1154217878 Q * meandtheshell Quit: bye bye ... 
1154218238 J * MooingLemur ~troy@shells200.pinchaser.com
1154219800 N * Bertl_oO Bertl
1154219806 M * Bertl evening folks!
1154220746 Q * KantankerousKid Ping timeout: 480 seconds
1154222955 M * lilo2 hiya Bertl!
1154223138 M * Bertl hey lilo2! still the funny name?
1154223161 M * lilo2 Bertl: yup, my main nick has been Forever Banned From OFTC 8)
1154223183 M * lilo2 Bertl: someone had a sense-of-humor deficiency on April Fool's Day *sighrin*
1154223191 M * Bertl really? that's kind of strange ... have you talked to the admins?
1154223204 M * lilo2 one of the founders banned it 8)
1154223227 M * Bertl well, bans can be removed, no? :)
1154223228 M * lilo2 apparently some latent angst transformed a 10-minute practical joke into a "denial-of-service attack"
1154223234 M * lilo2 in theory :)
1154223244 M * lilo2 but I think there are some long-standing grudges there, so maybe not
1154223270 M * Bertl the question is: ahve you talked to them lately?
1154223275 M * lilo2 I thought "surely he can't be serious about banning me forever" but apparently he is
1154223292 M * lilo2 well, I was here for about 2 years, minding my own business, and it seemed as if things were fine
1154223303 M * lilo2 (and after all the politics involved with OFTC's creation, that seemed pretty impressive)
1154223323 M * lilo2 but suddenly bing bang boom, we're back to "politics everytime I talk to someone associated with OFTC"
1154223331 M * lilo2 so after a couple of tries, I just gave up on it
1154223336 M * lilo2 it's no biggie
1154223339 M * Bertl well, you _are_ still here, which IMHO makes that a non-issue ... i.e. they do not keep you from being here ...
1154223343 M * lilo2 if anyone wants to know I'm really me, they can ask me on freenode
1154223359 M * lilo2 Bertl: yah, but I'm on Tor....they have kind of a love/hate relationship with Tor
1154223369 M * lilo2 Bertl: basically, they can't really block me
1154223376 A * lilo2 is just on OFTC for several channels
1154223381 A * lilo2 looks apologetic
1154223395 M * lilo2 (I mostly just want to get through all of the nonsense; I don't care if I don't have my own nick)
1154223451 M * Bertl okay, if you ever need a word from me (to the oftc admins, regarding the ban), then let me know (not that I think that my word has great value there :)
1154223468 M * lilo2 Bertl: appreciated....right now I just don't want to stir up any waves
1154223492 M * lilo2 I was really surprised....the joke was very public on freenode, I globaled it
1154223495 M * lilo2 it wasn't as if it was hidden
1154223509 M * lilo2 it lasted about 10 minutes here, on one channel, and as it was winding down, some OFTC admin started klining people
1154223528 M * lilo2 I guess people have trouble letting go of personal issues
1154223532 M * Bertl well, I remember, reg. nick: make sense to me .. something completely different: do you have any experience with dyndns.com, good or bad ...
1154223535 M * lilo2 anyway, I'm here, all's well
1154223545 M * lilo2 generall good, if I recall that service correctly
1154223548 M * lilo2 erm generally
1154223573 M * lilo2 fast response time
1154223575 M * lilo2 pretty easy to use
1154223585 M * Bertl okay, tx! well, enjoy your stay ... as you know, you're always welcome here :)
1154223586 M * daniel_hozac yeah, that's how i remember it too.
1154223589 M * lilo2 it seems as if it had a couple of interfaces
1154223598 M * lilo2 Bertl: thank you! :)
1154224431 M * Bertl daniel_hozac: if you are still awake :), I uploaded a few patches for 2.1.1 (mostly the lock and admin features)
