1191024229 Q * Punkie Quit: ChatZilla 0.9.78.1 [Firefox 2.0.0.7/2007091417] 1191024253 P * dowdle Don't want to miss my bus home. I actually learned a lot about Linux-VServer today. Thanks for all of the help. 1191028025 J * FireEgl FireEgl@4.0.0.0.1.0.0.0.c.d.4.8.0.c.5.0.1.0.0.2.ip6.arpa 1191028410 Q * fatgoose_ Quit: fatgoose_ 1191028979 J * ktwilight_ ~ktwilight@213.86-66-87.adsl-dyn.isp.belgacom.be 1191029162 Q * yarihm Quit: Leaving 1191029389 Q * ktwilight Ping timeout: 480 seconds 1191039272 J * _jmcaricand_zzz ~jmcarican@d83-179-153-107.cust.tele2.fr 1191042459 Q * zLinux Remote host closed the connection 1191047432 Q * edible Ping timeout: 480 seconds 1191047695 J * dna ~dna@148-244-dsl.kielnet.net 1191048485 J * Julius ~julius@p57B253DF.dip.t-dialin.net 1191049656 Q * Julius Quit: Verlassend 1191049660 J * Julius ~julius@p57B253DF.dip.t-dialin.net 1191049944 J * JonB ~NoSuchUse@kg0-231.kollegiegaarden.dk 1191050482 J * StOaN ~julius@p57B253DF.dip.t-dialin.net 1191050545 J * |jmcaricand|_ ~jmcarican@d83-179-232-38.cust.tele2.fr 1191050911 Q * _jmcaricand_zzz Ping timeout: 480 seconds 1191051036 Q * StOaN Quit: leaving 1191052165 Q * JonB Ping timeout: 480 seconds 1191052852 J * bonbons ~bonbons@2001:960:7ab:0:20b:5dff:fec7:6b33 1191053270 J * fatgoose ~samuel@206-248-132-37.dsl.teksavvy.com 1191053693 J * JonB ~NoSuchUse@130.227.63.19 1191053863 Q * FireEgl Read error: Connection reset by peer 1191054494 J * FireEgl FireEgl@4.0.0.0.1.0.0.0.c.d.4.8.0.c.5.0.1.0.0.2.ip6.arpa 1191054957 J * coderanger_ ~coderange@x-1-13.dynamic2.rpi.edu 1191055207 J * julius_ ~julius@p57B253DF.dip.t-dialin.net 1191055619 Q * Julius Ping timeout: 480 seconds 1191055821 J * julius__ ~julius@p57B253DF.dip.t-dialin.net 1191056160 Q * julius_ Ping timeout: 480 seconds 1191056293 Q * fatgoose Quit: fatgoose 1191057353 J * Baby ~miry@195.37.62.208 1191057591 Q * Baby 1191059603 N * Bertl_zZ Bertl 1191059609 M * Bertl morning folks! 1191059685 M * JonB hej 1191060012 M * blizz moin 1191060289 J * CWC ~CWC@89-215-37-177.2073053861.ddns-lan.pl.ekk.bg 1191061002 Q * CWC Ping timeout: 480 seconds 1191061714 Q * coderanger_ Quit: coderanger_ 1191061867 J * coderanger_ ~coderange@x-1-13.dynamic2.rpi.edu 1191061869 Q * coderanger_ 1191063259 J * Punkie ~Punkie@home.pekelny.net 1191063372 M * bzed morning Bertl 1191064023 M * JonB Bertl: yesterday we briefly talked about assigning the broadcast address to a particular vserver. But how do i do that in detail? can i have 2 lines in the ip file? 1191064082 M * Bertl JonB: /etc/vservers//interfaces//ip 1191064112 M * Bertl the part is the solution here :) 1191064163 M * JonB Bertl: i already have a interfaces/0/ip but can i just write 2 ip addresses in that file? 1191064267 M * Bertl no 1191064275 M * Bertl but you can make an interfaces/1/ip 1191064292 M * JonB and inside 1 there should be a nodev? 1191064299 M * Bertl precisely 1191064303 M * JonB thanks 1191064307 M * Bertl you're welcome! 1191064528 J * Piet ~piet@tor.noreply.org 1191064707 M * JonB Bertl: can interfaces/0/dev and interfaces/1/dev be identical? 1191064802 M * Bertl yes 1191064824 M * JonB okay 1191066241 J * DuckMaster ~Duck@tox.dyndns.org 1191066241 Q * duckx Read error: Connection reset by peer 1191066844 J * ecomp ~hmaier@62.218.223.240 1191066907 P * ecomp 1191067433 J * ema ~ema@rtfm.galliera.it 1191067962 J * julius_ ~julius@p57B24586.dip0.t-ipconnect.de 1191068347 Q * julius__ Ping timeout: 480 seconds 1191068754 J * drsource 0x2E2989@89-215-37-177.2073053861.ddns-lan.pl.ekk.bg 1191069642 M * Bertl welcome drsource! 1191070417 Q * drsource Ping timeout: 480 seconds 1191070699 M * epicbjorn moin1 1191070718 Q * michal_ Ping timeout: 480 seconds 1191070719 A * epicbjorn teaches himself iproute2 1191071735 J * michal_ ~michal@www.rsbac.org 1191071984 J * arekm ~arekm@chello089076024040.chello.pl 1191071993 M * arekm er 1191072032 M * epicbjorn er 1191072038 M * epicbjorn arrw 1191072207 Q * Piet Ping timeout: 480 seconds 1191072353 J * Piet ~piet@tor.noreply.org 1191072721 M * Bertl hey arekm! 