1101082401 M * Doener mugwump: around?
1101082556 M * Bertl ah, ping inside ngn context now works ;)
1101082636 M * Doener going the otherway round this time? ping first, then the rest? ;)
1101082680 M * Bertl well, we did investigate this last time and I thought this would be a good check/start
1101082715 M * Doener what's been the problem there?
1101082771 M * Bertl well, I'm not sure that I fixed it properly ... but on the turn around (whre the network stack answers the icmp_echo) the skb already contains an nfmark
1101082795 M * Bertl I added a check there, to leave that unchanged and not overwrite it with the sk value ...
1101082823 M * Bertl seems to work fine, but I added a FIXME comment .. because I don#t want to investigate now ...
1101082834 M * Doener so a conflict between 'normal' and vserver tagging?
1101082860 M * Bertl not really, more a conflict between vserver and vserver tagging ;)
1101082906 M * Bertl in dev_queue_xmit() ngn sets the skb to the context
1101082925 M * Bertl the nfmark to the xid of the socket
1101084290 Q * thh Quit: ChatZilla 0.9.61 [Mozilla rv:1.7/20040616]
1101084671 M * Bertl night everyone .. back tomorrow ...
1101084701 N * Bertl Bertl_zZ
1101086907 Q * albeiro Ping timeout: 480 seconds
1101086909 Q * matti Ping timeout: 480 seconds
1101086926 Q * redLED Read error: Connection reset by peer
1101086947 J * matti matti@linux.gentoo.pl
1101087201 J * albeiro albeiro@lexx.eu.org
1101111179 J * Val val@gj403.loria.fr
1101112307 J * BWare bware@212.26.196.41
1101113256 J * jsambrook jsambrook@aelfric.plus.com
1101117928 J * rs rs@ice.aspic.com
1101117934 M * rs hey dudes
1101118384 M * Doener night folks...
1101118389 N * Doener Doener_zZz
1101118632 Q * Loki|muh Ping timeout: 480 seconds
1101118850 J * Loki|muh loki@satanix.de
1101120110 Q * monrad Remote host closed the connection
1101120129 Q * jsambrook Ping timeout: 480 seconds
1101120722 M * eyck morning,
1101120725 M * eyck night?
1101120839 M * Loki|muh high noon ;)
1101120934 J * Session session@85.96.108.140
1101120965 Q * Session Quit: 
1101121387 J * meebey meebey@meebey.net
1101121404 M * meebey hi ya
1101121432 M * meebey Bertl is sleeping *waiting till he wakes up*
1101122415 M * Hollow Loki|muh: were you running gentoo as vserver guest?
1101125083 Q * Hollow Remote host closed the connection
1101125259 J * redLED redled@d51A4EFA7.kabel.telenet.be
1101126361 J * Hollow bene@home.xnull.de
1101127134 Q * Hollow Ping timeout: 480 seconds
1101127165 J * Hollow bene@home.xnull.de
1101130263 Q * Loki|muh Remote host closed the connection
1101130263 J * Loki|muh loki@satanix.de
1101131442 Q * albeiro Quit: leaving
1101131570 M * brc anyone experiencing server crashes with 2.6 ?
1101131716 M * infowolfe brc: none here
1101131720 M * infowolfe and i'm loading it pretty hard
1101131731 M * infowolfe what kind of crashes are you experiencing?
1101131759 M * infowolfe Hollow: did you have a questn about gentoo vserver guests?
1101132422 M * Loki|muh Hollow: yeah
1101132561 M * brc infowolfe: with 2.6.8.1 it crashed like once/twice a week
1101132567 M * brc wioth 2.6.9 one crash on 17 days
1101132577 M * brc the server just dies, i dont know exactly what happened cause it is a dedicated server.
1101132583 M * brc infowolfe: which kernel are you using?
1101132756 M * TheSeer brc: are those crashes vserver related?
1101132766 M * TheSeer brc: e.g. do they occur on a vanilla system too?
1101132997 M * brc i don't know if they would occur with a vanilla system :(
1101133018 M * brc and there is nothing on the logs
1101133020 M * brc after crashes
1101133236 M * TheSeer how much ram do you have on that box?
1101133247 M * infowolfe brc: 2.6.9-vs1.9.3
1101133256 M * TheSeer i have a (customer) server that runs fin with 1 GB ram..
1101133266 M * infowolfe TheSeer: i've got 1GB of ram too
1101133268 M * TheSeer as soon as i add a 2nd GB to it, the box keeps crashing
1101133278 M * TheSeer it just randomly dies
1101133292 M * infowolfe TheSeer: areyou using himem?
1101133293 M * TheSeer it doesn't matter "which" Gig to use
1101133311 M * infowolfe TheSeer: i'd give the info to Bertl...
1101133314 M * TheSeer e.g. i can swap the modules around to no effect.. just having them all in kills the box
1101133321 M * infowolfe TheSeer: current vps uptime is  14:20:57 up 4 days,  8:36,  3 users,  load average: 0.09, 0.55, 0.95
