1320797462 M * dkg Bertl: why would "naddress --remove --nid 111 --ip 192.168.13.8" fail?  the address is listed (3 times, somehow) in /proc/virtnet/111/info ?
1320797469 M * dkg naddress: vc_net_remove(): Invalid argument
1320798312 M * Bertl because it isn't implemented (yet)
1320798324 M * Bertl i.e. you can only remove all IPs not a single one
1320798349 M * dkg aha :)
1320798360 M * Bertl but it's safe to remove all of them and add one back
1320798467 Q * tty234 Remote host closed the connection
1320798533 M * dkg hm.  i did "naddress --remove --nid 111 && naddress --add --nid 111 --ip $oldaddr/24"
1320798554 M * dkg and now $oldaddr is listed more than once in /proc/virtnet/111/info
1320798568 M * dkg maybe --remove didn't work at all?
1320798602 M * dkg what should "naddress --set" do?
1320799166 M * Bertl from the kernel side, there is add and remove (in several versions)
1320799250 M * Bertl remove currently only implements NXA_TYPE_ANY
1320799259 M * Bertl i.e. remove all assigned addresses
1320799308 M * Bertl --set is a userspace 'macro' i.e. combination of remove all and add given to simplify changes
1320799345 M * Bertl of course, you can do those steps via remove and add as well
1320799368 M * dkg i see
1320799430 M * Bertl but I'm not sure you can specify NXA_TYPE_ANY via --remove (or at least it might depend on the util-vserver version)
1320799433 M * dkg so --set works, but i'm stillnot sure why "naddress --remove --nid 111" didn't remove them all
1320799440 M * dkg ah, ok
1320799469 M * Bertl you have to check the source (util-vserver) or ask daniel_hozac if there is a way to do the 'remove all' part
1320799481 M * dkg no need for me, --set does what i wanted.
1320799691 M * Bertl perfect then
1320800258 Q * dowdle Remote host closed the connection
1320804153 M * Bertl off to bed now ... have a good one everyone!
1320804160 N * Bertl Bertl_zZ
1320807069 Q * hparker Quit: Quit
1320808080 J * fisted_ ~fisted@xdsl-87-78-212-90.netcologne.de
1320808159 Q * fisted Ping timeout: 480 seconds
1320808888 Q * clopez Ping timeout: 480 seconds
1320817543 J * sannes1 ~ace@cm-84.209.106.118.getinternet.no
1320821621 J * jrklein ~osx@2001:470:1f0f:572::250:160
1320822024 J * ncopa ~ncopa@3.203.202.84.customer.cdi.no
1320823922 J * derjohn_mob ~aj@213.238.45.2
1320827615 Q * cehteh Quit: Coyote finally caught me
1320827898 J * cehteh ~ct@pipapo.org
1320828637 J * thierryp ~thierry@home.parmentelat.net
1320829079 J * kir ~kir@swsoft-msk-nat.sw.ru
1320829152 J * ghislain ~AQUEOS@adsl2.aqueos.com
1320829687 Q * thierryp Remote host closed the connection
1320831129 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com
1320833518 J * clopez ~clopez@155.99.117.91.static.mundo-r.com
1320833690 N * Bertl_zZ Bertl
1320833694 M * Bertl morning folks!
1320835856 T * * http://linux-vserver.org/ |stable 2.2.0.7, exp 2.3.0.36.38, grsec 2.3.0.36.28|util-vserver-0.30.216-pre2914| He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the Wiki, and we forget about the minute.