1154224450 M * daniel_hozac differences from the last iteration?
1154224468 M * Bertl well, it got drastically simpler now with the new switch :)
1154224496 M * Bertl kind of frightening simple actually ...
1154224601 M * daniel_hozac lol, wow.
1154224654 M * Bertl I think this already shows that the switch change were a good choice here ...
1154224659 M * daniel_hozac yeah.
1154227264 J * KantankerousKid debian-tor@tor-irc.dnsbl.oftc.net
1154227438 Q * KantankerousKid 
1154227453 J * KantankerousKid debian-tor@tor-irc.dnsbl.oftc.net
1154228129 P * click [IRSSI]
1154229562 J * DreamerC ~dreamerc@59-112-20-64.dynamic.hinet.net
1154229615 J * Zaki_ ~Zaki@212.118.100.102
1154229982 Q * Zaki Ping timeout: 480 seconds
1154231072 M * Bertl okay, I'm off to bed now ... back tomorrow! have fun!
1154231077 N * Bertl Bertl_zZ
1154233904 J * click click@ti511110a080-0663.bb.online.no
1154238472 J * MrX ~urk@219.95.238.72
1154239757 J * _Roey ~katz@h-69-3-4-130.mclnva23.covad.net
1154240040 Q * Roey Ping timeout: 480 seconds
1154241599 J * Viper0482 ~Viper0482@p54975387.dip.t-dialin.net
1154243096 J * dna ~naucki@dialer-171-97.kielnet.net
1154245017 J * matti matti@linux.gentoo.pl
1154246014 Q * SNy Ping timeout: 480 seconds
1154246263 M * Skram_Buffalo anyone around?
1154247541 J * bonbons ~bonbons@83.222.36.236
1154248179 Q * DreamerC Quit: leaving
1154249445 N * starlein starleinn
1154250486 J * SNy 5c1d4a12a4@bmx-chemnitz.de
1154250791 J * chand ~chand@gw.net81-65-27.noos.fr
1154250816 Q * chand 
1154250820 J * chand ~chand@gw.net81-65-27.noos.fr
1154251674 J * bleh blah@201-42-109-161.dsl.telesp.net.br
1154253210 N * Skram_Buffalo Skram_Air[port]
1154254781 J * meandtheshell ~markus@85-124-206-162.dynamic.xdsl-line.inode.at
1154256869 Q * brc_ Ping timeout: 480 seconds
1154257672 J * TEKiN ~MoRYeL@88.248.166.180
1154257729 Q * TEKiN 
1154257798 J * TEKiN ~MoRYeL@88.248.166.180
1154257951 Q * locksy Ping timeout: 480 seconds
1154258070 Q * TEKiN 
1154258105 J * DreamerC ~dreamerc@59-112-27-53.dynamic.hinet.net
1154258702 Q * mcp Read error: Connection reset by peer
1154258715 J * mcp ~hightower@wolk-project.de
1154258734 J * locksy ~locksy@mrtg.sisgroup.com.au
1154259103 Q * bleh Quit: UML is making me crazy... i'm walking in the walls ...
1154260994 J * yarihm ~yarihm@84-74-17-70.dclient.hispeed.ch
1154261254 Q * coocoon Ping timeout: 480 seconds
1154261273 Q * shedi Quit: Leaving
1154261775 N * Bertl_zZ Bertl
1154261779 M * Bertl morning folks!
1154261822 M * Hollow morning Bertl!
1154261862 M * Hollow Bertl: http://vwiki.croup.de
1154261901 M * Hollow if you still have old dns entries: http://85.10.237.61/
1154261918 J * coocoon ~coocoon@p54A05423.dip.t-dialin.net
1154261921 M * Bertl ah, mediawiki :)
1154261926 M * Hollow yep :)
1154261961 M * Hollow i tamed the beast (monobook.css) ;)
1154261992 M * Bertl cool!
1154262007 M * Hollow any comments on the nav structure?
1154262068 M * Bertl well, hmm, I get 'setup the wiki first' + can't write config file :)
1154262085 M * Hollow that's why i posted the second link in case of dns troubles
1154262099 M * Hollow i wonder when they will get updated
1154262106 M * Hollow i changed it two days ago
1154262109 M * Bertl ah, even better now :)
1154262117 M * Hollow heh
1154262119 M * Hollow indeed
1154262165 M * Bertl yep, looks definitely nice ...
1154262180 M * Hollow :)
1154262209 M * Bertl I think we have to update the logo, though ((R)+(TM))
1154262230 M * Bertl btw, did you try with a full color logo? did that look strange?
1154262279 M * Hollow yeah, i lightened it a bit, so that the css colors match, otherwise the left nav bar would be too grell(?)