1101133333 M * TheSeer infowolfe: the box is running a vanilla-kernel or a redhat/fedora default kernel
1101133334 M * infowolfe TheSeer: kernel config issue?
1101133335 M * TheSeer no difference
1101133350 M * infowolfe TheSeer: you should run a -bigmem kernel if you're going to be using over 1GB ram
1101133352 M * TheSeer it's not even having vserver installed
1101133368 M * TheSeer redhat doesn't have that crap ;>
1101133375 M * TheSeer -bigmem kernel reminds me of suse ;>
1101133391 M * infowolfe TheSeer: since when?
1101133399 A * infowolfe hasn't used redhat since 8.0
1101133455 M * TheSeer at least the kernel rpm's supplied on CD are only normal and SMP
1101133471 M * brc the box have 512 ram
1101133483 M * brc has
1101133490 M * brc 16 vservers and is allways swapping out
1101133507 M * TheSeer 16 vservers on 512 MB is a taff call ;)
1101133538 M * eyck why?
1101133546 M * eyck depends on what they're doing
1101133593 M * TheSeer sure.. but 16 vservers leaves each vserver with about 32 Megs (statistically ;) )
1101133615 M * TheSeer and 32 Meg isn't too much for any app these days
1101133648 M * TheSeer and given the fact he already stated that box is constantly swapping...
1101133738 M * TheSeer e.g. my root-server uses 70 Megs..
1101133751 M * TheSeer the qmail-mx vserver is currently at 250 Megs..
1101133755 M * TheSeer and it's hardly busy..
1101133760 M * TheSeer as of right now at least..
1101133835 J * albeiro albeiro@linux.gentoo.pl
1101133902 M * brc i know
1101133911 M * brc all the vservers belong to friends
1101133916 M * brc so we all try not to push out a lot the machine
1101133921 M * brc i am going to unify the vserver
1101133926 M * brc do you think that will help al ot ??
1101133939 M * brc anyway, even swaping a lot wouldn't cause crashing, would it ?
1101134035 J * id_ test@relax-media.softwarezentrum.de
1101134046 M * id_ Hi #vserver
1101134120 M * infowolfe hi id_ 
1101134145 M * infowolfe brc: each of my apache processes consume ~ 15MB ram each...
1101134162 M * infowolfe no matter how polite your friends are... 512M of ram just isn't enough
1101134198 M * brc right now:
1101134204 M * brc [root@localhost root]# free
1101134205 M * brc              total       used       free     shared    buffers     cached
1101134205 M * brc Mem:        513812     511692       2120          0       7232     151224
1101134205 M * brc -/+ buffers/cache:     353236     160576
1101134205 M * brc Swap:      2048276      78724    1969552
1101134206 M * brc [root@localhost root]#
1101134218 M * brc tehy are really polite ;)
1101134316 J * Pazzo thomas@host130-250.pool8172.interbusiness.it
1101134348 M * brc infowolfe: have you unified your vserver ?
1101134378 M * infowolfe brc: i use gentoo, unifying won't really save me much
1101134423 M * infowolfe not only that, but here: /dev/sda8             191G  5.4G  186G   3% /vservers
1101134426 M * infowolfe i don't think i really need to.
1101134429 M * infowolfe lol
1101134449 M * infowolfe that's 5 vservers
1101134565 M * brc unify also helps with MEMORY
1101134594 M * brc so maybe it would help you
1101134613 M * brc 200gb hard disk, i wish i had one of those :)
1101134645 M * eyck I'd prefer one of those 400G seagates
1101134655 M * infowolfe brc: it's not a 200GB disk... it's 4 160-GB disks in raid 10
1101134681 M * infowolfe and i've got a GB of ram and i don't use swap... so yah
1101134733 M * infowolfe eyck: i want 8 raptor 74GB 10krpm S/ATA with nice S/ATA 3ware card :-p
1101134758 M * eyck infowolfe: I like sleeping in quiet...
1101134782 M * eyck infowolfe: and for servers if I could wish I'd go for one of fire-wire racks
1101134810 M * brc hm
1101134811 M * infowolfe eyck: quiet? that's very quiet... lol, when put in the closet (after you've routed fiber in there)
1101134849 M * brc i have a machien with 2 80gb satas
1101134878 M * brc DUAL XEON 3.06 GHZ (ht)
1101134880 M * brc 2gb ram
1101134898 M * brc how much vservers do you think it is able to hold ?
1101134905 M * eyck 1024 ?
1101134926 M * infowolfe brc: how heavy are they going to be loaded?
1101134926 M * rs brc: memory is the most important thing for vserver hosting
1101134932 M * brc 2gb ram
1101134939 M * brc maybe i should add more 2 gb
1101134942 M * infowolfe brc: i could take that server down myself...
1101134947 M * infowolfe lol, without vservers.
1101134953 M * brc hehe ok
1101134976 M * infowolfe and rss limits are a bad thing if your people like to compile, 256M and a make -j9 will kill the rest of the vserver
1101135001 M * brc if there was a way to limit a vserver on X memory, and after that start swapping just for that vserver
1101135021 M * rs it's not possible for moment
1101135029 M * brc maybe everything would get slow due to IO of swapping?
1101135037 M * infowolfe brc: the code isn't there right now.
1101135042 M * brc even the vservers that are not swapping would get slow
1101135059 M * infowolfe brc: it's a moot point, the code doesn't exist to do that right now.
1101135084 M * brc i know, i was just wondering :)
1101135491 M * infowolfe brc: loaded the way my machine gets loaded, i'd say you might be able to get 4 vservers out of that dual xeon box