1320835856 T * ChanServ -
1320836203 J * SadMan sadman@sadman.net
1320836228 J * m_ueberall dircproxy4@subjektzentrisch.de
1320836239 J * jeroen__ ~jeroen@095-097-051-172.static.chello.nl
1320836835 J * thierryp ~thierry@home.parmentelat.net
1320838517 N * sid4windr sid3windr
1320840973 M * BenG hi vserverers
1320840991 M * BenG I'm looking at doing some rate limiting for my virts
1320841042 M * BenG I've played with this before, tagging packets from source vserver guest and putting them into particular queue through tc
1320841063 M * BenG I'm wondering if anyone else has done this, and if they have any documentation
1320841262 M * Bertl should be straight forward and the lartc howto should contain everything you need: http://lartc.org/howto/lartc.cookbook.html
1320841295 M * Bertl http://lartc.org/howto/index.html (here is the index)
1320841330 M * BenG it's a complex doc, and I've read it numerous times
1320841344 M * BenG was hoping someone had already done something vserver specific
1320841355 M * BenG I'll document on the wiki when I'm done anyhow
1320841363 M * Bertl http://lartc.org/howto/lartc.ratelimit.single.html
1320841373 M * Bertl that's probably the closest example
1320841402 M * Bertl IIRC, OVZ has a wiki page for that as well, so you might look there too
1320841456 M * BenG ah great, I will
1320841459 M * BenG cheers
1320841536 M * BenG http://wiki.openvz.org/Traffic_shaping_with_tc
1320841626 M * BenG I'll do this with cgroups' net_cls.classid though, for a more generic solution
1320841638 M * BenG I want per guest (and host) not per IP address
1320841679 M * Bertl okay, just keep in mind that lightweight (network isolated) guests will not have a network cgroup
1320841701 M * Bertl (i.e. better mention that and maybe explain how to do it with IPs as well on the wiki)
1320841728 M * BenG "lightweight (network isolated)" what are those?
1320841738 M * Bertl the default used in Linux-VServer 
1320841759 M * Bertl i.e. guests with network context but not network namespace
1320841766 M * BenG ah right
1320841773 M * BenG but they will have net_cls.classid though
1320841789 M * BenG and so can be tagged based on their cgroup can't they?
1320841841 M * BenG or will net_cls.classid value simply not effect any packets on guests without a network namespace?
1320841867 M * Bertl not sure, I haven't checked how that is implemented and on what cgroup it is based
1320841882 M * Bertl but should be simple to test
1320841919 M * BenG yep, my first test should be to see if any packets are tagged based on net_cls.classid
1320841938 M * BenG which is basically half the problems solved
1320842404 Q * thierryp Remote host closed the connection
1320842406 J * wurtel__ ~paul@gw-office.telegraaf.net
1320842988 Q * clopez Quit: Leaving
1320843596 J * thierryp ~thierry@home.parmentelat.net
1320850625 N * daniel_hozac_ daniel_hozac
1320851673 Q * kir Quit: Leaving.
1320851682 J * kir ~kir@swsoft-msk-nat.sw.ru
1320852097 M * Bertl off for now ... bbl
1320852102 N * Bertl Bertl_oO
1320852173 Q * fisted_ Ping timeout: 480 seconds
1320852310 J * fisted ~fisted@xdsl-87-78-218-105.netcologne.de
1320853126 N * zbyniu_ zbyniu
1320853132 Q * BenG Quit: I Leave
1320853474 J * mcp ~mcp@wolk-project.de
1320853504 J * dowdle ~dowdle@scott.coe.montana.edu
1320854916 Q * ncopa Quit: Leaving
1320855549 J * uranus_ ~uranus@p57B4F1E3.dip.t-dialin.net
1320857365 M * uranus_ is there any reason why the patch for kernel 3.1 has the rc tag?
1320858450 P * kir Leaving.
1320858757 J * bonbons ~bonbons@2001:960:7ab:0:d97c:78d4:5eac:24b2
1320859645 M * WMP is possible to limiten in vserver cpu time?
1320860185 N * Bertl_oO Bertl
1320860217 M * Bertl uranus_: because it isn't complete/tested yet
1320860232 M * Bertl WMP: yes, but usually that's not what you want
1320860269 M * WMP Bertl: i have vps on my server, and few vps use too much cpu, and i want to limit this
1320860293 M * Bertl yeah, but what you want to limit is the cpu usage, not the cpu time