1154262330 M * Bertl (colorfull?) well, what I mean is, ahve the logo in normal color, but leave the sidebar as is
1154262349 M * Bertl looks a little worn out atm :)
1154262370 M * Hollow well, maybe if we change the yellow/orange a bit.. the green was ok in full color imo
1154262377 M * Hollow i can upload it in full color if you want
1154262386 M * Bertl just for test please ...
1154262390 J * mire ~mire@180-166-222-85.COOL.ADSL.VLine.Verat.NET
1154262475 M * Hollow Bertl: ok, reload
1154262519 M * Bertl hmm, didn't change here ...
1154262568 M * Hollow did you try ctrl+f5 with firefox?
1154262581 M * Hollow mediawiki has serious cache probs imo ;)
1154262587 M * Bertl well, I have galeon here, and shift-reload does the same, IMHO
1154262592 M * Hollow sometimes i had to clear my cache in firefox to reload the css
1154262606 M * Hollow or some other html/image
1154262639 M * Hollow http://85.10.237.61/skins/common/images/wiki.png
1154262659 M * Bertl pastell
1154262665 M * Hollow wtf
1154262686 M * Bertl now it was updated
1154262713 M * Hollow i know why i hate web development :p
1154262730 M * Hollow it's just unpredictable ;)
1154262740 M * Bertl okay, to complicate this once again, could you set the side bar color to the same yellow the logo uses?
1154262791 M * Hollow try to reload
1154262809 M * Hollow (horrible)
1154262810 M * Hollow :)
1154262823 M * Bertl yeah, well, a little extreme, I ahve to play with that a little
1154262832 M * Bertl but the color logo looks much better
1154262851 M * doener Hollow: regarding dns, it might help if your ns2 and ns3 had the same idea about it as ns1 ;)
1154262855 M * Hollow yeah, maybe if you tune both css + logo a bit, it can look great ;)
1154262866 M * Hollow doener: ic.. ;)
1154262875 M * Hollow damn replication
1154262888 M * Bertl nah, synchronizing dns is for losers :)
1154262925 M * Hollow Bertl: you can play with it on your own if you want.. 
1154262940 M * doener heh, poor man's dns rr at its best, eh?
1154262972 M * Bertl Hollow: many thanks for the excellent example page, I will think a little about the colors and make a suggestion
1154262988 M * Hollow ok
1154263503 Q * chand Quit: chand
1154263733 J * chand ~chand@gw.net81-65-27.noos.fr
1154263761 Q * KantankerousKid Ping timeout: 480 seconds
1154263976 M * harry matti: yes
1154264076 J * brc_ bruce@201.79.112.195
1154264323 M * Bertl welcome brc_!
1154264467 Q * chand Quit: chand
1154266198 M * Bertl Hollow: could you try with #f7eccd for the sidebar please?
1154266226 M * Hollow ok, reload
1154266304 M * Bertl good, and maybe #d1c2ad for the 'top' line?
1154266341 M * Bertl (the gray one), but probably #e0ceb5 is a little better
1154266376 M * Hollow reload again
1154266411 M * Bertl ah, sidebar is off by one pixel (too high) here, how does that look for you?
1154266456 M * Hollow yeah, this is most likely because of font-size
1154266467 M * Hollow will fix this up later
1154266495 M * Bertl okay, np, the sidebar could use a little more red, IMHO, so maybe make that one #faeccd
1154266513 M * Bertl but it already looks quite nice to me, what's your opinion?
1154266534 M * Hollow hm, i like the former better
1154266538 M * Hollow but look yourself
1154266564 M * Bertl yeah, was a little too much, maybe somewhere inbetween
1154266564 Q * DreamerC Quit: leaving
1154266569 M * Bertl #f9eccd
1154266640 M * Hollow yup, looks ok
1154266834 M * Hollow what about the font? mediawiki defaults to sans-serif, i think we should explicitly define one too though
1154267013 M * Bertl well, sans-serif is fine, we can select helvetica if we like to
1154267306 J * DreamerC ~dreamerc@59-112-27-53.dynamic.hinet.net
1154267539 Q * FaUl Ping timeout: 480 seconds
1154267576 M * matti Bertl, harry, Hollow: :-)
1154269053 M * Hollow Bertl: anything else regarding color/font/layout?
1154269086 M * Hollow hey matti, btw ;)
1154269090 M * Bertl not at the moment, but I'm sure I will have some ideas on the way .. in general it looks quite nice to me
