1297469321 Q * imcsk8 Quit: Leaving
1297470168 Q * bonbons Quit: Leaving
1297471021 Q * yarihm Ping timeout: 480 seconds
1297472199 Q * dowdle Remote host closed the connection
1297486772 Q * hparker Quit: Quit
1297488920 Q * manana Remote host closed the connection
1297489355 Q * Piet Ping timeout: 480 seconds
1297489949 J * Piet ~Piet__@82VAABWS6.tor-irc.dnsbl.oftc.net
1297491945 M * Bertl off to bed now ... have a good one everyone!
1297491949 N * Bertl Bertl_zZ
1297495650 Q * padde Remote host closed the connection
1297495766 J * padde ~padde@patrick-nagel.net
1297498851 J * hijacker_ ~hijacker@87-126-142-51.btc-net.bg
1297502005 J * petzsch ~markus@dslb-092-078-229-149.pools.arcor-ip.net
1297502842 Q * hijacker_ Quit: Leaving
1297503010 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d
1297505184 J * manana ~mayday090@84.17.25.149
1297511675 Q * ntrs Remote host closed the connection
1297512871 J * ntrs ~ntrs@vault08.rosehosting.com
1297514943 Q * ntrs Remote host closed the connection
1297514996 J * ntrs ~ntrs@vault08.rosehosting.com
1297515199 Q * ntrs Read error: Operation timed out
1297515826 J * ntrs ~ntrs@209.135.140.188
1297516268 Q * ntrs Remote host closed the connection
1297516734 J * ntrs ~ntrs@vault08.rosehosting.com
1297517039 Q * ntrs Remote host closed the connection
1297517881 J * ntrs ~ntrs@vault08.rosehosting.com
1297520207 Q * rickytato Quit: rickytato
1297522788 Q * eyck Ping timeout: 480 seconds
1297523025 J * eyck ~eyck@77.79.198.66
1297523243 J * hparker ~hparker@2001:470:1f0f:32c:beae:c5ff:fe01:b647
1297523681 N * ensc Guest1165
1297523691 J * ensc ~irc-ensc@p5DF2D44E.dip.t-dialin.net
1297524087 Q * Guest1165 Ping timeout: 480 seconds
1297527310 N * Bertl_zZ Bertl
1297527313 M * Bertl morning folks!
1297527353 M * hparker Morning Bertl, how goes?
1297528059 M * Bertl well, I have a kswapd0 running with 95% cpu load on one of my servers, the funny part is, it has no swap enabled at all
1297528076 M * daniel_hozac lol
1297528105 M * Bertl but google tells me that might be related to transparent hugepage support
1297528107 M * daniel_hozac i'm debugging a system that is showing 1 GiB of memory usage, in spite of having essentially nothing but snmpd and a lighttpd instance running on it...
1297528125 M * daniel_hozac ah
1297528152 M * Bertl RSS usage?
1297528209 M * daniel_hozac totals 62 MiB across all processes.
1297528241 M * Bertl so what shows the 1 GiB memory usage?
1297528250 M * daniel_hozac slabtop shows 408 MiB used for radix_tree_nodes
1297528287 M * daniel_hozac free shows 1146 MiB of not-cached, not-buffered memory.
1297528362 M * Bertl the RTNs are usually for the slab allocator, 64bit system?
1297528368 M * daniel_hozac yeah.
1297528512 M * daniel_hozac ~257000 slabs used.
1297528515 M * Bertl many small files which are served by lighttp?
1297528535 M * daniel_hozac mostly large ones, it's a Fedora/CentOS mirror.
1297528542 M * daniel_hozac but there is very little to no traffic on it.
1297528567 M * hparker lighty caching them?
1297528574 M * Bertl try to flush the cache and check if the memory usage goes down
1297528587 M * Bertl (the slab/RTN part)
1297528628 M * daniel_hozac went down about 40 MiB, but it's still at 360.
1297528650 M * daniel_hozac and it still has 360 MiB of ext3_inode_caches...
1297528704 M * daniel_hozac i will likely just upgrade the kernel (it's running an older 2.6.32) and seeing if it goes away.
