daniel_hozac sure, why not.
1156643734 M * Bertl adjust the '8' to be different from your context ids
1156643737 M * daniel_hozac you'll have to make sure the nid doesn't collide with a guest or another service though.
1156643738 M * ntrs will there be any negative effect?
1156643773 M * Bertl no, it's a normal static network context
1156643791 M * ntrs Ok
1156643794 M * ntrs Let me try that
1156643797 M * Bertl and for the --ip part you want the 'allowed' ips
1156643798 M * daniel_hozac (which admittedly weren't normal at all until 0.30.210)
1156643815 M * ntrs Bertl, sure.
1156643844 M * daniel_hozac Bertl: OH! we could use vc_vci_config for checking for dynamic contexts, right?
1156643862 M * Bertl yep :)
1156643891 M * daniel_hozac so i still want it in stable then :)
1156643930 M * Bertl proc isn't enough for that, okay :)
1156643941 M * daniel_hozac parsing /proc is ugly, IMHO.
1156643975 M * Bertl agreed, will look into it, should be harmless
1156643980 M * ntrs Ok, thanks everyone, the service is working fine again. Thanks again for the help.
1156643981 M * daniel_hozac yeah.
1156643994 M * Bertl ntrs: you're welcome!
1156645360 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com
1156645363 Q * ntrs Read error: Connection reset by peer
1156647319 Q * yarihm Quit: This computer has gone to sleep
1156649103 Q * id23 Ping timeout: 480 seconds
1156649636 J * id23 ~id@p50811546.dip0.t-ipconnect.de
1156649644 M * Bertl wb id23
1156650144 Q * coocoon Quit: KVIrc 3.2.0 'Realia'
1156650196 J * mire ~mire@212-167-222-85.COOL.ADSL.VLine.Verat.NET
1156650628 M * Skram hi, Bertl 
1156650649 M * Bertl hey Skram! up early? or late?
1156650717 M * Skram Well, considering what all I have done today, I have been up for a while, but its only 11pm here
1156650954 M * Bertl ah, it's 6am here ...
1156650972 M * Bertl and I think I'm off to bed now ... have a good one everyone!
1156650982 N * Bertl Bertl_zZ
1156651274 M * Skram Ah
1156651275 M * Skram Peace
1156654488 Q * derjohn2 Ping timeout: 480 seconds
1156654901 J * derjohn2 ~aj@dslb-084-058-209-187.pools.arcor-ip.net
1156656938 Q * Hunger charon.oftc.net europa.oftc.net
1156656991 J * Hunger Hunger.hu@Hunger.hu
1156657064 J * _Hunger Hunger.hu@213.163.11.138
1156657222 Q * Hunger Ping timeout: 480 seconds
1156659287 J * dna ~naucki@196-250-dsl.kielnet.net
1156664316 J * Viper0482 ~Viper0482@p54976EF1.dip.t-dialin.net
1156666885 J * bonbons ~bonbons@83.222.36.236
1156669962 J * matti matti@linux.gentoo.pl
1156670111 M * harry damn
1156670115 M * harry vserver host is going down :(
1156670122 M * harry and i don't know why :((((((((
1156670152 M * harry Connected to gandalfi.cc.kuleuven.be.
1156670152 M * harry Escape character is '^]'.
1156670152 M * harry Connection closed by foreign host.
1156670187 M * matti Hi harry.
1156670362 M * matti harry: If you have any ttyS*, you can use serial console, and see... if there's something before host crashes - this is very useful in some cases.
1156671228 M * harry matti: i'm 3km away from the server
1156671237 M * matti harry: I am 700 km away.
1156671238 M * matti ;)
1156671238 M * harry and there is a consoleswitch... which i can reach
1156671243 M * harry i don't have the software
1156671246 M * matti harry: Still, serial console is handy :)
1156671283 M * matti harry: Have at least another machine near this one that crashes?
1156671472 M * matti harry: Maybe netconsole then?
1156671489 A * matti only want to help a bit :-(
1156671752 M * harry :)
1156671759 A * harry gotta run now... get home ;)
1156671760 M * harry sry
1156671766 M * harry collegue is fixing it ;)
1156671939 M * matti ;-)
1156671970 N * _Hunger Hunger
1156673778 J * nayco ~nayco@lns-bzn-26-82-254-108-216.adsl.proxad.net
1156673864 M * nayco Hello, all !
1156674127 M * daniel_hozac hi
1156674236 J * coocoon ~coocoon@p54A051B1.dip.t-dialin.net
1156674357 M * Hollow daniel_hozac: do you know the flag combination to display a fake init process? i.e. for init-less guests
1156674472 M * Hollow i can't get it to work
1156674494 M * Hollow i tried VXF_INFO_INIT, but that is for guests with init only
1156674522 M * daniel_hozac VXF_INFO_INIT|VXF_STATE_INIT?
1156674574 M * daniel_hozac at least, that's what my guests have, and they get a blend-through init.
1156674585 M * Hollow state_init triggers vx_set_init and vx_set_reaper, i tried to do the same with migrate flags, then the process is pid = 1 indeed, but for the init-less case this process (the rc script) vanishes
1156674608 M * daniel_hozac right.
1156674611 M * Hollow stable or devel?
1156674622 M * daniel_hozac that should automatically make it blend-through init, IIRC.
1156674630 M * daniel_hozac stable on that box, let me check the test box.
1156674637 M * Hollow i thought that too, but it doesn't seems to work on 2.1.1-rc31
1156674811 M * daniel_hozac same flags set on the devel box.
1156674859 M * daniel_hozac fs/proc/base.c:#define VXF_FAKE_INIT (VXF_INFO_INIT|VXF_STATE_INIT)
1156674895 M * Hollow let me check that again
1156674961 M * Hollow argh.. dump me.. state_init can only be set on ctx_create not afterwards
1156674966 M * Hollow *dumb
1156674979 M * daniel_hozac ah, heh.
1156674995 M * Hollow yup, works now
1156675059 M * Hollow hehe.. that's interesting..
1156675077 M * Hollow if i have state_init set, and migrate with set_init afterwards, i get two init processes :D
1156675090 M * daniel_hozac really?
1156675095 M * Hollow http://paste.linux-vserver.org/306
1156675168 M * daniel_hozac very interesting...
1156675194 M * Hollow ps ax dies of course... and if i migrate now without set_init the old blend-through is visible again ;)
1156675303 M * daniel_hozac yeah.
1156676684 Q * dna Quit: Verlassend
1156677395 M * daniel_hozac hmm, /proc gets two identical directories when you do that.
1156678390 M * derjohn2 is there a problem r-limiting a guests AS and RSS while not setting virt_mem? Since then the guests get "bash: fork: Cannot allocate memory", but with 1 GB AS the mem mgmt should be pretty happy with what we are doing inside.cat /proc/virtual/2000/limit gave me "VM:       261923               0/  262143         262144/  262144         3800": 3800 hits on VM !