1101135673 M * brc for people with that high load it is better to get a dedicated server, isn't it ?
1101135679 M * brc a GOOD ONE
1101135679 M * brc hehehe
1101137141 M * infowolfe brc, a dedicated celeron 2.4 can't handle the sites i'm migrating to my box right now...
1101137680 M * brc where do you host your servers?
1101137966 M * infowolfe brc: the cel 2.4 is at ovh in fr, the quad xeon box is at mesh solutions in de
1101137972 M * infowolfe brc: where are you located?
1101138169 M * infowolfe brazil... hrm... i don't know of any good hosting providers in brazil, sorry
1101138327 M * brc I am on brazil
1101138330 M * brc there aren't good hostings here
1101138333 M * brc our bandwith is pretty expensive
1101138337 M * brc so i need to get machines on US
1101139106 Q * id_ Quit: Verlassend
1101139446 J * DuckMaster Duck@dyn-83-152-173-228.ppp.tiscali.fr
1101139874 Q * DuckKing Ping timeout: 480 seconds
1101140913 M * infowolfe brc: i'd suggest www.hostventures.com or www.quik-host.net
1101140923 M * infowolfe they're both friends of mine and both good at what they do
1101141039 P * Pazzo 
1101141096 J * monrad monrad@213083190130.sonofon.dk
1101142671 J * DrJES chatzilla@TradeMart-2.EDnet.NS.CA
1101142941 J * sim0n simon@p508E8E34.dip0.t-ipconnect.de
1101142955 M * sim0n hi
1101143000 M * sim0n does anybody know if quota is already working for kernel-2.6 ?
1101143286 M * rs quota inside vserver or disk limits ?
1101143296 M * sim0n quota inside vserver
1101143312 M * rs it not yet implemented AFAIK
1101143318 M * sim0n i currently run v2.4, have every vserver on a seperated partition, and use quota inside vserver
1101143350 M * sim0n i can't switch to v2.6 until that works as well :-/
1101143966 M * brc infowolfe: thanks! gonna check them
1101143987 M * infowolfe brc: no problem
1101143999 M * brc infowolfe: so you have no crashing problems with 2.6.9 
1101144002 M * infowolfe brc: i'd really suggest quik-host, his prices should be a little better, but check them both out
1101144006 M * brc i have no luck hehehe
1101144023 M * infowolfe brc: nope, none at all... using a himem kernel config and i have smp enabled
1101144033 M * infowolfe attempt to remove acpi from the kernel, maybe i'tll fix it
1101144076 Q * BWare Ping timeout: 480 seconds
1101144100 M * brc hmm good idea
1101144102 M * brc gonna try that
1101144163 M * DrJES Why does my 2.6 kernel not all vserver enter to work? Gives :chcontext: vc_new_s_context(): Operation not permitted
1101144329 M * DrJES Latest 1.93 applied to 2.6.9.... ?
1101145202 N * Bertl_zZ Bertl
1101145238 Q * lilo Read error: Connection reset by peer
1101145239 J * lilo_ lilo@lilo.usercloak.oftc.net
1101145255 M * Bertl morning folks!
1101145277 M * Bertl DrJES: please elaborate!
1101145333 M * DrJES Doing a "vserver myserver enter" results in that error. So does vserver stop. So I can not stop/enter my vserver :(
1101145344 J * ensc ircensc@ultra.csn.tu-chemnitz.de
1101145360 M * Bertl welcome ensc!
1101145366 M * Bertl DrJES: did you start it?
1101145391 M * DrJES Yes, start works fine ;).
1101145416 M * Bertl ah, okay, so you start it, and does it show up with vserver-stat then?
1101145441 M * DrJES chcontext --ctx 34534 bash did not work until I rebooted (the main server) and tried again. Now it (chcontext) works, just no vserver enter.
1101145445 M * DrJES vserver-stat works fine.
1101145463 M * DrJES Oddly I'm sure I was testing all this ok when I had the netstat -nap problems a while ago.
1101145467 M * Bertl so it does show the vserver as running, right?
1101145472 M * DrJES Yes.
1101145494 Q * rs Quit: home
1101145499 M * Bertl okay, I got a few of those reports, but it couldn't be tracked down, are you interested in hunting that issue?
1101145503 M * DrJES I thought it didn't like to the big ctx number, so I even made it S_CONTEXT=6 in my conf.... no help ;(
1101145514 M * DrJES Tell me how to hunt :)
1101145538 M * Bertl do you ahve VSERVER_DEBUG enabled in your config
1101145543 M * DrJES I'm also seeing that tomcat will not start on it, but one thing at a time ;).
1101145546 M * DrJES No....
1101145554 M * DrJES I supose I could start another one with that enabled....
1101145559 M * Bertl okay, how easy is it for you to recompile a kernel?
1101145591 M * DrJES Easy to recompile. More difficult to reboot the live box :(. But I'll try to do what must be done!
1101145624 M * Bertl okay, because we probably have to add some checks in the questionable places ... but probably later ...