1297528709 N * ensc Guest1176
1297528719 J * ensc ~irc-ensc@p5DF2D44E.dip.t-dialin.net
1297528771 M * Bertl daniel_hozac: or upgrade :)
1297528786 M * Bertl s/upgrade/downgrade/
1297528821 M * daniel_hozac i don't think i've ever heard you suggest a downgrade before :)
1297528829 M * daniel_hozac what's your preferred kernel these days?
1297528870 M * Bertl I'm still running 2.6.22 on some machines, but most of them are currently trailing mainline so something between 2.6.36.4 and 2.6.38-rc5
1297528880 Q * Guest1176 Ping timeout: 480 seconds
1297528886 M * daniel_hozac i'm still running mostly 2.6.22 too...
1297528916 M * Bertl 2.6.38 seems to be a good choice at least for servers (putting the kswapd issue aside, which is new and might be fixed already)
1297528935 M * daniel_hozac okay. how come?
1297529111 M * Bertl the only issues I saw on a heavily loaded server with 2.6.38-rc3 is an nfs error message flooding the syslog (when set to log info level)
1297529138 M * Bertl i.e. raid over lvm with encryption works perfectly fine and quite performant
1297529170 M * Bertl no problems with OOM killings anymore, no problems with I/O in general
1297529229 M * Bertl and the overall latency is quite low as well
1297529264 M * Bertl i.e. running 3-4 rsyncs, moving 70M/s each and you can ssh to the machine as if it was idle
1297529309 J * yarihm ~yarihm@178-83-213-36.dynamic.hispeed.ch
1297529414 M * hparker So .38 has some new toys worth looking at?
1297529512 M * Bertl a bunch of new features as well, but mostly the old ones which were broken since 2.6.24 seem to work reasonably well there
1297529544 M * hparker Oh, nice
1297529555 M * Bertl it was already better with 2.6.36, but 2.6.38 seems to become a good release
1297529584 M * Bertl 2.6.37 is very similar to 2.6.36 in the way it behaves, so not really bad, but not really great either
1297529624 M * hparker Running .37 on my desktop atm, worth playing with .38?
1297529680 M * Bertl haven't tried it on a desktop yet, I usually do not reboot my desktop machines unless absolutely necessary :)
1297529758 M * Bertl (mostly because I'm lazy and have hundreds of tabs in many windows open and a reboot would require to open/arrange them again
1297529762 M * Bertl :)
1297529848 M * Bertl and for the kswapd issue, it seems that kswapd is also used to reclaim pages, which might be related to the rsyncs I'm running there, so it might be perfectly fine that kswapd uses up a whole cpu
1297529869 M * arekm Bertl: is MKNOD bcap enough for mknod to work in guest?
1297529882 M * hparker Migrating vserver guests to 2.6.35-vs2.3.0.36.32 from whatever is latest .22 patch.. I have a old RH server that won't start.. Complaining about libc.. Ideas?
1297529992 Q * yarihm Quit: This computer has gone to sleep
1297530015 J * yarihm ~yarihm@178-83-213-36.dynamic.hispeed.ch
1297530046 M * Bertl arekm: should be
1297530960 M * arekm Bertl: hm, then it isn't, added MKNOD to guest/bcapabilities but getting mknod("ble", S_IFCHR|0666, makedev(1, 1)) = -1 EPERM (Operation not permitted)
1297531080 M * Bertl what does /proc/virtual/<xid>/status say?
1297531109 M * arekm Bertl: http://pld.pastebin.com/MrjikbBk
1297531135 M * daniel_hozac do you have vserver device mapping enabled?
1297531280 M * arekm in kernel - yes (CONFIG_VSERVER_DEVICE=y) but if this needs some per gust additional configuration then I don't have any
1297531348 M * daniel_hozac you need to configure it.
1297531401 M * daniel_hozac vdevmap --xid 0 --set --open --create --target /dev/null; vdevmap --xid 0 --set --open --create --target /dev/sda
1297531405 M * arekm so I can't have this enabled in kernel but not use this feature?
1297531407 M * daniel_hozac should let any guests with mknod use it as expected.
1297531434 M * daniel_hozac nope
1297531438 M * daniel_hozac it's like proc security that way.
1297531458 M * arekm bah, have to disable it in kernel then ;/
1297531489 M * daniel_hozac or spend three seconds configuring it :-)
1297531491 M * Bertl daniel_hozac: btw, we should be able to remove that in 2.6.38, there is some kind of device mapper in mainline now