1154269122 M * Hollow you planned to rewrite the paper, right?
1154269134 M * Bertl yup
1154269194 M * Hollow i guess we need to form a wiki team to transfer all the content ;)
1154269201 M * Hollow anybody?
1154269228 M * Bertl btw, did you setup the media wiki in a linux-vserver guest?
1154269233 M * Hollow sure
1154269253 M * Bertl excellent, so we should have no problem to transfer that once it has been finished
1154269266 M * Hollow indeed
1154269321 M * Bertl Hollow: did you receive the nfs question from Martin? if so, could you bounce it to me?
1154269376 M * Hollow ah damnit
1154269389 M * Hollow i answered it instead of fwd
1154269400 M * Hollow fortunately to your address
1154269403 M * Hollow ;)
1154269461 M * Bertl np, it is now digging through my greylist filter  ...
1154269490 M * Hollow ok, bounced now
1154269549 M * Bertl do you have a moment for the details of the limit and accounting read commands?
1154269630 Q * matti Ping timeout: 480 seconds
1154269682 M * Hollow Bertl: sure
1154269709 M * Bertl I'm not 100% sure where to put them, i.e. in which category
1154269750 M * Bertl I'm inclined to move them to cat 40 (Stats/info/other
1154269777 M * Hollow yeah, that would have been my suggestion as well ;)
1154269780 J * matti matti@linux.gentoo.pl
1154269789 M * Bertl okay, sounds like a plan then ...
1154269876 M * Bertl I'll call the category VSTAT and use that for both of them
1154269963 M * Bertl and we better bump the VCI revision, yes?
1154270057 Q * michal` Ping timeout: 480 seconds
1154270076 M * Hollow Bertl: yep
1154270179 M * Bertl http://paste.linux-vserver.org/214
1154270195 M * Bertl does that look okay to you?
1154270214 M * Bertl ah, missing the #hits :)
1154270266 M * Bertl http://paste.linux-vserver.org/215
1154270275 J * KantankerousKid ~blah@tor-irc.dnsbl.oftc.net
1154270318 J * michal` ~michal@www.rsbac.org
1154270363 M * Hollow yup, looks good
1154270385 M * Bertl we probably need some extension to the reset_minmax
1154270413 M * Bertl first to reset the hits count too and second to report back the interval, no?
1154270514 M * Hollow ehm..?
1154270551 M * Bertl ehm, regarding hits? or regarding interval :)
1154270566 M * Hollow the interval.. you mean return the time since the last reset?
1154270585 M * Bertl yep, or the tick value of the last reset
1154270596 M * Bertl but probably 'uptime' in msec is best
1154270615 M * Hollow hm, where do i need this time?
1154270639 M * Bertl for example for statistics like hits/s or so
1154270713 M * Hollow ok.. we did not record any of the hits values until now
1154270759 Q * tdjb Ping timeout: 480 seconds
1154271591 Q * s0undt3ch Ping timeout: 480 seconds
1154272374 M * Bertl hmm, we have to change the 'current' to 'value', otherwise the clashes are too many :)
1154272442 M * Hollow heh
1154272449 M * doener :)
1154272880 M * matti harry: I just wonder, how much chroot restrictions from grsecurity can hurt vserver?
1154272931 M * Hollow doener: does vwiki.croup.de point to 85.10.237.61 now?
1154273050 M * Bertl vwiki.croup.de.		75266	IN	A	85.10.237.58
1154273072 M * Hollow well, now i probably still need to wait 24h ;)
1154273076 M * Bertl ns2 is wrong
1154273099 M * Bertl do you ahve control over the nameservers?
1154273106 M * Hollow yes
1154273114 M * Hollow the mysql replication somehow freaked out
1154273121 M * Bertl mysql?
1154273140 M * Hollow mydns
1154273155 M * Bertl interesting ... more than 1000 domains :)
1154273165 M * Hollow hm?
1154273199 M * Bertl well, I do not see much point in having a database for dns (for only a few domains)
1154273215 M * Bertl i.e. file based domain tables are much easier to handle here, no?
1154273262 M * Bertl and if you configure two of the servers as slaves, they should get a notification when the master is updated