1191072833 M * Bertl arekm: please give the following patch a try: http://vserver.13thfloor.at/Experimental/delta-earlyexit-fix01.diff 1191073043 M * arekm in a second 1191073092 Q * DavidS Ping timeout: 480 seconds 1191073692 M * Bertl arekm: btw, it would be interesting to figure why you do not have line information in vmlinux 1191073705 M * Bertl arekm: or maybe just the toolchain is messed up somehow 1191073971 M * JonB Bertl: i was thinking... you say that you can not run samba in more than one vserver guest. But what if the guests did not have the same physical netcard, but still both was attached to the same switch/network? 1191074003 M * Bertl did I say that? 1191074016 M * daniel_hozac i run samba in three guests without any problems... 1191074025 M * Bertl if so, fact is, you cannot bind more than one samba server to the same network broadcast address :) 1191074054 M * Bertl there is no problem serving on different networks or even share one network (if not broadcasting) 1191074091 M * JonB Bertl: and iptables can not copy a packet set to broadcast address and samba port number 2 both servers? 1191074102 M * JonB Bertl: into both vserver guests and their samba daemon 1191074180 Q * DuckMaster Ping timeout: 480 seconds 1191074194 M * daniel_hozac note that you don't need broadcast if you use WINS or DNS to resolve names... 1191074210 M * JonB daniel_hozac: okay, because i do use that 1191074220 M * Bertl yes, the broadcast thing is just a windows oddity 1191074224 M * JonB daniel_hozac: but i still could not get samba to work yesterday 1191074241 M * Bertl JonB: btw, do you use MacOS X on your laptop? 1191074268 M * JonB Bertl: yes 1191074293 M * Bertl hehe, thought I recongnize the quit: computer went to sleep message :) 1191074313 M * JonB heh 1191074516 M * arekm Bertl: crashed with that patch and no history unfortunately 1191074582 M * arekm trying again 1191074766 M * Bertl arekm: can you upload the oops for that one? 1191074851 M * arekm http://pastebin.com/mfd145d2 1191074856 M * JonB daniel_hozac: and samba does not work today either. Now i will try to move the existing server to become a guest and see if that works 1191074868 M * daniel_hozac why does it not work? 1191074889 M * JonB daniel_hozac: client says username or password is wrong 1191074910 M * daniel_hozac and broadcast would help.. how? 1191074924 M * daniel_hozac you did add the user to /etc/passwd and run smbpasswd -a , right? 1191074935 M * JonB daniel_hozac: i joined the domain 1191075017 M * daniel_hozac so you configured samba for that? 1191075024 M * JonB yes 1191075034 M * JonB copied the smb.conf file from the existing server 1191075046 M * JonB and then joined the domain 1191075050 M * JonB it did say i joined 1191075055 M * JonB and i can get winbind to work 1191075079 M * daniel_hozac what do you get in the logs? 1191075092 M * daniel_hozac (not that any of these problems seem Linux-VServer related to me...) 1191075141 M * JonB daniel_hozac: they dont? sorry, it seemed vserver related to me because i copied the data dirs, the config file, and joined the domain 1191075341 M * daniel_hozac Bertl: hmm, won't that earlyexit patch mess up the mm accounting? 1191075378 M * Bertl I thought so, but it was just a test, funny part is that it didn't mess up the accounting so far :) 1191075559 M * Bertl daniel_hozac: for me arekm issues look like somehow mm gets released over and over again, although there are still tasks accounted 1191075605 M * arekm http://pastebin.com/m46089dfe 1191075612 M * Bertl the history dump we got yesterday also looks like ref counting reaching zero with a lot of tasks 1191075615 M * arekm the same thing again 1191075661 M * daniel_hozac could it be some change related to clone(CLONE_VM)? 1191075668 M * daniel_hozac i.e. threads. 