1297531502 M * daniel_hozac ah?
1297531524 M * Bertl haven't had the time to look into it yet though ...
1297531551 M * arekm three seconds sure but I had no idea that such configuration is needed, so I doubt other people using this kernel will know about this
1297531561 M * daniel_hozac do you know what it's called?
1297531592 M * arekm would be nice if util-vserver scripts configured this or required configuration if it's enabled
1297531618 M * daniel_hozac well, probably easier to change the default in the kernel.
1297531620 Q * jrdnyquist Ping timeout: 480 seconds
1297531632 J * jrdnyquist ~jrdnyquis@slayer.caro.net
1297531681 M * daniel_hozac see dmap_defaults in kernel/vserver/device.c, add DATTR_CREATE
1297531784 N * ensc Guest1179
1297531794 J * ensc ~irc-ensc@p5DF2D44E.dip.t-dialin.net
1297531888 M * Bertl daniel_hozac: nah, sorry, can't remeber ...
1297531952 Q * Guest1179 Ping timeout: 480 seconds
1297532022 J * hijacker_ ~hijacker@87-126-142-51.btc-net.bg
1297532054 Q * yarihm Quit: This computer has gone to sleep
1297532057 M * Bertl maybe this might be a hint:
1297532063 M * Bertl lxc.cgroup.devices.allow = c 5:1 rwm
1297533237 J * ViRUS ~mp@p579B55B7.dip.t-dialin.net
1297533240 M * arekm daniel_hozac: are there any disadvantages or "weird" cases if I add DATTR_CREATE there?
1297533253 A * arekm googles what vserver device option actually is
1297534536 M * Bertl daniel_hozac: http://www.mjmwired.net/kernel/Documentation/cgroups/devices.txt
1297534562 M * Bertl doesn't allow mapping though, just access control
1297536280 M * daniel_hozac ah, yeah, that's been around for awhile though, right?
1297536304 M * daniel_hozac i remember seeing that before.
1297536315 M * daniel_hozac arekm: no, not really.
1297536322 M * Bertl yes, it was introduced some time ago, it seems to wrok since 2.6.36 (somewhat)
1297536336 M * daniel_hozac just restores default behaviour when you give CAP_MKNOD to guests.
1297536338 M * Bertl and it seems it was 'finished' so far in 2.6.38
1297536355 M * daniel_hozac 'finished'?
1297536357 M * daniel_hozac really? lol.
1297536372 M * daniel_hozac i think i reviewed those initial patches back in 2008 or something :)
1297536377 M * Bertl well, you know LXC, they never complete any subsystem :)
1297536382 M * daniel_hozac true...
1297536391 M * Bertl but it's actually even worse
1297536419 M * Bertl I'm trying to get some information about the LXC console handling (to get some ideas how to implement that for Linux-VServer)
1297536443 M * Bertl for over a week now, and there is nobody on the lxc or lxc-devel channel who would know a thing
1297536456 M * daniel_hozac hehe
1297536518 M * Bertl I'm constantly pointed to daniel lezcano, which simply is never oneline it seems :)
1297536540 M * Bertl (got the feeling he is the only one actively developing LXC atm :)
1297536586 M * daniel_hozac i've gotten that feeling since lxc was started.
1297536650 M * Bertl well, all the IBM folks are hanging around in the channel, but nobody knows any details ... really sad ...
1297536745 M * Bertl it seems they are assigning a certain number of ttys to a container, which in turn the contain uses as console(s)
1297536776 M * daniel_hozac that doesn't really sound scalable.
1297536789 M * Bertl from the kernel code, the tty (in contrast to the ptys) already provides all the 'console' stuff
1297536795 M * daniel_hozac right.
1297536815 M * Bertl well, I guess the ttys can scale similar to ptys
1297536829 M * daniel_hozac i suppose, as long as you never want to look at them... :-)
1297536829 M * Bertl so no loss there (from the kernel PoV)
1297536866 M * Bertl there is an lxc-console, which needs to be run to have a console for the guest
1297536885 M * Bertl i.e. an userspace app, obviously connecting to the tty from outside the container
1297536906 M * daniel_hozac ah, okay.
1297536944 M * Bertl might be interesting for you to check that out, I'll see what options we have from the kernel side today or tomorrow
1297536963 M * Bertl and maybe have a look at the device cgroup, we might be able to merge that with the mapper somehow
1297537098 M * daniel_hozac yeah...
1297538181 M * Mr_Smoke Hi there
1297538226 M * Mr_Smoke Please refresh my memory ... when starting a vserver with a dedicated (public) address, does the address need to exist on the host's eth0 for example BEFORE the guest starts, or not ?