1154273266 M * Hollow hm, i don't think so
1154273270 M * Hollow i like the mysql approach
1154273281 M * Bertl okay, it's fine with me :)
1154273301 M * Hollow i just changed default collation of one server, Bad Thing (TM)
1154273511 M * doener Hollow: ns1 and ns3 give me .61, ns2 gives me .58
1154273525 M * doener oh, should have read further than just the question...
1154273526 M * Hollow yeah.. i'm working on it ;)
1154273889 M * bonbons Hollow: for me they now all 3 announce .61 for vwiki
1154273949 M * bonbons Bertl: did you already hear from undead bind? that makes replication fun!
1154273991 M * Bertl okay, here are some steps on the way, please check and comment ...
1154274003 M * Bertl http://vserver.13thfloor.at/Experimental/delta-cacct-clean01.diff
1154274009 M * Bertl http://vserver.13thfloor.at/Experimental/delta-vstat-feat01.diff
1154274023 M * Bertl will be back in a few hours ...
1154274039 M * Bertl bonbons: undead bind?
1154274070 N * Bertl Bertl_oO
1154274071 M * bonbons yep, bind that partially started, say taking tcp socket and then doing nothing more
1154274109 M * Hollow bonbons: yep, i fixed it at this very moment ;)
1154274118 M * bonbons then you put second alive version (which doesn't bind tcp) and bye bye zone transfers
1154275260 Q * KantankerousKid Ping timeout: 480 seconds
1154276373 M * h01ger micah, in http://deb.riseup.net/vserver/usage/ in "Howto impose disk limits in each vserver" a script /etc/vservers/servername/scripts/post-start is described. should i put vdlimit-command in such a script and execute each time a vserver is started??
1154277207 J * Pazzo ~thomas@host130-250-static.72-81-b.business.telecomitalia.it
1154277964 M * daniel_hozac h01ger: that's old information.
1154277977 M * daniel_hozac 0.30.210 supports disk limits in the configuration, see the flower page.
1154278132 M * h01ger me has 0.30.204-5sarge3
1154278138 A * h01ger sighs
1154278150 M * daniel_hozac well, that's what you get for running sarge :)
1154278179 M * h01ger thats from backports.org
1154278183 M * h01ger (but still sarge)
1154278198 M * daniel_hozac bp.o has 0.30.210.
1154278236 M * h01ger right, 0.30.210-8bpo2 :)
1154278245 A * h01ger read apt-cache show output wrongly
1154278257 M * h01ger so i'll read the flower page again now :)
1154278261 M * h01ger thanx daniel_hozac 
1154278271 A * h01ger prepares to dive into flower power land
1154278276 A * h01ger looks for his pipe
1154278737 M * h01ger so /etc/vservers/vserver-name/dlimits is what i want..
1154278787 M * daniel_hozac yep.
1154279294 J * KantankerousKid debian-tor@tor-irc.dnsbl.oftc.net
1154280281 M * micah h01ger: thanks for pointing that old info out, i need to remove that
1154280369 Q * yarihm Read error: Connection reset by peer
1154280405 J * yarihm ~yarihm@84-74-17-70.dclient.hispeed.ch
1154281531 M * h01ger micah, denada :)
1154281763 N * Skram_Air[port] Skram
1154283747 M * matti :)
1154283862 J * chand ~chand@gw.net81-65-27.noos.fr
1154285365 Q * Viper0482 Remote host closed the connection
1154285764 J * _mcp ~hightower@wolk-project.de
1154285846 J * dna_ ~naucki@dialer-171-97.kielnet.net
1154285900 Q * michal` iridium.oftc.net nova.oftc.net
1154285900 Q * matti iridium.oftc.net nova.oftc.net
1154285900 Q * brc_ iridium.oftc.net nova.oftc.net
1154285900 Q * coocoon iridium.oftc.net nova.oftc.net
1154285900 Q * mcp iridium.oftc.net nova.oftc.net
1154285900 Q * dna iridium.oftc.net nova.oftc.net
1154285900 Q * Zaki_ iridium.oftc.net nova.oftc.net
1154285900 Q * meebey iridium.oftc.net nova.oftc.net
1154285900 Q * insomniac iridium.oftc.net nova.oftc.net
1154285900 Q * pagano iridium.oftc.net nova.oftc.net
1154285900 Q * kir iridium.oftc.net nova.oftc.net
1154285900 Q * Skram iridium.oftc.net nova.oftc.net
1154285900 Q * samuel_ iridium.oftc.net nova.oftc.net
1154285900 Q * m4z iridium.oftc.net nova.oftc.net
1154285900 N * _mcp mcp
1154286024 J * michal` ~michal@www.rsbac.org
1154286044 J * Skram ~MarkS@admins.sentiensystems.net
1154286048 J * meebey meebey@booster.qnetp.net
1154286052 J * matti matti@linux.gentoo.pl
1154286058 J * alexx ~alexx@proxy.ikse.net
1154286068 J * insomniac ~insomniac@slackware.it
1154286068 Q * alexx 
1154286878 Q * KantankerousKid 
1154286926 J * pagano ~pagano@131.154.5.20
1154286942 J * Viper0482 ~Viper0482@p54975387.dip.t-dialin.net
1154286946 J * coocoon ~coocoon@p54A05423.dip.t-dialin.net
1154286954 J * kir ~kir@swsoft-mipt-nat.sw.ru
1154286989 J * Zaki_ ~Zaki@212.118.100.102
1154287351 Q * mire Quit: Leaving
1154288522 J * x03 ~x03@10001268031.0000030345.acesso.oni.pt
1154288547 M * x03 hi all
1154288559 Q * Viper0482 Remote host closed the connection
1154288689 M * bonbons Bertl_oO: did you aready do some work for porting proc to seq_file? What is the size of the buffer for current code?