1156678438 M * daniel_hozac hehe, ouch.
1156678442 M * daniel_hozac i guess you set your limit too low ;)
1156678464 M * derjohn2 daniel_hozac, 256K*4k (page) = 1 GB ?
1156678472 M * daniel_hozac Hollow: IMVHO fs/proc/base.c:proc_pid_visible is to blame for this, would you agree?
1156678478 M * derjohn2 there where time machines had less physical mem !
1156678495 M * daniel_hozac Hollow: or maybe the fact that vx_set_init doesn't unset VXF_STATE_INIT.
1156678520 M * daniel_hozac derjohn2: it's address space. it doesn't really say anything about how much memory is actually used.
1156678636 M * derjohn2 daniel_hozac, hm, interesting. but I i dint limit that, on guest could exhaust all swap (or use so much swap the the box trashes)
1156678646 M * Hollow daniel_hozac: probably a combination of both ;)
1156678659 M * daniel_hozac Hollow: right, i guess the first is just a cascading effect of the latter.
1156678747 M * daniel_hozac derjohn2: sure, but that would require significant effort... i mean, you don't just tell the kernel "here, swap this out".
1156678750 M * Hollow imo you need a check if state_init is set, if yes, is there an initpid set? else blend-throug
1156678753 M * derjohn2 daniel_hozac, so the libc or such dont rely on "max physical mem" reported by the kernel? (i.e. in allocation strategies). Thats why i'll try virt_mem now.
1156678786 M * daniel_hozac derjohn2: i doubt it.
1156678802 M * daniel_hozac derjohn2: and even if they did, they wouldn't use it at all for address space allocation.
1156678814 J * ex ex@valis.net.pl
1156678851 M * derjohn2 daniel_hozac, if you allocate mem exactly in site of max RSS and write 111111 in the space, then all would get swapped?
1156678890 M * daniel_hozac derjohn2: hmm? why would it get swapped?
1156678919 M * derjohn2 because the guest has only 512MB AS 
1156678929 M * derjohn2 nor?
1156678933 M * daniel_hozac what?
1156678969 M * derjohn2 well, I limit the guest to lets say 512 MB AS and 1 GB RSS. That was the idea to cope with evil guests.
1156678985 M * derjohn2 lart me, if I do bullshit!
1156678987 M * daniel_hozac hmm, less AS than RSS? does that make sense at all?
1156678999 M * daniel_hozac i thought AS included RSS.
1156679016 M * derjohn2 vice versa ;)
1156679025 M * daniel_hozac what? RSS includes AS?
1156679033 M * derjohn2 as = 262144, rss = 131072
1156679044 M * derjohn2 sorry for the confusion
1156679127 M * daniel_hozac that's a rather odd ratio.
1156679140 M * daniel_hozac most of my guests have a 1:4 ratio for rss:as.
1156679144 M * derjohn2 lets asumme you have 4 guests and 2 GB RAM and 1 GB swap. My idea ws, that it would get pretty hard for them to exhaust all RAM.
1156679160 M * derjohn2 well, ok. how much swap do you use?
1156679201 M * daniel_hozac like right now, or the size of the swap partition?
1156679205 M * derjohn2 I try adapting the ratio and maybe add some swapspace
1156679210 M * derjohn2 the swap partition
1156679223 M * daniel_hozac 2 GiB RAM and 1 GiB swap :)
1156679254 M * s0undt3ch should the swap be at least equal to ram in size?
1156679273 M * derjohn2 under normal circumstances I dont see any advantage in having a swap > 1 GB. if that much swap is used, the disks a close to exploding 
1156679319 M * derjohn2 s0undt3ch, if you have 4 GB and 4 GB swap ... imagien a process that uses 8 GB and has continously swap 4 GB in and out....
1156679338 M * s0undt3ch yeah, pretty har on HD
1156679341 M * derjohn2 s0undt3ch, that machine wont doanything else then swapping
1156679342 M * s0undt3ch *hard
1156679406 M * daniel_hozac the idea of having at least the size of your RAM is that you can dump kernel core to swap, IIRC.
1156679448 Q * coocoon Quit: KVIrc 3.2.0 'Realia'
1156679549 M * s0undt3ch daniel_hozac: well, I just read that rule somewhere, and that rule was 2x more swap than ram
1156679560 M * daniel_hozac right.
1156679599 J * coocoon ~coocoon@p54A0580C.dip.t-dialin.net
1156679806 J * yarihm ~yarihm@84-74-17-70.dclient.hispeed.ch
1156680566 Q * lylix helium.oftc.net oxygen.oftc.net
1156680566 Q * bogus helium.oftc.net oxygen.oftc.net
1156680566 Q * hap helium.oftc.net oxygen.oftc.net
1156680566 Q * nokoya helium.oftc.net oxygen.oftc.net
1156680566 Q * [PUPPETS]Gonzo helium.oftc.net oxygen.oftc.net
1156680566 Q * mcp helium.oftc.net oxygen.oftc.net
1156680566 Q * wenchien helium.oftc.net oxygen.oftc.net
1156680566 Q * tokkee helium.oftc.net oxygen.oftc.net
1156680566 Q * abi helium.oftc.net oxygen.oftc.net
1156680579 J * lylix ~eric@dynamic-acs-24-154-53-234.zoominternet.net
1156680579 J * bogus ~bogusano@fengor.net
1156680579 J * abi ~abi@enz.schiach.de
1156680579 J * wenchien ~wenchien@221-169-69-23.adsl.static.seed.net.tw
1156680579 J * nokoya young@hi-230-82.tm.net.org.my
1156680579 J * mcp ~hightower@wolk-project.de
1156680579 J * tokkee tokkee@casella.verplant.org
1156680579 J * hap ~penso@212.27.33.226
1156680579 J * [PUPPETS]Gonzo gonzo@langweiligneutral.deswahnsinns.de
1156681654 J * dna ~naucki@51-228-dsl.kielnet.net
1156684853 J * snaqo ~ugah@20151168167.user.veloxzone.com.br
1156686247 M * derjohn "out of memory and no killable processes". might that have to do with the fix from rc28 to rc29 that daniel_hozac mention before?
1156686259 M * daniel_hozac hmm?
1156686299 M * derjohn daniel_hozac, you told about a fix that could make guest get s(t)uck, which was fixed in rc29 ?
1156686312 M * daniel_hozac that was a scheduler thing.
1156686367 M * daniel_hozac hmm, linux-vserver.org down?
1156686397 M * derjohn I get the error (kernel panic - I have a digi photo here), with the guest limited to 30000pages/600000pages, with a machine that has 2GB RAM and 1 GB swap ... only one guest.
1156686406 M * derjohn daniel_hozac, yup down here, too
1156686417 M * derjohn daniel_hozac, 3 mins ago a made a change ....
1156686489 M * daniel_hozac works again now.
1156686496 M * derjohn i an guest above there runs a (huge) postgres in. i assume i crashed when postgrss does it's auto vaccum analyse
1156686544 M * derjohn daniel_hozac, BTW: the scheulder fix is not mention in the changelog .. or is that the "locks assertion removed"
1156686568 M * daniel_hozac indeed, it's not mentioned...
1156686633 M * derjohn devel and stable?
1156686660 M * derjohn i add: * scheduler fix: scheduler could possibly 'hang' a guest
1156686661 M * daniel_hozac no, just devel.
1156686693 M * daniel_hozac 2006-07-30T23:01 has someone with the "no killable process" as well.
1156686698 M * daniel_hozac i knew i recognized it :)
1156686733 M * derjohn step by step ...  so .. changelog changed ;)
1156686743 M * derjohn grr .. wiki slow
1156686816 M * daniel_hozac btw, stable didn't get the stat syscalls, and thus not the IPv6 conflict resolution as it was never present.