1297538495 M * Bertl doesn't need to exist, it will be created by util-vserver if configured properly
1297538519 M * Bertl but often administrators prefer such addresses to be taken up/brought down by the host system instead
1297538847 M * Mr_Smoke cuz in one instance the vserver won't start (when the IP doesn't exist), and stops with a "cat" error.
1297538848 Q * ViRUS Quit: If there is Artificial Intelligence, then there's bound to be some artificial stupidity. (Thomas Edison)
1297538851 M * Mr_Smoke I'll dig it up and report later
1297538955 M * Bertl care to run the startup with --debug and upload the output to a pastebin?
1297538990 M * Mr_Smoke will do
1297539215 Q * hel Remote host closed the connection
1297539217 J * hel ~hel@porthos.lennackers.de
1297539787 M * Mr_Smoke Huh
1297539796 M * Mr_Smoke vserver-stat reports an empty NAME
1297539804 M * Mr_Smoke That's odd.
1297539829 M * daniel_hozac what version of util-vserver?
1297539831 M * Bertl which probably brings us to the question: what util-vserver version?
1297539897 M * Mr_Smoke 0.30.216_pre2924
1297539902 M * Mr_Smoke Never seen that before.
1297539905 M * Mr_Smoke Dinner, BBL.
1297539910 M * daniel_hozac hmm, what did you do to it? :)
1297544235 M * Mr_Smoke Heh, nothing really 
1297544238 M * Mr_Smoke Just compiled from source
1297544403 M * Mr_Smoke I'm building a clone of the "faulty" guest to experiment with
1297544510 M * Mr_Smoke The scenario for that guest was that, after a cold start of the host, it would not start on the first attempt, but only on the second one
1297544516 M * Mr_Smoke Other guests were not affected
1297544542 M * Mr_Smoke The differences are that this guest has 2 interfaces (IPV4 + IPv6), and that the addresses did not exist at at start time
1297544575 M * Mr_Smoke And now I've got this empty NAME in vserver-stat too
1297544580 M * Mr_Smoke that also affects only this host
1297544592 M * Mr_Smoke Debug output coming right up
1297544767 M * Mr_Smoke Hm
1297544784 M * Mr_Smoke Buggy one has interfaces/*/dev and ip, not prefix
1297545172 M * Mr_Smoke Hm doesn't seem to make a difference.
1297545176 M * Mr_Smoke I'll restart with --debug
1297545219 M * Mr_Smoke Aha
1297545230 M * Mr_Smoke How does one stop a vserver that vserver doesn't know is running ?
1297545233 M * Mr_Smoke empty NAME ..
1297545495 M * Mr_Smoke Ok got it
1297545598 M * Mr_Smoke < ++ cat /etc/vservers/ford/cgroup/memory.memsw.limit_in_bytes
1297545598 M * Mr_Smoke < cat: write error: Invalid argument
1297545601 M * Mr_Smoke O.o
1297545609 M * Mr_Smoke It just says 2.5G
1297545695 J * jordi ~jordi@115.Red-213-96-69.staticIP.rima-tde.net
1297545703 M * jordi ahoy!
1297545743 M * jordi does anyone have any experience wth the squeeze vserver kernel? Are there regressions wrt lenny?
1297545753 M * jordi or anything I need to know offhand before upgrading?
1297545810 M * Mr_Smoke Bertl: weird thing. The 1st start attempt will break while reading cgroup memorysettings
1297545813 M * Mr_Smoke The second doesnt
1297545817 M * Bertl jordi: depends on the kernel you're upgrading to/from 
1297545818 M * Mr_Smoke I'm pasting both logs
1297545822 M * Mr_Smoke in a pastebin
1297545841 M * jordi Bertl: 2.6.26-lenny -> 2.6.32-squeeze
1297545861 M * Bertl well, 2.6.26 is severely broken, so you have to clean up after that
1297545867 M * Bertl (see the wiki page)
1297545874 M * Mr_Smoke Bertl: http://paste.pocoo.org/show/337243/ and http://paste.pocoo.org/show/337244/ are, in that order, the results of 2 subsequent attempts to start this vserver