1191075697 M * Bertl yes, definitely, IIRC, arekm reports all kernels affected except for quite old 2.1 ones 1191075731 M * Bertl note: I think it is x86_64 specific, and I'm not sure it will trigger on UP 1191075732 M * daniel_hozac that still seems highly unlikely, as nobody else has seen it... 1191075749 M * Bertl well, this is a 8 way x86_64 :) 1191075771 M * daniel_hozac that doesn't seem like such an uncommon setup these days. 1191075808 M * Bertl yeah, I agree, this setup is the only one so far showing those issues 1191075824 M * daniel_hozac if we had a reproducer, i could give it a spin. 1191075832 M * Bertl nevertheless I think we should at least narrow it down, as it doesn't happen at so many places 1191075847 M * Bertl all arekm seems to be doing is 5min kernel compile in a guest 1191075862 M * Bertl arekm: please elaborate about your trigger setup 1191075865 M * daniel_hozac hmm, well, that definitely works fine here. 1191075874 M * daniel_hozac granted, it's just a 2-way x86_64. 1191075903 M * Bertl I was first suspecting NUMA, but there is no NUMA involved 1191075917 M * arekm Bertl: I'll stop all services in the guest and do compilation only - will see if it crashes. Right now guest runs few tcp services, too 1191075937 M * Bertl arekm: yeah, maybe try the following too: 1191075947 M * Bertl chcontext --xid 666 -- bash 1191075949 M * daniel_hozac btw, have we addressed the tag != xid in proc yet? 1191075956 M * Bertl arekm: then do the kernel compile 1191075963 M * Bertl daniel_hozac: nope 1191075976 M * daniel_hozac okay, i'll try to do that today then. 1191075982 M * Bertl excellent! 1191075985 M * daniel_hozac my logs are getting full on my build server ;) 1191076048 M * JonB success 1191076068 M * JonB but i do not know why :-( 1191076197 M * JonB i hate that 1191076205 M * Bertl quick, celebrate while it lasts! :) 1191076211 M * JonB only change is that time has passed 1191076219 M * JonB Bertl: yeah 1191076277 Q * Aiken Quit: Leaving 1191076284 M * JonB winbind suddently started working as well 1191076294 M * JonB maybe it needs to cache all the objects? 1191076545 J * CWC ~CWC@89-215-37-177.2073053861.ddns-lan.pl.ekk.bg 1191076570 M * arekm Bertl: http://pastebin.com/m250771fc chcontext --xid 666 -- chroot /vservers/carme-pld su - arekm + cd linux && make -j6 1191076603 M * JonB i have an idea to an explination. I joined using a backup domain controller since the primary is in normay and firewalled. so maybe the backup domain controller had to sync with the primary before me joined the network would actually work 1191076623 M * Bertl arekm: ah, great! 1191076655 M * arekm History: SEQ: d894 NR_CPUS: 16 1191076661 A * arekm wonders what's that nr_cpus 1191076682 M * Bertl it means that your kernel defined a max of 16 cpus 1191076691 M * arekm ah 1191076700 M * Bertl so the history buffers are using that too 1191076739 M * Bertl what we need now is to convert those @ addresses to acctual code lines 1191076868 M * JonB Bertl: you mean you actually have to do some work? 1191076900 M * Bertl you mean, besides explaining the obvious to folks on the channel? 1191076923 M * JonB Bertl: yeah, do you have time for anyone but me? 1191076954 M * Bertl not sure about that .. maybe when you're off to bed :) 1191076972 M * JonB Bertl: okay, i need to find some more questions ;-) 1191077138 M * JonB Bertl: would you be a right person to ask about speaking about the olpc bitfrost security and what can really be gained from vserver? 1191077275 M * daniel_hozac Ashsong, neuralis, and coderanger should be able to as well. 1191077290 M * JonB are they people in here? 1191077311 M * JonB what about yoou? 1191077356 M * JonB looks like there is alot more people in here than i noticed 1191077533 M * arekm Bertl: just to make sure (#d42a,*3):ffffffff8025903c release_vx_info ffff8101533e2000[#666,22.34] @ffff81014e615080 1191077547 M * arekm Bertl: and you want addr2line on 0xffff81014e615080 in this case? 1191077604 M * Bertl yep 1191077649 M * arekm every tested address gives ??:0 1191077667 M * arekm but for example some address of fs_* function gives correct result 1191077668 M * arekm $ addr2line -e vmlinux 0xffffffff802d8970 1191077668 M * arekm /home/users/arekm/rpm/SOURCES/linux-2.6.22/fs/stack.c:23 1191077775 M * Bertl okay, let's try a different approach, you have some space for kernel logging? 