1156686836 M * daniel_hozac i don't think the vc_[sg]et_vhi_name fixes were added to stable either.
1156687262 M * derjohn ok. I'll change.
1156687284 M * derjohn daniel_hozac, where do you see the mail 2006-07-30T23:01 ? TZ ?
1156687317 M * derjohn i searched my mozilla for "no killable process" -> {} 
1156687491 N * Bertl_zZ Bertl
1156687495 M * Bertl morning folks!
1156687526 M * phreak`` morning Bertl!
1156687572 M * derjohn Bertl, moin !
1156687580 M * derjohn phreak``, hi ;)
1156687589 M * phreak`` heya andreas :)
1156687615 M * phreak`` baaah
1156687625 A * phreak`` goes hurting util-vserver/dietlibc
1156687669 M * phreak`` I still can't figure why a -fPIE compiled dietlibc is breaking util-vserver
1156687801 M * Bertl probably because of register assignment oddities?
1156687823 M * Bertl btw, IIRC, doener and I did a fix on PIC + fp
1156687840 M * phreak`` Bertl: no idea :) thats what it makes it so hard for me to debug ;)
1156687901 M * daniel_hozac derjohn: IRC logs.
1156687904 M * derjohn daniel_hozac, I found something form blues in the chatlog.
1156687905 M * derjohn race ;)
1156688057 M * derjohn Bertl, was that "no killable proceses" probelm solved? (you talked to that to 'blues')
1156688119 M * daniel_hozac phreak``: breaking how? are you also compiling util-vserver with -fPIE?
1156688181 M * Bertl derjohn: url/details?
1156688181 M * phreak`` daniel_hozac: yeah, also compiling it with -fPIE ...
1156688192 M * phreak`` daniel_hozac: thats how it breaks: http://rafb.net/paste/results/GoD3yN79.html
1156688205 M * daniel_hozac Bertl: http://irc.13thfloor.at/LOG/2006-07/LOG_2006-07-30.txt
1156688219 M * phreak`` ^^ but that looks more like its ssp (stack-smashing-protection) related
1156688253 M * matti Yes.
1156688260 M * matti ssp in gcc.
1156688284 M * phreak`` matti: guess you came over that the other time (when you bugged solar in #gentoo-hardened) :)
1156688306 M * matti phreak``: Not really.
1156688316 M * matti phreak``: In my case pax was involved.
1156688544 M * FaUl Bertl: arround?
1156688552 M * FaUl do you still look for some mips-mashine?
1156688689 M * Bertl yep, mips is still missing here
1156688785 M * FaUl a friend of mine wants to give some away
1156688793 M * matti phreak``: Heh. Sorry if I "bugged" him to much. I only asked for help/hint. I'll not do this again.
1156688819 M * FaUl http://www.arcadezentrum.com/north/indigo/indigo1.jpg <- one like this
1156688825 P * snaqo 
1156688838 M * phreak`` matti: nah, don't think so (wasn't too much)
1156688845 J * admin-ybanafa ~admin@198.69.12.113
1156688863 M * derjohn Bertl, http://volatile.derjohn.de/kernel-no-killable.jpg
1156688888 M * derjohn Bertl, that what I call a screenshot ;)
1156688895 M * matti derjohn: LOL
1156688901 M * matti ;]
1156689013 M * phreak`` derjohn: haha, 900M swap and 0KB left ? oO :)
1156689051 M * matti phreak``: :)
1156689056 M * derjohn Bertl, daniel_hozac  the thing is that I limited the guest (only one guest) to 300000 rss and 600000 as. that guest shouldnt be able to get more than *calc* ~2.4 GB, while the machine has 2GB phy RAM + 1 GB swap
1156689146 M * Bertl welcome admin-ybanafa!
1156689154 M * Bertl derjohn: okay, and?
1156689198 M * Bertl derjohn: you get the OOM + no killable process?
1156689230 M * derjohn Bertl, yes, see scrennshot. All RAM exhausted. And: No processe son the host (except shh)
1156689246 M * derjohn s/RAM/VM/
1156689250 M * Bertl okay, that's interesting ...
1156689286 M * derjohn blues reported about the same thing: http://irc.13thfloor.at/LOG/2006-07/LOG_2006-07-30.txt
1156689289 M * Bertl no, the issue was discussed, but not further investigated/tested, because of lack of more info/test space
1156689309 M * Bertl yes, I remember, question is, can we reproduce that somehow?
1156689324 M * Bertl can we get a tasklist dump from the kernel?
1156689333 M * derjohn well, coped with the problem by changeing the AS/RSS limit, that solved the issue for some weeks.
1156689385 M * derjohn but now the problem is "back", but its an production machine .....
1156689397 M * derjohn *gruebel*
1156689403 M * Bertl would be interesting to make a test setup, with reduced main memory, let's say 256M or less and artificial limits?
1156689411 M * Bertl like having minimal swap, etc?
1156689453 M * derjohn OMG ... i forgot that there is a rsync process which is cron-ed froma  different machine. maybe that requests memory, too!
1156689468 M * derjohn but: all that shouldnt end up in a kernek panic, nor?
1156689480 M * derjohn *kernel
1156689502 M * Bertl it really depends
1156689516 M * Bertl by default, overcommitment is enabled in linux
1156689534 M * Bertl this basically hands out more memory than is actually available
1156689591 M * Bertl so once the real memory is exhausted, OOM killer pays a visit
1156689613 M * Bertl for whatever reason, all processes seem to be unkillable at that point
1156689632 M * Bertl (probably a nasty bug in the kernel)
1156689633 M * mnemoc Bertl: hi, do you know any updated document about network performance tweaks? 
1156689727 M * daniel_hozac Bertl: did you see Hollow's report about the double init process?
1156689737 M * Bertl mnemoc: no, nothing up-to-date .. sorry
1156689741 M * mnemoc :(
1156689743 M * Bertl daniel_hozac: no?
1156689782 M * Bertl daniel_hozac: but I remember removing a comment saying something like: 'remove duplicate fake init'
1156689783 M * mnemoc Bertl: and would you recommend any not up-to-date?
1156689812 M * Bertl mnemoc: I probably can in a few hours, no x11 atm
1156689816 M * derjohn Bertl, so i could try set the overcommitment quotra lower somewhere? proc? sysfs ?
1156689843 M * daniel_hozac Bertl: if you migrate with VXM_SET_INIT and the context already has VXF_STATE_INIT|VXF_INFO_INIT set, the init process will be shown twice.