1297545895 M * Mr_Smoke In this particular occurrence, /var/run/vserver was populated, so I get a name in vserver-stat
1297545904 M * Mr_Smoke As for the rest, I'm a bit lost
1297546020 M * jordi Bertl: what page, exactly?
1297546085 M * jordi because this doesn't say anything good
1297546085 M * jordi http://linux-vserver.org/Installation_on_Debian#Issues_with_Squeeze.27s_2.6.32_Kernel_and_Util-vserver
1297546086 M * Bertl http://linux-vserver.org/Installation_on_Debian#Issues_with_Lenny.27s_2.6.26_Kernel_and_Util-vserver
1297546117 M * Bertl basically you'll have to regen your xattrs with the new kernel
1297546201 M * jordi I see
1297546247 M * jordi 1242116428 M * Konsar my kernel sends messagess vxW: [-ls-,5554:#40001|40001|0] did lookup hidden md3:f1472850[#0,408] -/usr/share-.
1297546253 M * jordi lol, so this is caused by this?
1297546282 M * Bertl depends, it might be caused by all kind of misconfigurations
1297546330 M * jordi I happened after moving to lenny
1297546333 M * jordi from etch
1297546726 M * Mr_Smoke Bertl: did you get the pastes ?
1297546743 M * daniel_hozac Mr_Smoke: that's expected with cgroups.
1297546917 J * cuba33ci_ ~cuba33ci@111-240-166-1.dynamic.hinet.net
1297546956 M * Mr_Smoke daniel_hozac: how do you mean ?
1297546998 M * daniel_hozac if it fails, the second will succeed, as the cgroup setup will be skipped.
1297547022 M * Mr_Smoke Ok but why did it fail then ?
1297547105 M * Mr_Smoke And how can free -m report 3800M used when the total RSS for all guests is about 100M, that also bothers me a lot, suddenly.
1297547119 Q * cuba33ci Read error: Connection reset by peer
1297547125 N * cuba33ci_ cuba33ci
1297547220 M * daniel_hozac used without caches and buffers?
1297547223 M * daniel_hozac or just used?
1297547251 M * Mr_Smoke used
1297547254 M * Mr_Smoke that's very worrying
1297547257 M * daniel_hozac have you tried just using the number of bytes?
1297547270 M * Mr_Smoke 3253 cached
1297547293 M * Mr_Smoke              total       used       free     shared    buffers     cached
1297547294 M * Mr_Smoke Mem:       4120732    3891996     228736          0     304696    3331752
1297547303 M * Mr_Smoke I don't get it.
1297547320 M * daniel_hozac you have 3.6 GiB of flushable content...
1297547399 M * Mr_Smoke IS that what is causing the cgroup to fail ?
1297547773 M * daniel_hozac no
1297547829 M * Mr_Smoke What do you think could be failing then ?
1297547985 M * daniel_hozac like i said, try putting in the number of byets.
1297548011 M * Mr_Smoke Oh
1297548017 M * Mr_Smoke Sorry I had not understood that
1297548034 M * Mr_Smoke So basically 2.5*1024*1024 will do
1297548038 M * Mr_Smoke Let's give it a go
1297548052 M * daniel_hozac that would be mebibytes.
1297548150 M * Mr_Smoke daniel_hozac: that's not working anyways
1297548169 M * Mr_Smoke ah
1297548172 M * Mr_Smoke Silly me
1297548409 M * Mr_Smoke Something is very wrong
1297548435 M * Mr_Smoke the guest won't even start
1297548444 M * Mr_Smoke using --debug now ...
1297548549 M * Mr_Smoke ec/usr/sbin/chbind: line 135: 20227 Killed                  "${create_cmd[@]}" "${chain_cmd[@]}" -- "$@"