1191077841 M * arekm depends what you have in mind. free space on some fs - yes 1191077874 M * Bertl okay, we can use the normal kernel/vserver debugging to get the same info like the history 1191077888 M * Bertl but with proper information of the code lines 1191077889 M * daniel_hozac arekm: try ffffffff81014e61? 1191077915 M * arekm daniel_hozac: result: ??:0 1191077924 M * daniel_hozac ffffffff8025903c? 1191077939 M * arekm /home/users/arekm/rpm/SOURCES/linux-2.6.22/include/linux/vs_context.h:142 1191077963 M * arekm vxlprintk(VXD_CBIT(xid, 3), "release_vx_info(%p[#%d.%d.%d]) %p", 1191077964 M * arekm there 1191077999 M * daniel_hozac i guess it's a module then... 1191078158 M * arekm could try monolithic kernel 1191078165 M * Bertl I guess we are interested in xid bits: 1,2,3,4 and 5 1191078200 M * Bertl so you could try activating those with sysctl before you run the compile 1191078234 M * arekm how to activate? 1191078274 A * arekm doing monolithic kernel now 1191078394 M * Bertl arekm: sysctl -w vserver.debug_xid=62 1191079522 Q * Johnnie Ping timeout: 480 seconds 1191079583 M * Bertl arekm: btw, what gcc/binutil versions? 1191079652 M * arekm gcc-4.2.1-2.x86_64 1191079654 M * arekm binutils-2.18.50.0.1-1.x86_64 1191079910 Q * JonB Ping timeout: 480 seconds 1191080097 J * Johnnie ~jdlewis@c-67-163-142-234.hsd1.ct.comcast.net 1191080745 N * ensc Guest279 1191080750 M * arekm http://pastebin.com/m7edbf349 1191080755 J * ensc ~irc-ensc@p54B4CD26.dip.t-dialin.net 1191080810 Q * julius_ Ping timeout: 480 seconds 1191080835 M * arekm and syslog over network logged that as the last thing http://pastebin.com/m42ae832b (note that when oops happens network stops working so this log is just before ooops) 1191080855 M * arekm Bertl: but no history again in oops itself ;/ 1191080863 Q * Guest279 Ping timeout: 480 seconds 1191080882 M * Bertl arekm: do you have more of the debug backlog? 1191080957 M * Bertl we definitely see here that the ref count even gets negative at some points 1191080971 M * Bertl (which is clearly a bug) 1191080982 M * arekm http://pastebin.com/m42ae832b - have tons of this 1191081107 M * arekm Bertl: http://ep09.pld-linux.org/~arekm/syslog-1.txt 1191081117 M * Bertl tx 1191081454 M * Bertl daniel_hozac: we do a clr_vx_info on free_task(), but we do not take a reference on dup_task_struct(), right? 1191081583 M * yang Bertl: Must there be a license paid for the comercial usage of Linux-Vserver ? 1191081607 M * Bertl yang: no, but you can donate something :) 1191081608 M * daniel_hozac it's GPL. 1191081634 M * yang Bertl: off course ! I allready did :) 1191081640 M * daniel_hozac Bertl: but the only place dup_task_struct is used is in copy_process, in which we use init_vx_info. 1191081672 M * Bertl correct, just checking for error pathes atm 1191081863 N * phedny Guest281 1191081869 J * phedny ~mark@ip56538143.direct-adsl.nl 1191081987 J * JonB ~NoSuchUse@kg0-231.kollegiegaarden.dk 1191082179 Q * |jmcaricand|_ Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/ 1191082272 Q * Guest281 Ping timeout: 480 seconds 1191082402 J * edible edible@12-216-231-163.client.mchsi.com 1191082418 M * Bertl welcome edible! 