1101145631 M * DrJES by the way, config is a legacy config if that matters.
1101145645 M * Bertl first let's start with describing your setup/distro and configs
1101145668 M * Bertl 2.6.9 + vs1.9.3 right?
1101145933 A * redLED grins
1101145940 M * redLED Bertl, got more patchy fun for me? :)
1101145944 M * redLED oh yeah
1101145947 M * redLED hi btw :D
1101145948 M * Bertl hey redLED! not yet ... 
1101145971 M * Bertl but I'm on it ;)
1101145979 M * redLED good to hear :D
1101145992 M * Bertl hmm, no actually there is a new one, you tested ng1.1 right?
1101146021 M * redLED delta-vs1.9.3.7-ng1.1.diff
1101146043 M * Bertl http://vserver.13thfloor.at/Experimental/delta-vs1.9.3.7-ng1.2.diff
1101146047 M * Bertl hehe!
1101146067 M * redLED does that delta applies against clean, or against ng1.1?
1101146069 M * Bertl you have to back out the old one (patch -R)
1101146087 M * Bertl so it's against vs1.9.3.7 clean
1101146101 M * DrJES Fedora core 2 
1101146119 M * DrJES (busy doing other things)... 
1101146120 M * Bertl glibc version? gcc/binutils?
1101146132 M * Bertl np
1101146169 M * DrJES rpm -q glibc = glibc-2.3.3-27
1101146183 M * DrJES binutils-2.15.90.0.3-5
1101146261 M * DrJES np?
1101146269 M * Bertl np = no problem ;)
1101146293 M * Bertl gcc is 3.3.x or 3.4.x?
1101146423 M * Bertl redLED: this should allow you to make context local connections
1101146443 M * Bertl (e.g. lynx to in vserver apache, or ssh to localhost)
1101146472 M * redLED wow
1101146474 M * redLED thanks for that
1101146479 M * redLED i never even knew patch had a -R :p
1101146539 Q * eyck Ping timeout: 480 seconds
1101146716 M * Zoiah redLED: never had patch say: "this patch is already applied (maybe you meant -R?)"?
1101146810 M * DrJES gcc version 3.3.3 20040412 (Red Hat Linux 3.3.3-7)
1101146839 M * Bertl okay, you compiled the tools yourself?
1101146856 M * Bertl (util-vserver-0.X.Y?)
1101147050 M * DrJES util-vserver-0.30.196-1mdk
1101147063 M * DrJES No, did not compile my self.... Maybe I could try that ...
1101147075 M * Bertl that my mandrake binary then?
1101147104 M * Bertl +'s
1101147117 M * DrJES yes ;)
1101147138 M * Bertl didn't know that this would work on fc2 ;)
1101147161 M * DrJES Always did ;). Misses some deps (like perl) but worked .
1101147222 M * redLED Zoiah, i don't apply patches very often :D
1101147234 J * eyck eyck@81.219.64.71
1101147239 M * Bertl DrJES: okay, could you upload your vserver config (of the vserver which fails to be entered) somewhere? (you can replace sensible information)
1101147251 M * Bertl wb eyck!
1101147372 M * DrJES It's short
1101147375 M * DrJES IPROOT=10.10.10.10
1101147376 M * DrJES IPROOTDEV=eth0
1101147378 M * DrJES ONBOOT=no
1101147379 M * DrJES S_HOSTNAME=something.ednet.ns.ca
1101147381 M * DrJES S_FLAGS="lock"
1101147382 M * DrJES S_CAPS="CAP_KILL CAP_SETGID CAP_SETUID CAP_SETPCAP CAP_NET_BROADCAST CAP_SYS_NICE"
1101147384 M * DrJES S_CONTEXT=66
1101147410 M * Bertl hmm, didn't find a CAP_ALL eh? ;)
1101147443 M * Bertl okay, could you try to remove the "lock" flag and see if that makes it work as expected?
1101147458 M * DrJES 10-4...
1101147491 M * redLED Bertl, rebooting into ngn1.2 now
1101147506 M * Bertl good lap time!
1101147651 M * sim0n hi Bertl!
1101147661 M * Bertl hey sim0n! how are you?
1101147669 M * sim0n i'm fine thnx...how r u?
1101147689 M * Bertl good, thanks!
1101147698 M * sim0n i just wanted to ask how quota inside vserver was doing... for the v2.6 kernels?
1101147704 M * sim0n :-)
1101147719 M * Bertl not at all .. still on the todo list ...
1101147724 M * sim0n hmm kay
1101147736 M * Bertl currently I'm attacking networking ...
1101147764 M * sim0n i guess it's not just a copy/past-hack-hack thing which is "pretty" quickly done ? :-P
1101147766 M * sim0n k
1101147781 M * Bertl well, you can try ;)
1101147823 M * sim0n :-D... was just a question.. i have no idea what's different regarding that code, from 2.4-2.6
1101147848 M * Bertl but the following would be easy: port the vroot device and use separate partitions
1101147863 M * sim0n that's exactly what i'm using....
1101147890 M * Bertl well, that should be relatively easy, I thought you meant real context quota on shared partitions
1101147898 M * redLED Bertl, behaviour appears identical, the wrong interface is being bound for outgoing packets
1101147911 M * sim0n no no ... i'm using lvm and have every vserver on it's own partition..
1101147931 M * Bertl redLED: yes, that is expected, check for localhost conenctions and connections to your vserver ip from inside the vserver
1101147979 M * Bertl sim0n: did you already get used to the changes in 2.6/vs1.9.x?
1101148031 M * sim0n no.. i haven't tried that yet, because the lack of quota support kept me from upgrading. i'm using v2.6 on any other server though..
1101148071 M * Bertl well, I'd suggest the following: you make some test setup with 2.6 vserver (where you can testdrive your quota stuff)
1101148095 M * Bertl you can do so now by copying over the 'real' block device into the vserver
1101148136 M * Bertl this won't be secure, but it should work for testing, in a few days (maybe tomorrow) I can hack-port the vroot device, and you can then test this ...