1297548554 M * Mr_Smoke O.o
1297548603 M * Mr_Smoke I have an idea
1297548706 M * Mr_Smoke Hm nope, still nothing
1297548707 M * Mr_Smoke F$ck
1297548868 M * Bertl kernel?
1297548891 M * Mr_Smoke 2.6.36-grsec2.2.0-vs2.3.0.36.38
1297548895 M * Mr_Smoke but grsec disabled here
1297548902 M * Mr_Smoke Heck the guest won't start at all now
1297548922 M * Bertl that sounds like one of the older 2.6.36 patches
1297548935 M * Bertl you might be hitting the cred issue
1297548939 M * Mr_Smoke could be, but I never had a glitch before
1297548942 M * Mr_Smoke cred issue ?
1297548978 M * Bertl arekm can give you the details ...
1297549040 M * Mr_Smoke What kernel version should I upgrade to ?
1297549056 M * Bertl you can stay with 2.6.36, but don't use the old patches
1297549063 M * Mr_Smoke Can a reboot help bring the guest back up momentarily ?
1297549063 M * Bertl or apply the fixes
1297549074 M * Bertl very likely
1297549082 M * Mr_Smoke Of is there a hotfix ?
1297549116 M * Mr_Smoke LEt's reboot then
1297549117 M * Bertl nothing besides not entering a guest or not using the old patches
1297549144 M * Mr_Smoke Well rebooting for now. Let's look for a more recent patch
1297549194 M * Mr_Smoke Bertl: are there known issues with the latest one ? or can I use that ? (vs2.3.0.37-rc4)
1297549204 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com
1297549262 M * Mr_Smoke Also, where can I read about this partiular issue ?
1297549312 M * Bertl yes, the recent patches should be fine, as I said
1297549328 M * Bertl you can read up on the issue in the IRC logs, I guess
1297549419 M * Mr_Smoke Wow that's gonna take a while to grep, day by day
1297549445 M * Bertl that's what google is for 
1297549496 M * Mr_Smoke And this is known as ... credential issue ?
1297549507 M * Bertl google for 'site:irc.13thfloor.at cred arekm'
1297549547 M * Mr_Smoke ok thnx
1297549598 M * Mr_Smoke Credentials accounting, right
1297549690 M * Mr_Smoke So vserver <x> exec is safe, but vserver <x> enter will mess it up then ?
1297549706 M * daniel_hozac no.
1297549745 M * daniel_hozac there is no difference between the two, except for the terminal allocation in the enter case.
1297549800 Q * hijacker_ Quit: Leaving
1297549853 M * Mr_Smoke Ok so I just better update to the latest 38-rc then
1297549873 J * areq_ ~areq@q.areq.eu.org
1297549885 M * Bertl yep, definitely
1297549898 M * Mr_Smoke And forget grsec for the moment
1297549899 M * Mr_Smoke ok
1297549957 M * Mr_Smoke Heh, .38 must be out, the rc's gone
1297549969 M * areq_ hi
1297550010 M * areq_ handleDeviceMap should works in util-vserver-0.30.216-1.pre2926 ?
1297550110 M * daniel_hozac yes.
1297550111 M * areq_ I try add vdevmap --xid 42 --set --open --create --target /dev/null; vdevmap --xid 42 --set --open --create --target /dev/root to config
1297550152 M * Mr_Smoke Let's go with rc4 then
1297550220 M * areq_ I creared: /etc/vservers/builder/apps/vdevmap/{root,null}/{create,open,target)
1297550257 M * areq_ # cat /etc/vservers/builder/apps/vdevmap/root/target
1297550257 M * areq_ /dev/root
1297550424 M * areq_ during vserver start handleDeviceMap is called with empty xid ($2):
1297550424 M * areq_ $1 --set
1297550424 M * areq_ $2
1297550424 M * areq_ $3 /etc/vservers/builder/apps/vdevmap
1297550454 M * areq_ with vserver stop looks better:
1297550469 M * areq_ $1 --unset
1297550469 M * areq_ $2 100
1297550469 M * areq_ $3 /etc/vservers/builder/apps/vdevmap
1297550479 M * Bertl and does your guest 'builder' have a context?
1297550490 M * areq_ vc_unset_mapping(): No such process
1297550504 M * areq_ # cat /etc/vservers/builder/context
1297550504 M * areq_ 100
1297550511 M * Mr_Smoke FWIW, -vs2.3.0.37-rc4 applies to 2.6.38-rc4 with 2 minor rejects that are easily fixed by hand
1297550695 M * Bertl k, tx
1297550846 M * areq_ $S_CONTEXT in /usr/lib64/util-vserver/vserver.start is empty
1297550938 M * BenG Bertl, I want a good environment to put production servers on. Out of the currently available kernel versions and patches, which would you suggest?