1156689849 M * Bertl derjohn: you can increase the swappiness, but that wont help you there
1156689849 M * mnemoc Bertl: cool, i'll wait all you need :p
1156689925 M * daniel_hozac Bertl: http://paste.linux-vserver.org/307
1156689937 M * derjohn what about /proc/sys/vm/min_free_kbytes ?
1156689958 M * daniel_hozac Bertl: IMHO vx_set_init should unset VXF_STATE_INIT.
1156690051 M * Bertl daniel_hozac: I have to state that here officially! this is the _best_ bug report I did get for Linux-VServer yet! I'm impressed! I wish all were like this one.
1156690080 A * Hollow looks
1156690099 M * Bertl two lines vcmd, showing the problem, we can easily collect those for automated regression testing
1156690116 M * Hollow oh.. and i thought it was ironical
1156690117 M * Hollow :P
1156690147 M * Bertl no, seriously, I meant it ...
1156690167 M * Bertl daniel_hozac: good work!
1156690185 M * daniel_hozac hehe, i just converted Hollow's snippet to something i could reproduce.
1156690193 M * daniel_hozac (http://paste.linux-vserver.org/306)
1156690215 M * Bertl yes, but that's the way we can handle bugs properly
1156690235 M * Hollow imo vcmd has an awful syntax
1156690252 M * daniel_hozac it's a developer's tool, not something you expose to users ;)
1156690252 M * derjohn ah tehre is also /proc/sys/vm/overcommit_ratio. it's 50. hm.
1156690264 M * Bertl Hollow: doesn't matter, as long as it gets the job done
1156690275 M * daniel_hozac i think of it as syscalls from bash :)
1156690282 Q * coocoon Quit: KVIrc 3.2.0 'Realia'
1156690289 M * Bertl yes, that's the idea behind it
1156690437 P * admin-ybanafa Leaving
1156690448 M * derjohn Bertl, I rose min_free_kbytes from 3831 to 8196. let's see if it helps. i dunno what to do with overcommit_memory and overcommit_ratio.
1156690577 M * phreak`` ok, its ssp ;(
1156690659 M * Bertl derjohn: IMHO you should track down the issue
1156690703 M * derjohn Bertl, I willing to install the test system mentioned and even give you _root_ on the host.
1156690754 M * derjohn Bertl, currently it seems to be reproducable every night, i only changes the value mentioned above.
1156690780 M * derjohn i should add it's a 2.6.17.6-vs2.1.1-rc26-p3-squash-drbd-256ip
1156690877 M * matti derjohn: What is "squash-drbd-256ip"?
1156690878 M * matti ;]
1156690912 M * Bertl derjohn: try to get it down to a few minutes (on the test system) as suggested (low memory, almost no swap, etc)
1156690932 M * derjohn matti, realted to a sport like tennis with many extra balls :)
1156690964 M * matti ;p
1156691026 J * lilalinux ~plasma@80.69.35.186
1156691035 M * derjohn matti, i assume your Q to be a joke, nor?
1156691138 M * derjohn Bertl, how to lower the mem? append line is OK mem= ? I have to look how to shrink the swap partition
1156691142 M * matti derjohn: Well, I presume that squash is a squashfs? But I don't know what drbd is... Heh.
1156691184 M * derjohn drbd - distributed block device a.k.a. network raid one 
1156691254 M * derjohn you can make virtual decive that stores its data over network on two machines. if one crahed, you can go on with the second , only losing a second of data
1156691276 M * matti ;-)
1156691375 M * derjohn matti, i am not using it in the machine in question , but I usually use only one kernel for all out hosts
1156691401 M * derjohn i think about adding unionfs to my kernel release, but other things first ;)
1156691452 M * matti :)
1156691458 J * coocoon ~coocoon@p54A05C96.dip.t-dialin.net
1156691667 M * Bertl derjohn: yes, regarding swap, just add a new swap partition (small one) and remove the old one
1156691833 M * derjohn Bertl, I have an appointment with the family. After I return (I guess 22h) I will be here any try to reproduce.
1156691862 M * Bertl excellent, should have my X running till then too
1156692682 Q * lilalinux Remote host closed the connection
1156692900 M * daniel_hozac Bertl: http://daniel.hozac.com/vserver/delta-initpid-fix01.diff
1156692943 M * Hollow daniel_hozac: well, the state_init trigger uses vx_set_init as well... wouldn't that render fakeinit useless?
1156693086 M * daniel_hozac Hollow: hmm?
1156693107 M * Hollow in vc_set_flags
1156693116 M * daniel_hozac right.
1156693196 M * daniel_hozac IIUC, vc_set_cflags should be called with VXF_STATE_INIT in the mask and 0 in the flagword to make the current process the fake init one.
1156693205 Q * Viper0482 Ping timeout: 480 seconds
1156693219 M * daniel_hozac i.e. unsetting it is what triggers that.
1156693254 M * Hollow ah, right..
1156693359 M * daniel_hozac they both seem to work on my test system, but once you add a real init to the guest, you won't get the blend-through one back.
1156693397 M * daniel_hozac (both == guest with fake and guest with real init both work)
1156693457 M * Hollow i wonder why VXF_FAKE_INIT is not a normal context flag
1156693465 M * daniel_hozac if we want to get the blend-through init back, i guess we'll have to fix fs/proc/base.c:proc_pid_visible to also check vxi->initpid != 0.
1156693472 M * daniel_hozac because it's a combination of two :)
1156693543 M * daniel_hozac s/!=/==/
1156693559 M * Hollow i mean make it a new one
1156693774 J * Viper0482 ~Viper0482@p549760E3.dip.t-dialin.net
1156693782 M * phreak`` now i need a way to see why util-vserver/dietlibc is dying when ssp-enabled ;( (http://paste.linux-vserver.org/308)
1156693813 M * phreak`` Bertl: iirc you used objdump on something (when you helped me fixing some syscall issues wrt. to pie)
1156693941 M * daniel_hozac phreak``: are you sure dietlibc supports that?
1156693962 M * phreak`` daniel_hozac: nope, I'm not sure, thus I'm trying to debug it :)
1156693968 M * daniel_hozac hehe, ok.
1156694048 M * daniel_hozac objdump -d is usually pretty helpful.
1156694097 M * daniel_hozac (when you've stared at it for a few days trying to grasp the logic...)
1156694110 M * Wonka *g*
1156694187 M * phreak`` daniel_hozac: :)
1156694311 Q * FireEgl Ping timeout: 480 seconds
1156694794 Q * lylix Quit: Download Gaim: http://gaim.sourceforge.net/
1156695596 J * |gerrit| ~kvirc@dslb-084-060-197-027.pools.arcor-ip.net
1156695614 Q * GNUcifer Remote host closed the connection
1156695662 M * daniel_hozac phreak``: can you even compile a simple int main() { return 0; } with that dietlibc?
1156695938 M * phreak`` daniel_hozac: compiling yeah, but not executing
1156695951 M * daniel_hozac right, that's what i meant :)
1156696008 M * phreak`` daniel_hozac: http://paste.linux-vserver.org/309
1156696063 M * daniel_hozac heh.
1156696095 M * daniel_hozac what is creating that message?
1156696116 M * Hollow we compile dietlibc with WANT_STACKGAP, right?
1156696150 M * phreak`` Hollow: I didn't disable it (and didn't filter -fstack-protector(-all)) on this run
1156696305 M * Bertl FaUl: hmm, hmm, did he get linux running on that one?
1156696323 M * Bertl FaUl: (sorry for the long delay, no graphical browser till now)
1156696366 J * FireEgl FireEgl@Sebastian.Atlantica.US
1156696383 M * Bertl FaUl: and did he use it as 'Big Iron' (judging from the picture :)
1156697481 J * cehteh ~ct@cehteh.homeunix.org
1156697645 M * cehteh i am trying to setup the new vpn feature of ssh in vserver but it cant create the tun devices 
1156697647 M * cehteh SIOCSIFADDR: Permission denied
1156697647 M * cehteh tun0: ERROR while getting interface flags: No such device
1156697689 M * cehteh the device node is there and and i granted CAP_NET_ADMIN and CAP_NET_RAW .. but still doesnt work, any idea?