1101148174 M * sim0n what do you mean by "copying over the real block device into the vserver" ?
1101148180 M * DrJES Second vserser without  S_FLAGS="lock" is same error
1101148187 M * redLED Bertl, no connectivity between the vservers, or between vservers and host
1101148192 M * sim0n ah... hmm... mknod.. set it up in fstab... us the real one.. ?
1101148218 M * Bertl redLED: yeah, expected too, but what about connectivity within a vserver?
1101148221 M * DrJES Odd, First vserser (no changes made yet) is not giving : chcontext: sethostname(): Operation not permitted
1101148237 M * DrJES s/not/now/
1101148244 M * Bertl sim0n: or just cp -va /dev/vgx/lala /vservers/dev/hdv1
1101148315 M * Bertl DrJES: okay, could you check if you kernel config contains CONFIG_VSERVER_DEBUG and if it is enabled?
1101148354 M * Bertl redLED: for example ping 127.0.0.1 or ssh root@localhost
1101148380 M * redLED Bertl, 127.0.0.1 indeed appears to work, and even exist individually between vservers (each vserver connects to own services for 127.0.0.1)
1101148381 M * sim0n bertl ok. do i have to change a lot regarding config files... when migrating to v2.6 ? i have to regenerate the quota files, right ?
1101148418 M * Bertl quota files are not such an issue, old files should work fine
1101148441 M * Bertl the legacy config will work (at least to some extend) with 2.6./vs1.9.3 and newer tools
1101148456 M * Bertl but it is advised to migrate to the new config ...
1101148520 M * redLED hahaha
1101148525 M * redLED *cracks himself up*
1101148550 M * redLED default fc3 vserver rollout ends up with a blank (::) rather than blocked (:*:) root password
1101148557 M * redLED so i roll out sshd, and start it
1101148563 M * redLED (this is elsewhere, not the ngn boxes)
1101148573 M * redLED it took it 20 seconds to get remotely rooted by an ssh worm :)
1101148583 Q * eyck Remote host closed the connection
1101148585 M * Bertl lol!
1101148594 J * eyck eyck@81.219.64.71
1101148596 M * DrJES It is not enabled. I'll build one with that turned on...
1101148607 M * Bertl okay, please do so ...
1101148618 M * redLED Bertl, indeed, 127.0.0.1 is well behaved :)
1101148624 M * Bertl excellent!
1101148639 M * Bertl let's see if you can sniff packets between the vservers
1101148650 M * Bertl (you might need to give CAP_NET_RAW for that
1101148665 M * redLED hmm
1101148672 M * redLED you mean you want me to tcpdump from *inside* a vserver?
1101148677 M * Bertl yep
1101148710 M * redLED hey, i was wondering btw
1101148729 M * redLED remember i told you i need some kind of seperate, detached routing between the host machine, and the vservers in it for some sick purpose?
1101148735 M * redLED and you sent me the UML way
1101148762 M * redLED i was wondering, could i create a bunch of ettertap devices, and let each vserver see only one, and be able to mess around with that ettertap device for as much as it pleases?
1101148792 M * Bertl yes, you can, but you won't be able to use it as you think you could ;)
1101148824 M * Bertl the reason for that is the _shared_ network stack
1101148844 M * Bertl you assign ip 10.0.0.1 to the tun/tap device
1101148858 M * Bertl give that ip to the vserver, send some packets and ...
1101148866 M * Bertl they never get routed through the tun/tap
1101148874 M * sim0n bertl ok... i will give it a try
1101148906 M * Bertl let me know if/when it works, I'll make sure that the vroot device is available then ...
1101148928 M * redLED Bertl, oh :|
1101148947 M * Bertl that is, why UML is better for this purpose ...
1101148963 M * Bertl (the packets will be forced to pass such an interface)
1101148966 M * redLED uml is too much overhead for my purposes :|
1101149014 M * Bertl I'm pretty sure linux/linux-vserver can probably do what you want, but you ahve to rethink your approach ...
1101149080 M * sim0n bertl ok..thnx a lot
1101149087 M * Bertl you're welcome!
1101149158 M * redLED Bertl, tcpdump from inside vserver (with the appropriate cap) works, and sees entire traffic of interface
1101149176 M * redLED Bertl, just as if i'd run it on the host
1101149203 M * redLED Bertl, what are we interested in, packets between a vserver and it's host, or packets between two vservers?
1101149238 M * DrJES Bertl: I'm ready for a reboot but have to wait 'til users are off. Guess I will email to vserver list when I get it back if you are not available?
1101149272 M * redLED DrJES, that's what 'wall' is for :p
1101149280 M * Bertl DrJES: well, you can, but usually you'll find me here, when I'm available
1101149300 M * redLED DrJES, echo "Emergency, this is not a drill! Server going for emergenecy maintenance in 25 seconds! RUN!" | wall :p
1101149364 M * Bertl redLED: we are looking for vserver 'local' traffic (e.g. ping/ssh inside vserver A) and if we can see that on vserver B (looking at lo)
1101149401 M * Bertl next step after that is vserver outbound traffic (from vserver A) and visibility on ethX inside vserver B
