1346891011 J * clopez ~clopez@17.28.165.83.dynamic.mundo-r.com 1346891488 Q * Aiken Remote host closed the connection 1346891887 Q * click Ping timeout: 480 seconds 1346893410 Q * sannes Remote host closed the connection 1346896349 Q * nlm_ Ping timeout: 480 seconds 1346896673 M * Bertl_oO off to bed now ... have a good one everyone! 1346896678 N * Bertl_oO Bertl_zZ 1346897732 Q * clopez Ping timeout: 480 seconds 1346898296 J * click click@ice.vcon.no 1346901221 Q * nkukard Ping timeout: 480 seconds 1346901248 J * nkukard ~nkukard@41-133-165-27.dsl.mweb.co.za 1346902958 J * ghislain ~AQUEOS@adsl2.aqueos.com 1346903252 Q * nkukard Read error: Operation timed out 1346904187 J * nkukard ~nkukard@41-133-165-27.dsl.mweb.co.za 1346909396 Q * PowerKe Read error: Operation timed out 1346909430 Q * dkg Ping timeout: 480 seconds 1346909938 J * PowerKe ~tom@94-226-105-17.access.telenet.be 1346913652 J * dkg ~dkg@lair.fifthhorseman.net 1346915452 Q * nkukard Read error: Operation timed out 1346918596 J * nakacya ~nakacya@s1106124.xgsspn.imtp.tachikawa.spmode.ne.jp 1346920994 J * fisted_ ~fisted@xdsl-84-44-224-5.netcologne.de 1346921315 Q * fisted Ping timeout: 480 seconds 1346921536 Q * nakacya Ping timeout: 480 seconds 1346921924 J * nakacya ~nakacya@s1106149.xgsspn.imtp.tachikawa.spmode.ne.jp 1346922093 N * Bertl_zZ Bertl 1346922098 M * Bertl morning folks! 1346922133 M * uranus moin Bertl 1346922156 M * uranus Bertl, did you read my posts from yesterday? about the rcu lock? 1346922214 M * Bertl yup 1346922336 M * uranus do you see problems if the lock is unlocked before the fetch of the cache value? 1346922339 M * Bertl but I'm not sure what you changed how and what the last paste means? 1346922368 M * uranus i moved rcu_read_unlock(); in front of cache = mem_cgroup_stat_read_cache(mcg); 1346922387 M * uranus your patch makes it the other way rcu_read_unlock() is behind cache = mem_cgroup_stat_read_cache(mcg); 1346922430 M * Bertl well, that I see, but your version definitely leaves it locked when mcg = NULL 1346922435 M * Bertl (which is wrong) 1346922451 M * uranus oh i see 1346922473 M * Bertl but I wonder how you got there 1346922488 M * uranus so maybe in that way? http://paste.linux-vserver.org/22932 1346922537 M * Bertl that would work, not sure if that isn't just papering over the issue though 1346922538 M * uranus well I got lots of more traces and these traces showed all that the double locking was after mem_cgroup_stat_read_cache was called 1346922563 M * Bertl okay, do you have them somewhere? 1346922574 M * uranus just have to look at the last logs 1346922596 M * Bertl 22930? 1346922686 M * Bertl let's assume mem_cgroup_stat_read_cache() causes a problem because it takes the rcu_lock somewhere 1346922690 M * uranus 22930 was my test script to get the trace with only 2 guests on a test system, where the guests have a very small memory limit 1346922720 M * Bertl now what I wonder is, where does it actually end up taking the rcu lock? 1346922751 M * uranus http://paste.linux-vserver.org/22933 1346922754 M * Bertl don't get me wrong, I'm not against the changes, probably we'll do similar in other places as well 1346922788 M * uranus mem_cgroup_read_stat calls get_online_cpus and there the lock issue happens in my opinion 1346922791 M * Bertl ah, in get_online_cpus? 1346922856 M * Bertl okay, and probly similar when CONFIG_HOTPLUG_CPU is enabled 1346922872 M * uranus that's enabled in my kernel 1346922883 M * uranus addr2line points to kernel/cpu.c:67 1346922890 M * uranus address ffffffff8103e0d4 1346923009 J * nkukard ~nkukard@41-133-165-27.dsl.mweb.co.za 1346923059 M * uranus Bertl, i'll be back in around 1 hour if you have more questions 1346923089 M * Bertl okay, I'll take a deeper look at it and see what I can come up with 1346923095 M * Bertl you can test a new patch then? 1346923172 M * uranus yes 1346923257 M * uranus i hope that because of this bug some other strange things get resolved which seems to be also locking issue (processes in state D or R) 1346924270 J * BlackPanx 4d6f0224@ircip1.mibbit.com 1346924291 M * Bertl daniel_hozac: ping? 1346924315 M * BlackPanx can server freeze becouse of packet/io queues ? 1346924347 M * BlackPanx do these queues go on disk or are they stored in memory ? 1346924389 M * Bertl define packet/io queues 1346924393 M * BlackPanx is it possible that server becomes unresponsive becouse of too many connections to it ? doesnt accept new connections, but reponds to ping 1346924418 M * BlackPanx well i happen to hit one limit in lighttpd, and all of the sudden server became unresponsive 1346924424 M * Bertl ping is not a good test on linux systems, it usually works even if the machine is already crashed 1346924448 M * Bertl hit one limit means? 