1297550969 M * Bertl 2.6.38, then 2.6.36 and then probably 2.6.22
1297551002 M * BenG 2.6.38 patches are marked rc though
1297551009 M * BenG still a good bet though?
1297551051 M * Bertl they work quite fine here, and I have them marked rc because I didn't get to finish the cleanup planned for vs2.3.0.37 yet
1297551076 M * BenG ah okay, I'll build an image up then, thankyou
1297551090 M * Bertl there should be a vs2.3.0.37 soon though
1297551121 M * BenG and hopefully a final 2.6.38 to go with it
1297551130 M * areq_ any other files in vdevmap are required ?
1297551146 M * Mr_Smoke Bertl: one last question, i've read about namespaces somewhere, but didn't quite put my head around it yet. Where should i read about that in the vserver context ? wiki only has what looks like "old" stuff
1297551212 M * Bertl depends on what you want to know about them
1297551228 M * Bertl you are already using them, unless you explicitely configured nonamespace
1297551295 M * Mr_Smoke oh ok then, nothing to specifically configure in the kernel then
1297551328 M * Mr_Smoke I mean
1297551347 M * Mr_Smoke Should they all be enabled ? user, pid, network, uts, ipc ,
1297551418 M * Bertl you usually want most of them, but you do not want to enable the namespace cgroup
1297551425 M * Mr_Smoke ok
1297551454 M * Mr_Smoke yeah i had that set up already apparently
1297551466 M * Mr_Smoke there's this new swap extension by default i need to read about
1297551488 M * Bertl the pid namespace is not used (yet) the network namespace not integrated into util-vserver (at least not completely, daniel_hozac, please correct me if wrong)
1297551512 M * Bertl memory and swap controller are already supported and used when enabled
1297551551 M * Mr_Smoke ok
1297551726 M * Mr_Smoke i'll finish this kernel later i think
1297551769 M * Mr_Smoke Thanks for all the help :)
1297551778 M * Mr_Smoke looking forward to the next stable kernel
1297551787 M * Mr_Smoke I must donate too, soon :)
1297551824 M * Bertl you're welcome! and thanks in advance :)
1297551865 Q * BenG Quit: I Leave
1297551958 M * Mr_Smoke daniel_hozac: oh, also, memory limits in cgroups work fairly well if you specify them as 2048M for example
1297551958 M * areq_ what about my vdevmap problem ? any idea what I did wrong ?
1297551970 M * Mr_Smoke The issue must have been the credentials accounting
1297551988 M * Mr_Smoke I set mine to 2048 and 2560M, and the guest started like a charm
1297552021 M * Bertl areq_: just give it a moment, I'm sure daniel_hozac will look into it when he finds the time
1297552034 M * Bertl (if he's not already doing so :)
1297552058 M * Mr_Smoke :)
1297552061 M * Mr_Smoke Probably is
1297552068 M * Mr_Smoke You guys are so dedicated
1297552071 M * Mr_Smoke I mean it
1297552076 M * Mr_Smoke *always* excellent support 
1297552081 M * Mr_Smoke Hence my pledge to donate :)
1297552115 M * areq_ Bertl: ok ;)
1297552199 M * Mr_Smoke Also, I now have a decent (core2Due) machine I can use to experiment with
1297552217 M * Mr_Smoke so maybe that can be useful for debugging stuff later
1297552252 M * Bertl giving the latest 2.6.38-rc a shakedown won't hurt, and if you encounter any issues, just let us know
1297552262 Q * petzsch Quit: Leaving.
1297552291 M * Mr_Smoke Yeah, I'll see what i can do
1297552425 M * Mr_Smoke There :) night for now !
1297552434 M * Bertl nn
1297554431 M * Guy- Bertl: I hit some bugs with 2.6.38-rc, but nothing that looked vserver specific
1297554444 M * Bertl what kind of issues?
1297554457 M * Guy- an mdadm related spinlock problem resulting in a lockup
1297554479 M * Bertl okay, unlikely that is related
1297554483 M * Guy- and another lockup caused by removing something from the conntrack table using conntrack -D
1297554508 M * Guy- yeah, Neil Brown already has a fix for the spinlock issue
1297554568 M * Bertl okay, as we do not modify anything in iptables, most likely unrelated too
1297554579 M * Guy- that's what I thought