1156697691 M * Bertl precreate the tun device on the host instead
1156697712 M * Bertl IIRC, derjohn did make a howto for that, just ask google
1156697731 M * cehteh not openvpn but ssh
1156697748 M * Bertl yes, same problem, same solution
1156697749 M * cehteh and i am reading the page on the wiki 
1156697762 M * Bertl I have ssh running with -w and a persistant tun device
1156697765 M * cehteh lemme retry 
1156697771 M * cehteh ok
1156697804 M * cehteh are devices also in diffrent namespaces?
1156697829 M * daniel_hozac hmm?
1156697853 M * daniel_hozac tun0 is an interface. if it has an IP address that the guest is allowed to use, you should be able to see it.
1156697862 M * cehteh i get the same error when trying to create a tun device in the root
1156697878 M * cehteh cant create the tun device
1156697899 M * daniel_hozac with openssh?
1156697917 M * cehteh just ifconfig on the root server
1156697931 M * daniel_hozac huh?
1156697946 M * daniel_hozac you need to create the tun before you can configure it.
1156697950 M * cehteh starmaster:~# ls /dev/net/
1156697950 M * cehteh tun
1156697950 M * cehteh starmaster:~# lsmod | grep tun
1156697950 M * cehteh tun                     7296  0 
1156697951 M * cehteh starmaster:~# ifconfig tun1 10.0.50.1 10.0.99.1 netmask 255.255.255.252
1156697953 M * cehteh SIOCSIFADDR: No such device
1156697959 M * daniel_hozac http://daniel.hozac.com/vserver/test-tun.c
1156697974 M * daniel_hozac ./test-tun -d tun0 -p
1156697983 M * daniel_hozac would create the tun0 interface.
1156697990 M * cehteh ah ok
1156697998 M * daniel_hozac it's not an alias.
1156698006 M * daniel_hozac it's not created because you configure it.
1156698084 M * cehteh gotcha
1156698100 M * cehteh is there no standard tool for creating tun devices this way?
1156698119 M * daniel_hozac not AFAIK.
1156698128 M * daniel_hozac derjohn used some UML utility, IIRC.
1156698149 M * Bertl yes, there is one, tunctl
1156698155 M * daniel_hozac part of?
1156698168 M * Bertl originally part of UML, now a separate tool, iirc
1156698187 M * daniel_hozac oh.
1156698194 M * cehteh ah ok
1156698222 M * cehteh and i sstill need to put/configure this tun in /etc/vserver/xxx/interfaces ?
1156698328 M * daniel_hozac yep.
1156698340 M * daniel_hozac unless you want to give the guest that ability.
1156698358 M * cehteh not really
1156698410 M * cehteh ok ... i think that helps me for now .. and in debian its still in uml-utilities
1156698799 Q * mire Quit: Leaving
1156699153 J * s0undt3ch_ ~s0undt3ch@bl8-4-93.dsl.telepac.pt
1156699162 J * balbir ~balbir@59.144.51.152
1156699172 Q * s0undt3ch Read error: Connection reset by peer
1156699175 N * s0undt3ch_ s0undt3ch
1156699686 Q * balbir Ping timeout: 480 seconds
1156700251 J * adm-ybanafa ~admin@82.114.176.197
1156700631 P * adm-ybanafa Leaving
1156700691 Q * michal_ Ping timeout: 480 seconds
1156700761 J * balbir ~balbir@59.144.51.152
1156701064 J * michal_ ~michal@www.rsbac.org
1156703206 Q * Viper0482 Remote host closed the connection
1156703947 N * Belu_zZz Belu
1156703991 Q * ntrs_ Read error: Connection reset by peer
1156704004 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com
1156704595 J * ntrs__ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com
1156704595 Q * ntrs_ Read error: Connection reset by peer
1156704765 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com
1156704765 Q * ntrs__ Read error: Connection reset by peer
1156704833 A * Belu is away (iŽll be back later...)
1156704833 N * Belu Belu_zZz
1156704989 M * Hollow Bertl: can we extend the vx_info syscall with information in /proc/virtual/xid/cvirt ?
1156705260 J * ntrs__ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com
1156705260 Q * ntrs_ Read error: Connection reset by peer
1156705369 M * Bertl Hollow: sure, please keep bugging me until a patch is ther :)
1156705376 M * Hollow will do ;)
1156705377 M * Bertl off for now .. back later
1156705385 N * Bertl Bertl_oO
1156705671 M * cehteh mhm, damnit .. why are things in linux sometimes so complicated
1156705687 M * Hollow it wouldn't be fun otherwise
1156705688 M * Hollow ;)
1156705701 M * cehteh tried to tunnel over tap and bridge, no success
1156705718 M * cehteh (in the root server for now)
1156705761 M * cehteh seems the bridge didnt bridge :P
1156705783 M * Hollow heh, no bridge knowledge in my mind, sorry ;)
1156705793 M * cehteh guess about me
1156706051 J * natas ~what@82.78.3.180
1156706068 Q * natas 
1156706125 M * daniel_hozac Hollow: which information?
1156706150 M * Hollow nr_*, load* and uptime
1156706180 M * daniel_hozac ok.
1156706191 Q * baggins Ping timeout: 480 seconds
1156706779 M * daniel_hozac Hollow: about *.linux-vserver.org, did we already discuss a bug tracker?
1156706787 M * Hollow not really
1156706805 M * Hollow iirc bertl said he does not care as long as bugs are still sent to the ml
1156706821 J * baggins ~baggins@kenny.mimuw.edu.pl
1156706822 M * Hollow maybe a bugzilla list or so would be approriate
1156706866 M * daniel_hozac my memory sucks, so some way to track the issues would be nice, IMHO.
1156706883 M * Hollow yeah, bugzie would be nice
1156707038 M * Hollow daniel_hozac: btw, do you want an account on the svn/wiki/git etc vserver?
1156707066 M * daniel_hozac i.e. shell account? what would it be used for?
1156707111 M * Hollow yep, shel account included, but also valid for svn, git (via ssh), the ftp archives and we could also setup ~userdirs for apache