1101149429 M * redLED Bertl, we see it on the 'lo' interface, yes
1101149483 M * Bertl okay, could you strace -fF the tcpdump inside vserver B and upload that somewhere?
1101149500 M * DrJES :). Hard to send out to web users :). 
1101149508 M * Bertl redLED: (just watching a single ping)
1101149730 M * redLED Bertl, done, available on same URL
1101149754 Q * eyck Ping timeout: 480 seconds
1101150852 M * Bertl redLED: do you have ipv6 enabled, I guess yes, right?
1101151088 M * redLED Bertl, yep
1101151113 M * Bertl could you try with ipv6 disabled? at kernel compile time or module not loaded
1101151129 M * Bertl hmm, compile time preferred actually ...
1101151145 M * redLED i could unload the module ofcourse, that's easy
1101151152 M * redLED i could also uncompile it completely
1101151154 M * Bertl okay, try that first
1101151156 M * redLED as in, no v6 support at all
1101151185 M * Bertl if you can sniff across vserver with ipv6 unloaded, please retry with ipv6 compiled out ...
1101151209 M * redLED hmm
1101151214 M * redLED it refuses to get gracefully unloaded
1101151220 M * redLED i guess i'll rebuild a kernel without it
1101151229 M * Bertl okay, is fine with me too ...
1101151311 M * redLED on it
1101152231 A * sim0n is away: i'm gone.... 
1101152238 M * Bertl cya
1101152717 M * brc bertl
1101152724 M * Bertl yup?
1101152786 M * brc has anyone, bsides me, reported crashes with 2.6.9-vs1.9.3 ?
1101152824 M * Bertl hmm ..guess so, but it seems they are mainline issues ...
1101152847 M * brc would netconsole help me trace the issue ?
1101152867 M * Bertl read your discussion about that, a few thinks worth to consider:
1101152882 M * Bertl - avoid highmem especially with 2GB memory
1101152896 M * Bertl - enable the NMI watchdog
1101152911 M * Bertl - disable unrelated and unused stuff in the kernel
1101152926 M * Bertl - do not enable netconsole unless you use it
1101152950 M * Bertl - make sure that irqs and apic is working properly 
1101152966 M * Bertl - if possible get a serial console
1101153017 M * brc never head about nmi watchdog, gonna google for it
1101153020 M * brc apic ?
1101153075 M * Bertl see Documentation/nmi_watchdog.txt (in your kernel source)
1101153114 M * Bertl APIC = Advanced Programmable Interrupt Controller (something every SMP system has, and most newer system have too)
1101153227 M * brc ok
1101153242 M * brc took note about everything
1101153261 M * Bertl also testing system stability with stress testing tools is a good idea
1101153280 M * brc with 2.6.8.1 i couldn't get 5 days uptime. with 2.6.9 i got 15 days
1101153304 M * Bertl for example, I once had a system which was 'stable' as long as I didn't add a 3rd memory stick ...
1101153323 M * Bertl didn#t matter which memory or what combination ...
1101153370 M * brc would memtest report that problem ?
1101153385 M * Bertl really depends, in this case it wasn't even memory related
1101153403 M * Bertl it seems that the load of the 3rd stick caused issues with the cpu bus
1101153405 Q * DrJES Quit: no reason
1101153420 M * Bertl but it can't hurt to test with various tools
1101153430 M * Bertl including memtest and 24h kernel compiles
1101153447 M * brc weird, i though machines were more stable :)
1101153482 M * Bertl most hardware is _cheap_ ...
1101153511 M * Bertl and the typical windows user can't tell a system crash from a harware issue ...
1101153514 M * brc bad quality hardware
1101153522 M * brc that's true hheheh
1101153527 M * Bertl but it could also be a kernel or vserver issue ...
1101153542 M * Bertl we just don't know until we see some traces ...
1101153549 M * brc My desktop is on a cheap asrock mb, woth xp2.4+ and the cheapest ddr 512mb ram
1101153556 M * brc Never crashes, and i am almost everyday here
1101153564 M * Bertl btw, if you have a serial console, and the system locks hard, then chances are good that you can get a trace anyway
1101153588 M * brc not possible to have serial console. that's why i liked the idea of netconsole
1101153611 M * Bertl yeah, netconsole is better than nothing, but it might cause issues itself
1101153626 M * brc hehehe
1101153628 M * Bertl it is at a very low level in the network stack/drivers
1101153712 M * brc there should be something like hdconsole
1101153721 M * brc so we would have a small partitionf or logs
1101153725 M * brc and it would just >> to it
1101153730 M * brc would be less risky than netconsole ?
1101153764 M * Bertl guess that won't work, a possible solution might be the kernel exec stuff which is currently tested
1101153850 M * brc you told em to avoid highmem specially with 2gb
1101153854 M * brc what should i use ?
1101153875 M * brc i was using a kind of "generic" kernel, with 2gb enabled
1101153899 M * brc one of the machiens has jsut 512mb
1101153921 M * Bertl vs1.9.3 has the split patches for that reason
1101153948 M * Bertl selecting the 2/2 split instead of the usual 3/1 will allow to use up to 2GB without configuring highmem