1154288802 J * bubulak ~bubulak@whisky.pendo.sk
1154290155 M * h01ger whats wrong with this dlimits: http://paste.debian.net/9925 ? the directory is the right one... 
1154290162 M * h01ger do i need to do something to enable this?
1154290164 Q * meandtheshell Read error: Connection reset by peer
1154290624 J * brc_ bruce@201.79.112.195
1154290726 N * virtuoso_ virtuoso
1154290786 Q * virtuoso Killed (services.oftc.net (Too many invalid passwords))
1154290797 J * virtuoso ~s0t0na@80.253.205.251
1154291792 Q * yarihm Quit: Leaving
1154292148 M * daniel_hozac bonbons: a page, i think.
1154292193 M * daniel_hozac h01ger: you're lacking the index directory. those files should be in dlimits/0.
1154292199 M * bonbons daniel_hozac: that's what I guess :) but nowhere we ever check whatwe write there...
1154292216 M * daniel_hozac bonbons: well, we know how long all the files are.
1154292277 M * bonbons possibly harmful for those with e.g. many interfaces/ip-addresses in a context (ok, they still have to adjust define for it)
1154292302 M * daniel_hozac true, and moving to a dynamic allocation method will require some work in that area.
1154292366 M * bonbons yep... seq_file should handle the buffer size behind the scenes if I understood it right
1154292830 M * bonbons daniel_hozac: do you know if Bertl_oO started some work in seq_file direction?
1154293228 J * blues blues@blysk.ds.pg.gda.pl
1154293274 M * blues hello
1154293294 M * blues I have some serious problem with system running vserver
1154293342 M * cehteh that is?
1154293355 M * blues my system hangs with "out of memory"
1154293372 M * blues "kerne panic - no killable processes found
1154293378 J * samuel_ ~samuel@jupe.quebectelephone.com
1154293388 M * blues it started when I started to play with vservers
1154293400 M * cehteh well then check where which and why the memory is exhausted
1154293407 M * cehteh did you set ulimits?
1154293409 J * m4z m4z@bastard-operator.from-hell.net
1154293434 M * bonbons blues: what kernel version?
1154293462 M * blues bonbons: currently 2.6.16.23
1154293486 M * bonbons what vserver patch on it?
1154293486 M * blues bonbons: I've tried many previous 2.6.14-2.6.16
1154293492 M * blues bonbons: moment
1154293497 M * blues some 2.1
1154293564 M * blues looks like it was patch-2.6.16-vs2.1.1-rc15
1154293648 M * blues was problem like that reported previously?
1154293682 M * blues The system freezes for good - even Sysrq doesn't work
1154293714 M * blues Only once I've catched something in logs - I can post it
1154293738 M * h01ger daniel_hozac, yay! works! thx :)
1154293785 M * bonbons blues: would be useful to have some info like what used the memory, and what triggers the starvation
1154293819 M * blues bonbons: I don't have any idea to figure what eats the memory
1154293849 M * blues aaa - one more thing may be usefull - it's SMP system with 4xPIII Xeon
1154293862 M * bonbons top when system dies, cat /proc/memory and the like
1154293902 P * mnemoc 
1154293904 M * blues bonbons: there is no time. When it happens I can't do anything
1154293909 M * blues bonbons: even login
1154293923 J * Aiken ~james@tooax6-133.dialup.optusnet.com.au
1154293977 Q * dna_ Quit: Verlassend
1154293996 M * blues bonbons: and it happens ~1-2 times per week
1154294013 M * h01ger daniel_hozac, it works, as in df displays a limited filesystem size, but df also displays the number of used blocks incorrectly..