1320860318 M * WMP yhy, and this is possible?
1320860329 M * WMP limit to one thread on ps?
1320860331 M * WMP vps?
1320860338 M * Bertl CPU time is the actual 'time' a process gets
1320860353 M * Bertl so, for example 5 seconds, then it gets killed
1320860369 M * WMP yhy, ok ;)
1320860374 M * WMP so i want to limit cpu usage
1320860380 M * Bertl and yes, both, cpu 'time' and cpu usage can be limited with recent and older kernels
1320860400 M * Bertl on recent kernels it is done via the cgroup subsystem
1320860709 M * WMP Bertl: and on 2.6.38?
1320860776 M * Bertl 2.6.38(.8) is recent
1320860790 M * arekm oh, someone else having the same problem as I with cgroups http://serverfault.com/questions/211509/how-to-stop-apache-from-crashing-my-entire-server
1320860972 M * Bertl yeah, so maybe file a bug report on LKML, might lead to somebody looking into the cgroup oom handling :)
1320862463 Q * derjohn_mob Ping timeout: 480 seconds
1320862511 Q * thierryp Remote host closed the connection
1320868750 J * hijacker_ ~hijacker@cable-84-43-136-96.mnet.bg
1320870846 J * thierryp ~thierry@home.parmentelat.net
1320872054 Q * hijacker Read error: Connection reset by peer
1320872151 J * hijacker ~hijacker@213.91.163.5
1320872380 M * arekm Bertl: http://lists.linuxfoundation.org/pipermail/containers/2011-November/028382.html
1320873172 Q * sannes1 Remote host closed the connection
1320873329 M * Bertl tx
1320873636 Q * thierryp Remote host closed the connection
1320874652 M * daniel_hozac arekm: we've had similar reports before too.
1320874660 M * daniel_hozac i don't remember who it was though.
1320875299 M * arekm "net: Add network priority cgroup" - soon there will be cgroup for everything
1320875438 M * Bertl cgroups are in nowadays :)
1320875448 M * Bertl s/in/'in'/
1320875491 M * Bertl nobody cares if it works or not, as long as there is a cgroup for it :)
1320875534 M * arekm Nov  9 20:48:53 tm2 kernel: [18300.349110] memory: usage 5222400kB, limit 5222400kB, failcnt 282869
1320875537 M * arekm Nov  9 20:48:53 tm2 kernel: [18300.349113] memory+swap: usage 0kB, limit 9007199254740991kB, failcnt 0
1320875543 M * arekm that value looks soo cool heh.
1320875579 M * arekm (hoping that this isn't vserver related bug ;P)
1320875609 M * Bertl nah, that's vanilla and strictly speaking it's not a bug
1320875670 M * arekm but I have no swap here
1320875696 Q * bonbons Quit: Leaving
1320875708 M * arekm and even if I did I would expect limit to be size of mem+swap by default and not some huge number
1320875718 M * Bertl first, it's swap+mem, and second, the huge number is the result from using -1/-2 for special no-limit
1320875729 M * arekm ahh
1320875765 M * Bertl the interesting part (which I consider a bug and which we worked around recently) is the usage = 0
1320875781 M * Bertl because strictly speaking it should be the same as the memory usage
1320875801 M * Bertl (remember the negative/huge values in meminfo)
1320875827 M * arekm but is this vserver introduced bug just like in meminfo or vanilla crazynes?
1320875839 M * Bertl that message is vanilla
1320875895 M * Bertl at least regular Linux-VServer messages should be properly prefixed
1320875913 M * Bertl and I don't remember having similar debug output
1320875926 M * Bertl (which, of course, might have slipped through)
1320875939 M * arekm well, message could be vanilla but calculacions could be vserver-affected
1320875961 M * Bertl nah, we do not modify anything cgroup related
1320875975 M * Bertl too broken to mess with it atm :)
1320876144 M * Bertl in a few years, cgroups will be stable and completely useable ... I hope
1320876197 M * arekm and vserver won't exist? ;)
1320876213 M * daniel_hozac ideally...
1320877514 J * Aiken ~Aiken@2001:44b8:2168:1000:21f:d0ff:fed6:d63f
1320878563 M * Bertl off to bed now .. have a good one everyone!
1320878569 N * Bertl Bertl_zZ
1320879090 Q * hijacker_ Quit: Leaving
1320880473 Q * ghislain Quit: Leaving.