1191082567 Q * CWC Ping timeout: 480 seconds 1191082676 M * phedny I just finished the job of moving some vservers from one host to another 1191082684 M * phedny and this all went with very few problems :) 1191082701 M * phedny much easier than moving services in the old days... 1191082725 M * Bertl daniel_hozac: dup_mm() assumes that the passed in task_struct is identical to current, no? 1191082810 M * Bertl nah, it assumes that the copy will be done _from_ current, and will get assigned to tsk 1191083043 Q * JonB Read error: Connection reset by peer 1191083255 M * Punkie Is it possible bind service from guest to IP from host? I have disabled hide_netif, I see IPs from host in guest, but I cant bind service to that IPs. 1191083308 M * Bertl you simply have to assign the host ip to the guest 1191083320 M * Bertl but note: that isn't really advised 1191083340 M * Punkie mean in configuration of guest? 1191083345 M * Bertl a better approach is to use S/DNAT to map certain host ports to the guest ip (which can be private) 1191083391 M * Bertl yes, note that if you assign the host ip to the guest, the guest can hijack and block host services 1191083547 M * Punkie thanks a lot :) (NAT is not usable for me) the guest is only for me, not for any other person 1191083587 M * Bertl btw, why is NAT not an option? 1191083831 M * Punkie I have on server 12000+ conections , how weighting would it be to NAT them? (sorry for my english :( ) 1191083949 M * Punkie Ariadne:~# netstat -aon | wc -l 1191083955 M * Punkie 13102 1191084089 M * Bertl if you do DNAT, then it would require one additional lookup for each connection 1191084115 M * Bertl granted, that is more than without, but I doubt it would be measureable 1191084140 M * Bertl but if you are the only one using the guest, then sharing the host ip is fine 1191084176 M * Punkie :) 1191084223 M * arekm http://pastebin.com/m3d3a255e history available this time 1191084280 M * arekm checking addr2line 1191084641 M * arekm http://pastebin.com/m45e57e0f 1191085944 Q * the-dude Ping timeout: 480 seconds 1191085945 J * NoFearrr ~1@dxb-as62375.alshamil.net.ae 1191086025 P * NoFearrr 1191086238 J * the-dude ~martijn@senturparks.xs4all.nl 1191086546 M * arekm Bertl: still digging? 1191086572 M * Bertl arekm: yep 1191086622 M * Bertl arekm: let's try to trigger it differently 1191086640 M * Bertl arekm: maybe with a short burst of task creations? 1191086679 M * Bertl something like: for n in `seq 0 100`; do true & done (or something similar) 1191086684 Q * ema Quit: leaving 1191087702 Q * Piet Ping timeout: 480 seconds 1191087743 M * Bertl arekm: could you give that one a try too (with the old 'known to work' kernel compile)? 1191087746 M * Bertl http://vserver.13thfloor.at/Experimental/delta-mminit-fix01.diff 1191087794 M * daniel_hozac isn't that missing an assignment somewhere? 1191087827 M * arekm Bertl: with 2.6.17? 1191087839 M * daniel_hozac (and are there really no callers of mm_alloc that need to be updated? 1191087905 M * Bertl hmm, yeah, I forgot mm_alloc :/ 1191088017 M * Bertl daniel_hozac: which assignment? 1191088033 M * daniel_hozac well, shouldn't the vxi argument be used? 1191088062 M * Bertl it is passed on to the mm_init() 1191088063 M * arekm running "true" even 10000 doesn't trigger it 1191088074 M * daniel_hozac yeah, shouldn't mm_init use it? 1191088115 M * Bertl ah, damn, right :) 1191088157 M * Bertl forget that one, I'll update it in place shortly :) 1191088168 M * Bertl daniel_hozac: thanks for double checking! 1191088183 M * daniel_hozac hehe, no problem. 1191088491 J * zLinux ~zLinux@88.213.17.231 1191088497 J * Piet ~piet@tor.noreply.org 1191089239 J * _jmcaricand_zzz ~jmcarican@d77-216-167-1.cust.tele2.fr 1191089246 M * Bertl arekm: okay, patch was updated 1191089519 J * DavidS ~david@p57A487D0.dip0.t-ipconnect.de 1191089839 J * Julius ~julius@p57B24586.dip0.t-ipconnect.de 1191089851 M * Bertl wb DavidS! Julius! 1191089861 M * DavidS hey bertl! 1191089881 M * Julius hiho 1191089993 P * _jmcaricand_zzz Time makes no sense 1191090013 J * |jmcaricand|_ ~jmcarican@d77-216-167-1.cust.tele2.fr 1191090103 M * matti Hi B. 1191090162 Q * Piet Ping timeout: 480 seconds 1191090239 M * Bertl hey m! 1191090334 J * jmc ~user@d83-179-210-152.cust.tele2.fr 1191090338 J * Piet ~piet@tor.noreply.org 1191090400 M * DavidS Bertl, I stumbled upon the nice article about upcoming linux container stuff from the 2007 kernel summit: http://lwn.net/Articles/249080/ how much would that relate to further vserver development? 