1156707169 J * robig ~rsoft@83.221.253.82
1156707179 M * robig hello
1156707186 M * Hollow hi robig
1156707230 M * daniel_hozac ok, sure.
1156707233 M * robig can you help me with a little vserver problem, please?
1156707242 M * Hollow daniel_hozac: send me your pubkey please
1156707246 M * daniel_hozac perhaps, but only if you tell us more about it.
1156707279 M * robig when i tried to start a created vserver the following message was displayed:
1156707280 M * robig The following problem(s) were encountered while verifying vshelper
1156707280 M * robig functionality:
1156707280 M * robig * The vshelper state-directory '/var/run/vshelper' does not exist;
1156707299 M * robig ive created this direcory manually
1156707306 Q * baggins Ping timeout: 480 seconds
1156707372 M * robig and also the /var/run/vservers directory
1156707383 M * daniel_hozac Hollow: http://daniel.hozac.com/tmp/vserver_id.pub
1156707402 M * Hollow username?
1156707436 M * daniel_hozac dhozac
1156707449 M * daniel_hozac robig: and what happens then?
1156707458 M * daniel_hozac robig: what distro are you on?
1156707461 M * robig ah.. ive to create /var/run/vservers.rev too
1156707649 M * Hollow daniel_hozac: ok, try to login, helios.dev.croup.de (85.10.237.61)
1156707650 M * waldi hmm, can I allow a vserver to do remount,rw?
1156707661 M * daniel_hozac waldi: secure_remount?
1156707676 M * waldi hmm
1156707690 M * daniel_hozac (ccap)
1156707697 M * waldi yep, works
1156707718 M * daniel_hozac Hollow: works fine.
1156707723 M * Hollow good
1156707739 M * Hollow please change your password
1156707740 A * waldi needs to finaly get the no-nodev-bindmount patch up and running
1156707759 M * waldi i need this for this machine ...
1156707886 M * daniel_hozac Hollow: old password? empty?
1156707903 M * Hollow query
1156708059 M * daniel_hozac done.
1156708070 M * Hollow okay, fine
1156708100 M * Hollow any suggestion for devspace? dev.l-v.org?
1156708166 M * daniel_hozac devspace? like ~dhozac?
1156708170 M * Hollow yup
1156708200 M * daniel_hozac maybe users would be more correct?
1156708257 M * Hollow well, being pedantic i'd say we aren't users.. ;)
1156708262 M * Hollow but maybe people
1156708265 M * Hollow like on kernel.org
1156708267 M * daniel_hozac yeah, that works.
1156708279 M * Hollow ok
1156709320 M * Hollow Bertl_oO: please add people.linux-vserver.org with 85.10.237.61 to DNS
1156709824 A * waldi have to redo that patch ...
1156709894 M * derjohn Hollow, the .html files to fetches for the wiki-transition: are they updated? Of should I always look on the src=
1156709915 M * Hollow you mean the sorted htmls on home.xnull.de? not updated..
1156709982 M * derjohn Hollow, k. as soon as I "transctibe" a page, I will a a link to the new wiki on the old wiki page and a warning "please do not update!"
1156710023 M * daniel_hozac hmm, i thought we had a rewrite map that showed the new page for pages that had a new version?
1156710039 M * daniel_hozac though i suppose that won't work for pages where we change the name, which i think we should do for most of them...
1156710074 M * derjohn daniel_hozac, I think the map will be actived if all content ins transferred?
1156710076 M * Hollow daniel_hozac: http://home.xnull.de/work/vserver/wikipages.txt
1156710082 M * derjohn daniel_hozac, I wnat to do it page by page ---
1156710083 M * Hollow this is the rewrite map
1156710127 M * daniel_hozac i think we really need to standardize the naming.
1156710141 M * Hollow derjohn: the map will be activated as long as not all content is transferred, and a some time after that so google links will still work
1156710147 M * Hollow daniel_hozac: indeed.
1156710152 M * daniel_hozac like spaces...
1156710185 M * Hollow page names like these are as good as no name: http://wiki.linux-vserver.org/Linux-VServer-Hacks
1156710216 M * derjohn Hollow, so I simply delete the page on the oldwiki and the rest is done by the map?
1156710236 M * daniel_hozac no, the old wiki is still the one pointed to by linux-vserver.org.
1156710255 M * Hollow derjohn: no, just leave the page as-is, we will add a notice on all pages in the old wiki, but let it read-write for some time, then read-only, then remove
1156710266 M * Hollow daniel_hozac: well, yes and no
1156710274 M * Hollow linux-vserver.org will get the redirect map
1156710283 M * daniel_hozac but it's not active right now.
1156710293 M * Hollow i.e. new pages are redirected to wiki. and not migrated pages will be redirected to oldwiki.
1156710297 M * derjohn Hollow, "we will add" means I ping you as soon as i finish some page?
1156710309 M * daniel_hozac ok.
1156710311 M * Hollow no, as soon as the new wiki goes official
1156710315 M * Hollow ^ derjohn
1156710363 M * Hollow the notice will be on all pages
1156710365 M * derjohn Hollow, so my part is simply adding the note to oldwiki page.
1156710397 M * Hollow no, as i said, we will add a global notice to all pages once the new wiki is official
1156710414 M * Hollow to all old pages
1156710450 M * derjohn sure and in between? until we read the point that will be weeks. and it would be sad to lose changes users make in this time.
1156710470 M * derjohn *reach
1156710547 M * Hollow well, the new wiki will be official as soon as 2.0.2 is out
1156710562 M * derjohn within 3 days ? c00l ;) then i'll hurry up
1156710602 M * Hollow derjohn: yeah, it will encourage users to add new pages to the new wiki
1156710613 M * Hollow :)
1156710623 M * derjohn Hollow, I punt a hint on the "News" of the oldwiki that there will be new one soon
1156710643 M * Hollow please don't add a link for now
1156710649 M * derjohn why?
1156710690 Q * |gerrit| Remote host closed the connection
1156710700 M * Hollow because i will do an announcement once 2.0.2 + wiki is out
1156710770 M * derjohn Hollow, and what about a general hint: ||{B,r}  27 August 2006:|| We made plans for a new wiki. Stay tuned. ||
1156710795 M * daniel_hozac the plans have been around for months though ;)
1156710796 M * derjohn it's yout project and I wont disturb your plans.
1156710818 M * derjohn ||{B,r}  for month now || We made plans for a new wiki. Stay tuned. ||
1156710835 M * Hollow well, as i said, i will do an announcement...
1156710846 Q * dna Quit: Verlassend
1156711027 Q * phedny Ping timeout: 480 seconds
1156711115 Q * bonbons Quit: Leaving
1156711189 M * waldi Bertl_oO: recursive bind does only apply new flags to the top mount
1156711197 Q * yarihm Quit: This computer has gone to sleep
1156711321 M * derjohn Bertl_oO, the machine is now @ 256MMb and 128 swap. I try to reproduce the "no killable processes" phenomenon.