1346924461 M * BlackPanx lighttpd has 1024 limit of requests per second 1346924487 M * BlackPanx i had one problem and forgot, so all requests weren't balanced, instead went to one server 1346924489 M * Bertl and you achieve 1024 requests per second? 1346924489 M * BlackPanx and it was fine 1346924508 Q * nkukard Read error: Operation timed out 1346924518 M * BlackPanx i did it yesterday, and it was fine, until it hit 1024 requests per second at 9:00 this morning 1346924523 M * BlackPanx when server became unresponsive 1346924527 M * BlackPanx all of the sudden 1346924538 M * BlackPanx but that was guest 1346924565 M * Bertl the question to me is: did it hit the limit and then got unresponsive or did it get unresponsive and thus hit the limit 1346924600 M * Bertl anything in the host logs of that machine? 1346924773 M * BlackPanx no error in host log. 1346924777 M * BlackPanx seems like server was fine 1346924780 M * BlackPanx but very overloaded 1346924789 M * BlackPanx or at least hit limit where he couldnt accept new connections 1346924806 M * BlackPanx cause too many were established 1346924819 M * BlackPanx i'll investigate a bit more 1346924820 M * BlackPanx and get back 1346925246 M * Bertl okay, check for memory and paging/swapping 1346925275 M * Bertl if the machine goes over the physical memory and starts swapping, it will get quite slow and thus seem completely unresponsive 1346925283 M * Bertl (compared to normal operation) 1346925517 J * nkukard ~nkukard@41-133-165-27.dsl.mweb.co.za 1346925648 J * clopez ~clopez@fanzine.igalia.com 1346926503 M * uranus Bertl, how long do you need for the testing patch? 1346926580 M * Bertl I'd like to chat with daniel_hozac first, because I'd like to get his opinion 1346926600 M * Bertl after that it shouldn't take too long (maybe 20 minutes) 1346926739 M * uranus thx 1346926921 Q * clopez Quit: Leaving 1346927011 Q * Romster Quit: Geeks shall inherit properties and methods of object earth. 1346927714 Q * vspas Read error: Connection reset by peer 1346928212 Q * nakacya Remote host closed the connection 1346929941 J * arekm ~arekm@ixion.pld-linux.org 1346930877 Q * arekm Quit: leaving 1346931240 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com 1346935263 J * BenG ~bengreen@cpc10-aztw24-2-0-cust114.aztw.cable.virginmedia.com 1346935811 Q * Romster Read error: Connection reset by peer 1346937881 Q * ensc|w Remote host closed the connection 1346937892 J * ensc|w ~ensc@www.sigma-chemnitz.de 1346937963 Q * BlackPanx Quit: http://www.mibbit.com ajax IRC Client 1346937993 M * daniel_hozac Bertl: pong 1346939067 M * daniel_hozac (unpong, internet here is too bad to use) 1346939718 M * Bertl okay :) 1346939762 M * Bertl well basically my thought is the following: when we request guest specific memory/swap info, the process doing those requests should keep the memcg alive 1346939802 M * Bertl so I think it should be safe to not 'get' the memcg for proc related adjustments, and instead retrieve it under rcu lock and then just use it 1346939885 M * Bertl I'll take a nap, if you have some comments/ideas/opinions (and the network means to share them :), just let me know 1346939903 M * Bertl off for a nap ... bbl 1346939907 N * Bertl Bertl_zZ 1346941191 J * Romster ~romster@202.168.100.149.dynamic.rev.eftel.com 1346942672 J * nlm_ ~nlm@host77.190-30-39.telecom.net.ar 1346942981 Q * BenG Quit: I Leave 1346944780 J * nakacya ~nakacya@KD118152083243.ppp-bb.dion.ne.jp 1346947417 J * bonbons ~bonbons@2001:960:7ab:0:9826:5b3a:c9f4:228b 1346950990 J * clopez ~clopez@fanzine.igalia.com 1346951274 J * arekm ~arekm@ixion.pld-linux.org 1346951646 Q * clopez Quit: Leaving 1346951679 J * clopez ~clopez@fanzine.igalia.com 1346952342 Q * clopez Quit: Leaving 1346952373 J * clopez ~clopez@fanzine.igalia.com 1346952444 Q * clopez Remote host closed the connection 1346952505 J * clopez ~clopez@fanzine.igalia.com 1346952599 Q * clopez 1346952630 J * clopez ~clopez@fanzine.igalia.com 1346952677 N * Bertl_zZ Bertl 1346952687 M * Bertl clopez: testing something? 