1101153952 M * brc ahh ok just remembered i saw that on the kernel configuration
1101155033 M * albeiro hm...
1101155068 M * albeiro Bertl: forgive me offtopic question, but as we are talking about hardware...
1101155098 M * albeiro my machine behaves in a very strange way - it restarts frequently during compilation
1101155127 M * albeiro it can for example compile such a monsters as qt/gcc/glibc
1101155138 M * albeiro but restart at something simple
1101155162 M * albeiro burnP6,burnBX could not lock or restart it,
1101155169 M * Bertl restart as in 'shuts down and reboots' or '*bang* was reset'
1101155171 M * albeiro do you have any clues what should i look for ?
1101155184 M * albeiro BANG ! and /me can see bios
1101155187 M * albeiro messages
1101155212 M * Bertl hmm .. there are two obvious possibilites
1101155225 M * Bertl - power management code is kicking in 
1101155237 M * Bertl - you system power is not stable enough
1101155253 M * albeiro hm...
1101155267 M * Bertl the former can be tested by disabling all power management in the bios and the kernel
1101155270 M * brc have you tried memtest86 ?
1101155282 M * albeiro yeah, memtest once restarted machine after 15 minutes
1101155290 M * Bertl the latter probably just by replacing the power supply adn/or motherboard
1101155291 M * albeiro and second time machine survived all night
1101155329 M * albeiro would lowering power consumpiton help a bit ? (by the time i  will get spare parts for testing)
1101155364 M * Bertl depends on what actually is the cause ...
1101155390 M * Bertl if it is broken power management code (either bios or kernel) that won't help a bit
1101155402 M * albeiro sure it will not
1101155422 A * mugwump shows up with sunglasses pushed up onto his forehead, a very loud shirt and a tan
1101155423 M * albeiro btw - do i need to enable highmem when using 256 MB of ram and 1 GB swap ?
1101155466 M * mugwump highmem is the per-process limit, not system VM iirc
1101155470 M * Bertl mugwump: hawaii?
1101155485 M * mugwump no, the Coromandel - I'll get a link
1101155505 M * Bertl albeiro: highmem just makes sense if you have more memory installed than can be handled by lowmem
1101155529 M * Bertl the typical split for 2.4 and 2.6 uses 3/1 where kernel gets 1GB address space
1101155551 M * Bertl so you can use up to 1GB without having himem enabled (a little less actually)
1101155553 M * mugwump http://tour.thepeninsula.co.nz/cookscoast.htm  # where I just spent my weekend
1101155589 M * Bertl looks nice ...
1101155642 M * mugwump I was there because of a wedding on the beach for a friend
1101155652 M * albeiro Bertl: thx, i will recompile my kernel carefully reconfiguring it, and will see if it helps
1101155677 M * brc mg: you should come to itacoatiara(niteroi, rio de janeiro, brazil): www.itacoatiara.com
1101155899 M * mugwump Esporte Eventos ... finalistas da categoria feminina que miran agradables
1101155940 M * Bertl albeiro: you're welcome, keep us up-to-date on that issues ...
1101156092 M * albeiro sure i will
1101156119 M * albeiro oh - btw - remember me beeing so happy with small cost repaired laptop ?
1101156137 M * albeiro it is doing the same again - still beeing broken ;p
1101156150 M * albeiro nothing comes easy it seems ;]
1101156181 M * albeiro (this laptop just locks hard from time to time, rather frequently, even without os ;p)
1101156191 M * Bertl hmm .. so it was a q&d repair then ...
1101156199 M * albeiro heh
1101156212 M * albeiro yes, both of my machines broke almost in the same time :/
1101156217 M * albeiro only router works ;]
1101156225 M * Bertl linux based?
1101156230 M * albeiro yes
1101156243 M * Bertl well, then jsut move development/production there ...
1101156251 M * Bertl ;)
1101156299 M * albeiro heh, not so easy, but managed to get my desktop (this *bang* restarting) working more or less stable, just can not put it under preasure ;]
1101156317 M * albeiro but even hell does not kow why m,y laptop hard locks
1101156335 M * Bertl har locks sound like timing issues on the bus ...
1101156348 M * Bertl how old is it?
1101156357 M * Bertl and what model?
1101156392 M * albeiro IBM A21m, old one
1101156418 M * albeiro PII 750 with 128 MB of rtam and everything on board i need (15'1 nice screen)
1101156429 M * albeiro PIII of course ;]
1101156485 M * albeiro timing issues you say... i had loaded default cmos seetings, but it can even lock up during system loading
1101156512 M * albeiro that is not pretty much i cna do to laptop, yes ?
1101156702 M * Bertl how good are you with a soldering iron and voltage/ohm meter?
1101156908 M * albeiro not bad, and i have very good in this things friends, so it is pretty ok :)
1101156956 M * albeiro (nice :-] it managed to lock up before knoppix could even show its boot menu
1101158718 Q * sim0n Quit: Client exiting
1101158756 M * Bertl redLED: still there, or already fallen asleep?
1101159528 M * redLED still here, but busy with my legal advisor
1101159531 M * redLED i'll bbl in a few
1101159561 M * Bertl np
1101159587 M * albeiro Bertl: so, if you had any clues about this strange laptop behavior share it ;]
1101159645 Q * Hollow Remote host closed the connection
1101159650 M * Bertl just asking, ebcause if you are used to such tools, I'd suggest looking for the capacitors, especially the eltrolytic ones (tantal probably in your laptop) and see if some of them have unusual impedance ...
1101159680 M * albeiro i will...
1101159743 M * Bertl typically aged tantal capacitors (and normal ones too) are responsible for changes in bus timings ...
1101159802 J * Hollow bene@home.xnull.de
1101160012 M * Bertl wb Hollow!
1101160035 M * albeiro nice tip, thx, i will check them carefully
1101160154 M * Bertl and ESR meter would be of great help, I guess ...
1101160344 M * albeiro i can look for some, have friends in place where i am studing, and since it is technical university i hope to sucess
1101160371 A * albeiro automatic control and robotics, silesian university of technology
1101160714 Q * flock Ping timeout: 480 seconds
1101160839 M * albeiro Bertl: lol - my laptop was turned off, but connected to power. and... it just had... turned itself on ! LOL