1156712187 Q * nayco Quit: Bonne nuit !
1156712210 A * waldi likes real machines:
1156712211 M * waldi              total       used       free     shared    buffers     cached
1156712211 M * waldi Mem:      33257432     824064   32433368          0     114128     501552
1156712323 M * Wonka nice
1156712339 M * Wonka not really needed, though
1156712395 M * derjohn waldi, hey, that's one digit more than I expected :)
1156712410 M * waldi derjohn: yep
1156712444 M * waldi derjohn: http://bblank.thinkmo.de/blog/archive/2006/08/26/new-machine-to-play-with
1156712488 M * waldi -bash: fork: Resource temporarily unavailable 
1156712490 M * waldi *gna*
1156712525 M * Wonka ah yes, had one of those in my fingers too
1156712537 M * Wonka no qemu for it, though
1156712542 M * waldi qemu?
1156712549 M * waldi what do you want to do?
1156712558 M * Wonka else i would've run some BOINC client on it
1156712562 M * derjohn waldi, 32 CPU? 16 HT? or real?
1156712572 M * waldi 8 real
1156712573 M * Wonka derjohn: 8 cores, 4 threads each
1156712586 M * derjohn ah, so it was the wrong question ;)
1156712629 M * derjohn waldi, about the fork: did you limit AS/RSS ?
1156712647 M * waldi nope
1156712678 M * derjohn did you overflow the process table ?
1156712707 M * waldi possible
1156712733 M * waldi the machine was rather dead since days
1156712758 M * derjohn that was the 32-hearts machine ?
1156712761 M * waldi no
1156712770 M * waldi my 6 hearts zseries
1156712823 M * derjohn ah, funny, because I had similar probs with AS/RSS limit. the guest's VM got exhausted in a way that to fork or anything was posbbile.
1156712839 M * waldi this was the context 0
1156712858 M * waldi lets see if low memory does not break any longer
1156712863 M * derjohn waldi, so the guets ate up all your resources? time to rlimtit ;)
1156712926 M * waldi the last time i tried that, the context locks up hard if it reached the limit
1156712961 M * waldi i finaly redo the nodev-bind patch
1156712962 M * waldi http://lophos.multibuild.org/linux/vserver-bind-dev.patch
1156712972 M * derjohn hard lock? In a way you had to reboot the context ?
1156713003 M * waldi no, i had to reboot the machine
1156713016 M * waldi it had many processes in D and Z state
1156713034 M * waldi hmm, does secure_remount imply secure_mount?
1156713046 M * derjohn from within guests which were out-of-everything i assume?
1156713056 M * waldi yes
1156713085 M * daniel_hozac waldi: no.
1156713110 M * waldi daniel_hozac: what happens if i have secure_remount and do mount -o remount,dev?
1156713151 M * waldi the check which sets nodev is only active for secure_mount
1156713257 M * daniel_hozac yep.
1156713260 M * daniel_hozac you're right.
1156713266 M * waldi problem?
1156713290 M * daniel_hozac i'd say so.
1156713298 M * waldi okay, have to test that
1156713331 M * daniel_hozac it works for me (i.e. remount,dev lets me access devices) on 2.1.1-rc31.
1156713338 M * waldi not good
1156713346 M * daniel_hozac right.
1156713370 M * waldi will try to get a cve, this needs to be fixed
1156713488 M * daniel_hozac we should probably wait for someone else to confirm it though.
1156713496 M * waldi take a look at my patch, it should fix this problem and at the same time allow dev bindmounts
1156713518 M * daniel_hozac do_remount should use SECURE_REMOUNT, no?
1156713527 M * waldi yes
1156713533 M * waldi will test it
1156713543 M * daniel_hozac ok, you've updated the patch since i first viewed it :)
1156713548 M * waldi yes
1156713603 Q * id23 Remote host closed the connection
1156713689 M * daniel_hozac the patch makes sense to me.
1156713712 M * waldi hmm, does not work for me currently
1156713718 M * waldi ah, it have secure_mount
1156713757 M * waldi yes, works
1156713849 M * waldi can you please test the patch?
1156713871 M * waldi remount should allow to set nodev but not to clear
1156713893 M * waldi the ssame for bind
1156713970 M * daniel_hozac rebuilding now.
1156713980 M * waldi bind mount does not check for secure_mount?
1156714073 M * daniel_hozac hmm, appears not...
1156714100 M * daniel_hozac but i get permission denied for it.
1156714104 M * robig hm... how can i mount the hosts cdrom in a vserver?
1156714107 M * daniel_hozac so it has to be blocked somewhere.
1156714126 M * waldi ah, via mount_is_safe
1156714158 M * daniel_hozac ah.
1156714186 M * derjohn robig, do it from the host
1156714212 M * waldi so this is okay
1156714224 M * derjohn robig, and maybe use a bind --mount /media/cdrom /var/lib/vservers/media/cdrom or sucgh
1156714264 M * waldi daniel_hozac: okay, for now, secure_remount without secure_mount is a bad idea ...
1156714273 M * daniel_hozac yep.
1156714288 M * daniel_hozac but i guess they're usually used together.
1156714302 M * waldi yep
1156714308 M * daniel_hozac rebooting to the new kernel.
1156714324 M * robig works also on already running vs?
1156714354 M * daniel_hozac robig: you'll have to do it in the namespace then, with vnamespace -e <guest> ...
1156714379 M * robig i can reboot the vserver.. wouldnt be a problem
1156714412 M * daniel_hozac waldi: patch works for at least the previous test-case, it keeps the mount nodev.
1156714436 M * daniel_hozac (though silently... would be nice to fail but i guess we don't have sufficient info available)