1154294020 M * blues bonbons: I can't sit and wait for that....
1154294023 M * bonbons blues: keep a session open that permanently outputs that info (while true; do cat /proc/mem; done)
1154294024 A * h01ger goes to bed
1154294044 M * bonbons on screen you should be able to read the most recent outputs before crash
1154294093 N * Bertl_oO Bertl
1154294097 M * Bertl evening folks!
1154294099 M * blues bonbons: I've tried such a things. The sessions get broken without any info on screen. It gets too quickly...
1154294129 M * blues bonbons: only once I get ANY oops on system console
1154294144 M * blues bonbons: on screen
1154294170 M * blues the system freezes completely.
1154294189 M * Bertl blues: how much swap space does your system have?
1154294216 M * blues (aaaa - the hardware is ok. I have few identical boxes and I've replaced/tested them)
1154294227 M * blues 1G RAM and 1G swap
1154294241 M * Bertl and all identical boxes hang?
1154294269 M * blues Bertl: yes
1154294346 M * cehteh so you have no limits on memory usage on the vservers?
1154294377 M * blues cehteh: I have limits
1154294386 M * cehteh mhm
1154294438 M * blues cehteh: is any problem with that?
1154294440 M * cehteh well you can set them lower so that the offending program gets killed earlier mabye
1154294461 M * cehteh dunno .. i would say you have to track down what causes the hang
1154294480 M * blues cehteh: I counted on some help with that... :-/
1154294509 M * Bertl blues: yeah, you can count on that :) haven't got any reports in this direction so far
1154294526 M * Bertl blues: so we have to narrow down what causes this 'unusual' behaviour
1154294534 M * blues I don't have any ideas what to do with that 
1154294545 M * cehteh using some monitor logging the 5 programs which use the most memory to syslog every minute or such
1154294550 M * blues I've spend a lot of time trying to track this...
1154294571 M * Bertl first, try to give me a good idea what programs are running inside the guests (and on the hosts)
1154294576 M * derjohn Bertl, what is highest possible XID?
1154294593 M * Bertl 49151 (for now) 65535 for dynamic
1154294625 M * derjohn hm, any plans to reactivate dynamic xid-ing ?
1154294668 M * blues Bertl: there are 3 vservers - one is icecast, second is small system with lighttpd and some really small services
1154294668 M * Bertl no, will be done completely in userspace in the near future
1154294693 M * Bertl (that was regarding the dynamic xids)
1154294717 M * derjohn Bertl, you mean the 'allocation' of the XID will be done by util-vserver? nice idea.
1154294735 M * Bertl vserver-utils or util-vserver or ... :)
1154294738 M * blues Bertl: third is complete system with postfix, apache, courier and few other things
1154294794 M * derjohn Bertl, well the numbers are more or less internal numbers as we have guest names (e.g. ovz uses the number afair), Who is working on the userspace stuff?
1154294810 M * blues Bertl: on host system is running apache, postfix - this is because I'm moving it to new vserver on-line.
1154294827 M * blues Bertl: and thats all...
1154294831 M * Bertl derjohn: IIRC hollow has already some kind of dynamic userspace id assignment running
1154294868 M * derjohn Bertl, ok. I expected that answer ;)
1154294961 M * Bertl blues: okay, I'm currently looking at the code which selects a process to be killed
1154294973 M * Bertl mm/oom_kill.c ~200
1154294991 M * Bertl there are a few cases where a task does not qualify for oom killing
1154295002 M * derjohn Bertl, argh! rc27! changes?
1154295014 M * blues Bertl: the most interesting thing is that there way on console "no killable processes"
1154295020 M * Bertl blues: in your case, the search ends without finding _any_ process
1154295022 M * blues s/way/was/
1154295035 M * Bertl yes, I backtracked from there
1154295058 M * Bertl so, there is one process which will not be touched, the init process
1154295074 M * Bertl also processes with OOM_DISABLE set will not be selected
1154295107 M * Bertl then non overlaping cpusets (this is new in recent kernels) will not qualify
1154295156 M * Bertl finally we have two special cases, which seem a little complicated but could as well cause your issues
1154295165 M * blues I don't use cpusets (yet - I wanted to play with that)
1154295184 J * meandtheshell ~markus@85-124-206-162.dynamic.xdsl-line.inode.at
1154295264 M * Bertl but the special cases are handled special in the return too, and do not produce the beforementioned message