1101160876 M * Bertl maybe you should start feeding it instead of repairing it ;)
1101160885 M * albeiro rotfl :)
1101161208 Q * berni Quit: os upgrade
1101164394 J * berni berni@2001:1b18:202::2
1101164405 M * Bertl wb berni!
1101164412 M * berni thnx
1101164423 M * Bertl you are famous, you know?
1101164451 M * berni of course
1101164458 M * berni I'm the most famous guy in the universe :-)
1101164470 M * Bertl ah, okay, ... ;)
1101164804 N * Doener_zZz Doener
1101164810 M * Bertl morning Doener!
1101164820 M * Doener morning (well, kind of ;)
1101164835 M * Bertl ah, just a few days and it _is_ morning ;)
1101164970 J * rs rs@imhotep.rhapsodyk.net
1101164976 M * rs re
1101164978 M * Bertl welcome rs!
1101164984 M * rs hey bertl!
1101164989 M * rs how are you?
1101165013 M * Bertl a little tired .. but otherwise fine, and you?
1101165063 M * rs yeah fine
1101165069 M * Doener ah, ngn is making progress?
1101165077 M * Bertl yep, it is ...
1101165102 M * rs really ?
1101165107 M * rs wow cool!
1101165121 M * Bertl http://irc.13thfloor.at/LOG/
1101165368 M * rs nice,  your logger is now in production :)
1101165388 M * Bertl yep, pinky seems to behave well ...
1101165393 M * Doener hm, guess i'll start my day with q3a :)
1101165418 M * rs Doener: q3a = quake 3 arena ?
1101165429 M * Doener yep
1101165448 M * Bertl anarchy was fun!
1101165462 M * rs If you want a chalanger, I have a friend really good at this game :)
1101165501 M * rs Bertl: so you don't need my logger anymore ?
1101165535 M * Bertl the logger itself, probably no, the stats page could be fun ...
1101165554 M * Doener i just started playing again two days ago... i need to get my skills back to challenge Bertl sometime...
1101165587 M * rs Doener: Bertl is a quake player ?
1101165594 M * rs amazing...
1101165613 M * rs Bertl: what didn't you do ? :)
1101165655 M * Bertl I'm actually not the climber type ... so no mountains and such ... :)
1101165686 M * Bertl diving is another thing I do not do lightly ...
1101165737 M * rs that's all ?  :)
1101165780 M * Bertl and a milion other things I didn't do yet ;)
1101165849 M * rs about the stats system, maybe it would be simpler to make translate your log format to a well known one instead of adding the support in the statistics system :)
1101165862 M * rs s/make//
1101165880 M * Bertl yep, 'my' log format should be easily expanded to typical log formats
1101165909 M * Bertl I jsut wanted to make it lean and keep it simple
1101165937 M * rs could be nice to have a search engine too
1101165958 M * rs maybe htdig or mnogosearch would fit
1101166027 M * Bertl yeah, I'll be spending a few hours on a dynamic php frontend
1101166054 M * Bertl with some basic displaying features maybe even search, but I guess google can do a better job on that ...
1101166074 M * rs yeah right
1101166246 M * Bertl btw, you should forward your bug report to the network dev list (if not done already)
1101166246 Q * ensc Read error: Connection reset by peer
1101166270 M * rs Bertl: it was answered
1101166276 M * Bertl ah, really?
1101166290 M * Bertl let's hear/see ...
1101166292 J * ensc ircensc@ultra.csn.tu-chemnitz.de
1101166294 M * rs yeah, seems to be known issue and fixed in the 10-rc2
1101166303 M * Bertl excellent!
1101166329 M * rs I spend a part of my day to find the fix... but nothing is talking about this issue in the changelog
1101166350 M * Bertl was it answered on the netdev lsit or lkml?
1101166357 M * rs lkml
1101166402 M * Bertl i.c. Xu ;)
1101166426 M * rs you know him ? :)
1101166470 M * rs I assumed that with his first name, he shouldn't be a bad guy ;)
1101166480 M * Bertl well, not in person ...
1101166505 M * Bertl yeah, the first name is acceptable ;)
1101166585 M * rs so what is the problem with this guy ?
1101166596 M * rs is he a bullshiter ? :)
1101166689 M * Bertl no, he is fine ... and probably right .. 
1101166719 M * Bertl but the response is typical ...
1101166753 M * rs oh ok, he gives not that much info :)
1101166754 M * Bertl he's doing ipsec stuff and such ...
1101166783 M * rs because there is nothing (I can find) on the list talking about this problem
1101166829 M * rs OTOH I didn't searched into the netdev list 
1101167147 M * Bertl I guess it can't hurt to ask him what patch is relevant there ...
1101167176 M * rs already done in private
1101167190 M * Bertl no feedback yet?
1101167191 M * rs I guess I have to wait for the timezone
1101167196 M * Bertl ah, okay ...
1101167792 M * redLED ok
1101167797 M * redLED i'm back, but i'm about to drop off
1101167806 M * redLED lets try this first
1101167810 M * redLED no ipv6 module now
1101167816 M * Bertl okay ;)
1101167856 M * redLED hmm
1101167865 M * redLED i still see 127.0.0.1 on both
1101167871 M * redLED via tcpdump
1101167882 M * Bertl you see the packets?
1101167889 M * redLED yep
1101167902 M * Bertl tcpdump on one vserver, packet stuff on the other?
1101167915 M * redLED yeah
1101167921 M * matti :D
1101167924 M * redLED i *do* have CAP_NET_RAW enabled on both though
1101167930 M * redLED in case that matters
1101167939 M * Bertl okay, np, we'll continue another time ...
1101167958 M * redLED should i prepare another strace for you tomorrow first thing in the morning?
1101167972 M * Bertl no, the one you did is fine ...