1156714448 M * waldi this is also possible
1156714457 M * waldi should I change that?
1156714475 M * derjohn daniel_hozac, waldi: do i get it right that this smells like a rc32 soon?
1156714481 M * waldi yes
1156714490 M * waldi and another CVE entry
1156714512 M * daniel_hozac IMHO yes, specifying ,dev and then having the mount turn out nodev anyway is weird.
1156714531 M * waldi daniel_hozac: okay, will bail out
1156714535 M * derjohn waldi, debian sec will love it ;)
1156714543 M * waldi yes
1156714553 M * waldi i just requested the CVE from them
1156714614 M * derjohn i think i'll wait with my rc31 rollout ;)
1156714799 M * waldi daniel_hozac: please reload
1156714953 M * waldi this also changes the behaviour for normal mount, bail out if it don't set nodev
1156714965 M * waldi which is a drastic change to the current behaviour
1156715055 Q * meandtheshell Quit: bye bye ... 
1156715197 M * daniel_hozac is that really what we want?
1156715288 M * daniel_hozac will NODEV be in the mnt_flags if ,nodev wasn't explicitly?
1156715290 M * daniel_hozac +set
1156715339 M * waldi no
1156715423 M * waldi so i don't think we want that now
1156715424 M * derjohn Bertl_oO, I was successfull: rsync-ing the whole host killed the machine again 
1156715439 M * waldi maybe change it for 2.1
1156715457 M * waldi but not for 2.0
1156715476 M * daniel_hozac do we want that behavior in 2.1?
1156715500 M * daniel_hozac seems rather unexpected to get -EPERM just because you didn't specify -o remount,nodev.
1156715517 M * waldi mount copies the old flags for remount
1156715524 M * waldi to nodev is set if it was before
1156715533 M * daniel_hozac ok.
1156715540 M * waldi only for bind and normal mounts, it changes
1156715582 M * waldi anyway, I have both patches
1156715862 M * waldi oh, oh
1156715866 M * waldi one error
1156715914 M * waldi check_mnt should be before i test the flags
1156716116 M * waldi okay, final version: http://lophos.multibuild.org/linux/vserver-mount.patch
1156716194 J * baggins baggins@kenny.mimuw.edu.pl
1156716652 M * daniel_hozac any reason we can't just fail on remount nodev->dev?
1156716753 M * waldi the same reason than bind and normal mount
1156716763 M * daniel_hozac how so?
1156716768 M * waldi something which just works, fails no
1156717414 Q * baggins Ping timeout: 480 seconds
1156717453 M * robig i've a locales problem in all my vservers... 
1156717463 M * derjohn robig, debian?
1156717480 M * robig debian like ;)
1156717498 M * derjohn apt-get install locales or  dpkg-reconfigure locales ?
1156717508 M * derjohn debootstrap doesnt by default
1156717517 M * robig have done already
1156717526 M * robig does noting
1156717530 M * derjohn details about the prob?
1156717549 M * robig also using localeconf
1156717589 M * derjohn hm, why? maybe you borke wth there? (i never used it, so i dont know)
1156717614 M * robig locale -a only shows:
1156717630 M * robig locale: Cannot set LC_CTYPE to default locale: No such file or directory
1156717634 M * robig ...
1156717694 M * derjohn robig, are you sure you did a dpkg-reconfigire loacles? maybe with "-plow"?
1156717708 M * derjohn robig, debian or ubuntu 
1156717709 M * robig what does plow?
1156717720 M * derjohn debconf prio low = plow 
1156717724 M * robig ubuntu :)
1156717729 M * derjohn edgy ?
1156717735 M * derjohn then I know the problem ;)
1156717748 M * robig dapper
1156717764 M * cehteh is it possible to migrate a runing process from one context to another?
1156717770 M * derjohn hm, does the dialog come up when you do a dpkg-reconfigire locales ?
1156717795 N * ntrs__ ntrs
1156717818 M * ntrs Is anyone using devel rc29 or rc31 in production with multiple guests?
1156717819 M * derjohn cehteh, I doubt that, as it would have to change the namespace
1156717832 M * daniel_hozac ntrs: running it on my test box, why?
1156717836 M * robig derjohn: nope
1156717838 M * derjohn cehteh, (not sure if everything is 'on place'
1156717839 M * cehteh .. or rather clone it/create a new context
1156717854 M * ntrs I am just wondering about the stability on a production system with like 20-30 guests.
1156717872 M * derjohn robig, there must come up a curses based dialog where you can choose locales. I had the phenomenon on edgy ...
1156717876 M * cehteh moving into anoter existing context is likely impossible by design
1156717890 M * daniel_hozac cehteh: you can't migrate from a context.
1156717910 M * daniel_hozac cehteh: you won't even get passed the basic security checks in sys_vserver.
1156717912 M * robig the locales -a should also list all available locales, eh?
1156717915 M * derjohn I apt-ed around loacles and all suggests and recommeneds of the locales package and friends and then it vanished
1156717933 M * robig could this be a kernel misconfiguration?
1156717936 M * derjohn robig, cannot help you much there, but I could provide a naked dapper template.
1156717946 M * derjohn robig, no
1156717975 M * robig template?
1156717976 M * daniel_hozac s/passed/past/, sigh.
1156717983 M * derjohn robig, apt-get install --reinstall locales ? or apt-get remove --purge locales before?
1156718008 Q * FireEgl Read error: Connection reset by peer
1156718010 M * derjohn robig, yes a "working" dapper you just copy in /var/lib/vservers/newdapper
1156718016 M * robig the problem occures on all vservers
1156718025 M * derjohn robig, all?
1156718033 M * robig ive also an old one, which runs nice on my old server
1156718045 M * robig all 3 ;)
1156718092 M * robig 2 new debootstrapped, and a old copy of my running server
1156718293 M * derjohn robig, well, with my dapper it worked. I bootstr... *gruebel* ah, I did an breezy bootstrap and dist-upgraded.
1156718313 M * derjohn so the phenomenen may be in dapper and in edgy.-
1156718354 M * robig what do you mean with template?
1156718378 M * robig my old vserver was an debian sarge system
1156718385 M * robig (debootstrap)
1156718416 M * robig maybe something wrong in the host system?
1156718500 M * derjohn robig, I think you host is fine. just the guest didnt correctly configure the locales. locale-gen UTF-8 or whatever
1156718523 M * derjohn I really cannot remember what I did to solve the problem. sry :/
1156718568 M * robig :(
1156718596 J * FireEgl FireEgl@Sebastian.Atlantica.US
1156719053 Q * robig Ping timeout: 480 seconds
1156719067 J * baggins baggins@kenny.mimuw.edu.pl
1156719379 M * ntrs So, I guess no real production use of devel yet.
1156719453 M * derjohn ntrs, I use it in production 
1156719476 M * ntrs derjohn, any issues? How many guests on average?
1156719485 M * derjohn up to 10-12 guests-
1156719532 M * ntrs derjohn, no problems?
1156719560 M * derjohn ntrs, no problems that are unique to devel.
1156719571 M * derjohn why do you think so ?
1156719573 M * ntrs derjohn, should I go up to rc31 or is rc29 good enough?
1156719599 M * ntrs What non-unique problems are you referring to?
1156719607 M * derjohn ntrs, I just compiled rc31, but rc32 seems to be very close, so I wait till rollout :)
1156719635 Q * baggins Ping timeout: 480 seconds
1156719636 M * derjohn ntrs, currently: setting 'good' rlimits
1156719656 M * ntrs good rlimits?
1156719662 M * derjohn ntrs, hunting down a "no killable processes" bug (Or hunting my own idioty ;))
1156719674 M * derjohn ntrs, you know rlimits?
1156719692 M * ntrs yes, I know rlimits, but I didn't understand what are 'good' rlimits.
1156719731 M * derjohn ntrs, those which keep a guest alive and guarantee enough resources for the other guests.
1156719740 M * ntrs I see.
1156719762 M * derjohn ntrs, i.e. I set AS about 2*RSS, which is not enough to e.g. compile larger stuff inside a guest.
1156719791 M * ntrs I only set rss, never bothered with AS
1156719795 M * derjohn ntrs, danirl recommened a 4:1 ratio ... I do an 'empirical' analysis now :)
1156719838 M * derjohn ntrs, i wanted to prevent that one guest uses up all swap.
1156719883 M * ntrs how could one guest use up all the swap if it's rss is limited, he could only use up to the rss limit, right?
1156720023 M * daniel_hozac swap should not count against the RSS limit.
1156720032 M * derjohn ntrs, hm, RSS is resident , i.e. thats what in physical RAM is. 
1156720643 Q * coocoon Quit: KVIrc 3.2.0 'Realia'