1154295291 M * Bertl derjohn: should be in the irc logs ...
1154295331 M * Bertl blues: so ... as you are not using cpusets, it looks to me like all processes have OOM_DISABLE set somehow
1154295378 M * Bertl blues: what does /proc/<pid>/oom_adj show for some of the more memory intensive apps?
1154295499 M * blues Bertl: # cat /proc/*/oom_adj | grep -v 0
1154295501 M * blues #
1154295572 M * Bertl looks definitely strange
1154295619 M * Bertl first, I'd say, whatever it is, it seems to be a mainline issue (because we do not modify the search per se)
1154295647 M * daniel_hozac derjohn: potential deadlock fix, some minor cleanup.
1154295670 M * Bertl blues: i.e. I'd expect your issues to occur without the linux-vserver patches (if you manage to run as many threads)
1154295719 M * Bertl blues: second, I don't think we can get more insight without adding some kind of additional debug code to the killer (which dumps some task states or so)
1154295722 M * blues Bertl: The problem is that I can't figure reproducable test-case
1154295724 M * blues :(
1154295726 M * derjohn daniel_hozac, thx, I was not able to find this in the chatlogs yet via google. do both changes go to devel and stable?
1154295735 M * daniel_hozac derjohn: yep.
1154295794 M * Bertl blues: the machines are SMP not NUMA, yes?
1154295802 M * blues Bertl: SMP.
1154295913 M * blues Bertl: 4xi686 Pentium_III_(Cascades) Xeon
1154295993 M * blues Bertl: so - what do you suggest?
1154296012 M * Bertl how many systems do you ahve in this setup?
1154296067 M * blues Bertl: one? :D
1154296096 M * Bertl aha :) well, you were talking of several identical systems :)
1154296125 M * blues Bertl: _boxes_ :)
1154296160 M * blues Bertl: I have few identical hardware boxes - I could use them to replace working one
1154296191 M * Bertl you said, once a week that this happens?
1154296291 M * blues Bertl: ~1-2 per week.
1154296319 M * Bertl well, that could be sufficient to get some clues
1154296330 M * Bertl so I'd suggest the following:
1154296359 M * Bertl - get a serial console (with 24/7 logging)
1154296384 M * Bertl - enable the existing OOM debug stuff
1154296404 M * Bertl   (#define DEBUG in mm/oom_kill.c)
1154296420 M * Bertl - enable history tracing and debug info (just to make sure)
1154296448 M * Bertl - have vmstat running remotely 
1154296481 M * Bertl then you just wait ... already the first 'hang' should give some clues
1154296495 M * Bertl (probably it's a good idea to enable the nmi watchdog too)
1154296587 M * blues Bertl: well....
1154296599 M * blues Bertl: I have watchdog there :)
1154296623 M * Bertl okay, then you will get an oops via serial
1154296650 M * blues Bertl: tried to run serial console - the result was the same like on the regular screen
1154296680 M * daniel_hozac the watchdog should panic if the system hangs.
1154296680 M * blues Bertl: only info about the "no killable procesess"
1154296716 M * Bertl if nmi oopser or some hardware watchdog is enabled, you will get an oops via serial, or your hardware locks up (which I do not think is the case here)
1154296749 M * Bertl but the debug from the oom_kill will spew a lot of info before that oops
1154296769 M * blues Bertl: hmmm... Sysrq doesn't work...
1154296792 M * blues yes, thats true
1154296862 M * Bertl theoretically it could be that your init process leaks memory, and it is the only process left after half a week (and the oom killer's run :)
1154296911 M * Bertl nevertheless, if that's the case, we'll see that in the debug output too
1154297005 Q * m4z iridium.oftc.net nova.oftc.net
1154297069 M * blues hem... theoretically you're right, but it's quite not likely... :)
1154297105 M * blues Bertl: I think that I would see it on other machines...
1154297224 Q * chand Quit: chand
1154297302 J * m4z m4z@bastard-operator.from-hell.net
1154297633 M * blues Bertl: aaa - I forgot... here is oops I catched: http://www.blues.gda.pl/kernel-29.07.2006.log
1154297651 M * blues if it helps...
1154297756 M * blues (my dogs need to go out - I have to go with them...)
1154299171 M * Bertl okay, off for now .. back later ...
1154299177 N * Bertl Bertl_oO
1154300596 Q * meandtheshell Quit: bye bye ... 
1154301033 Q * bonbons Quit: Leaving
1154302405 Q * coocoon Quit: KVIrc 3.2.0 'Realia'