1346952773 M * clopez yep... some irc scripts.. sorry for the continuous re-connecting notices 1346955457 J * sannes ~ace@cm-84.211.87.28.getinternet.no 1346958204 N * ensc Guest6029 1346958213 J * ensc ~irc-ensc@84.173.235.201 1346958621 Q * Guest6029 Ping timeout: 480 seconds 1346959451 M * nlm_ hi there, I have an issue that I saw recently on this IRC too. all processes has an open fd (fd 3) linking to the root mountpoint (the mountpoint outside the chroot I mean). using kernel 3.4.5-vs2.3.3.6 and utils 0.30.216-pre3034 1346959626 Q * clopez Ping timeout: 480 seconds 1346959755 M * Bertl IIRC, daniel_hozac was going to fix that in util-vserver (probably it is already fixed in svn) 1346959781 M * nlm_ cool, I'll update now. thanks 1346960164 M * nlm_ could you give me the url of the svn? I only found a git repo on github but that was last updated 5 months ago 1346960285 M * nlm_ I see you dropped SVN some time ago, git.linux-vserver.org also shows latest commit 4months ago 1346960482 M * Bertl please check with daniel_hozac where he keeps his development branch 1346960502 M * Bertl I remeber that the info was available on the wiki, but I don't see it right now 1346960559 M * Bertl (and please add it to the wiki once you find it :) 1346960585 M * nlm_ I'll search the oldwiki and port it if I find it 1346960879 J * kermie 18224ae4@ircip4.mibbit.com 1346960963 M * kermie Hi - I'm having some trouble with sound in my test Vserver (I'm using x2go to connect, and am running Debian Squeeze on both guest and host). I can connect, start firefox, go to Pandora, but the sound comes out of the host,not the guest. If I play an MP3 on the guest, that sound comes out of the guest. 1346961032 M * Bertl sounds interesting (pun intended :) 1346961057 M * Bertl what do you mean by 'the sound comes out of the guest'? 1346961239 M * kermie Badly phased on my part....I have a netbook using x2go that connects to a guest vserver on a host machine. My goal is to get sound to come out of the netbook when connected to the guest vserver. Using x2go, looking at the netbook, I can double click on an MP3 that is on the guest Vserver and the sound comes out of the netbook. When I do Youtube or Pandora, 1346961264 M * kermie the sound comes out of the speakers attached to the box that is the "host" to the "guest" vserver. 1346961288 M * Bertl okay, now I got it ... I presume it is using different sound systems 1346961301 M * Bertl e.g. pulseaudio vs. direct alsa 1346961328 M * Bertl and one is configured for network audio, the other is directly patched through to the sound device 1346961370 M * Bertl you might want to check the filedescriptors/sound settings but I'm not sure you have much luck with flash plugins or other proprietary stuff 1346961389 M * Bertl (i.e. you can just hope that it works or install a workaround to redirect) 1346961422 M * kermie Quite possibly....when I've used x2go connecting directly from one device to another this has worked. Also, I think when I tried kde/kdm it worked (am now using gnome). 1346961450 M * kermie Sorry....in the first case was not using vserver; in the 2nd case (kde/kdm) was using vserver. 1346961455 M * Bertl well then I'm at least sure that you can configure it in such way that it will work 1346961478 M * Bertl (if it works on the host, there is no reason it shouldn't work on a guest) 1346961501 M * kermie Would this be not having the right "dev" items 1346961521 M * Bertl maybe, but maybe it's just having the wrong config 1346961549 M * Bertl first, try to identify the devices actually used in both cases 1346961564 M * Bertl (the correct and the wron one) 1346961568 M * Bertl *wrong 1346961617 M * kermie OK...FYI - if I do cat /proc/asound/version in the host, I get "Advanced Linux Sound Architecture Driver Version 1.0.21." If I do the same in the guest, I get no file found. 1346961625 M * kermie Not sure this means anything or matters... 1346961648 M * Bertl well, it just means that this proc entry is hidden 1346961666 M * Bertl but that is probably a good thing, as this would be the raw hardware audio device 1346961677 M * Bertl (which you do not want for network audio anyway) 1346961724 M * kermie OK...cool. Last question. Where would look to fix "wrong config"? Is that using the various alsa tools or possibly some pulse audio files? If x2go is using pulseaudio, should I be aiming at that? 1346961777 M * Bertl hard to tell, haven't worked with x2go yet, IMHO a simple test/check would be to examine open filedescriptors of a working and non working app (as simple as possible) 1346961800 M * kermie OK thanks! Will keep at it...really appreciate the help. 1346961801 M * Bertl if one goes to an alsa device node, that's a clear 'ALSA' sign 1346961815 M * Bertl np, feel free to hang around 1346961833 M * kermie thanks! 1346962965 Q * sannes Remote host closed the connection 1346963288 Q * uranus Quit: Verlassend 1346964982 J * fisted ~fisted@xdsl-87-78-10-244.netcologne.de 1346965119 Q * fisted_ Ping timeout: 480 seconds 1346966400 J * clopez ~clopez@17.28.165.83.dynamic.mundo-r.com 1346966690 Q * bonbons Quit: Leaving 1346970973 Q * ghislain Quit: Leaving. 1346975290 Q * clopez Read error: Operation timed out