1191090509 M * Bertl well, we are part of the kernel virtualization approach (although maybe not with the manpower of ibm or google) 1191090526 N * |jmcaricand|_ jmcaricand 1191090535 M * Bertl and what has been included so far is quite similar to what we have/had in Linux-VServer 1191090563 M * Bertl some approaches are too heavy for my taste, but if they get into mainline, we will definitely support them 1191090586 M * Bertl of course, we will continue to provide light-weight alternatives as we do now 1191090751 M * DavidS ah, good, 1191090982 J * virtuoso_ ~s0t0na@ppp91-122-94-171.pppoe.avangard-dsl.ru 1191091382 Q * virtuoso Ping timeout: 480 seconds 1191092290 M * arekm Bertl: oopsed 1191092316 M * arekm Bertl: http://pastebin.com/m406572a5 1191092318 M * Bertl well, I wasn't expecting it to fix it :) 1191092336 M * Bertl although it would have been nice :) 1191092723 M * arekm so what now? 1191092736 M * Bertl well, I see a few options here 1191092782 M * Bertl a) we could try to create a very small test case which triggers this (if we succeed, going through the history/debug info should reveal the true issue) 1191092818 M * Bertl b) we can try to change some more parameters in that equation and see if they shed some more light on it 1191092852 Q * jmcaricand Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/ 1191092886 M * jmc quit 1191092888 M * daniel_hozac did you get the addr2line info for the monolithic kernel yet? 1191092890 Q * jmc Quit: ERC Version 5.0.2 $Revision: 1.726.2.11 $ (IRC client for Emacs) 1191092901 M * Bertl c) we can go through the source code and look for missing/superfluous get/put set/clr 1191092902 M * arekm daniel_hozac: yes, posted here already 1191092914 M * arekm daniel_hozac: http://pastebin.com/m45e57e0f 1191092926 J * Blissex ~Blissex@82-69-39-138.dsl.in-addr.zen.co.uk 1191092948 M * arekm daniel_hozac: doesn't look helpful 1191092982 M * daniel_hozac that's primarily because it's full of garbage... 1191093037 M * arekm http://pastebin.com/m3d3a255e this is that oops for which I did addr2line 1191093051 M * daniel_hozac fix your snippet do to it right. 1191093091 M * arekm it does it right for most of addresses 1191093104 M * arekm but for example (#0758,*5):ffffffff80236d66 init_vx_info ffff81014879b000[#666,22.30] @ffff810141e6,22.32] @ffff81015432e420 - which address is interesting? 1191093124 M * daniel_hozac but there's too much crap too even see what's interesting. 1191093142 M * daniel_hozac -o 1191093144 M * Bertl this line looks broken to me 1191093148 M * daniel_hozac indeed. 1191093165 M * Bertl probably the output got mangled in the logger or so 1191093225 M * Bertl I'm still a little confused why the history works in certain cases, and completely fails in others 1191093258 M * arekm it's possible that it always works but IPMI console messes things and doesn't show entire thing 1191093297 Q * igraltist Ping timeout: 480 seconds 1191093660 M * arekm looks like we are going nowhere with this stuff ;-/ back to 2.6.17 1191093738 Q * arekm Quit: leaving 1191093881 M * Bertl so 2.6.17 is the last kernel which works? 1191093899 M * daniel_hozac too late ;) 1191093941 J * arekm arekm@carme.pld-linux.org 1191094082 M * Bertl so 2.6.17 is the last kernel which works? 1191094112 M * Bertl arekm: what Linux-VServer versions did you test so far? 1191094147 M * Bertl (maybe we can narrow it down this way) 1191094191 M * arekm 2.6.17+vs2.1 good, 2.6.20+vs2.3.0.12 bad 1191094209 M * Bertl okay, what about in-betweens? 1191094236 M * daniel_hozac note that such early 2.3 versions were basically untested. 1191094266 M * Bertl well, nevertheless, we are testing 2.3.0.24 now :) 1191094281 M * arekm daniel_hozac: 2.6.22.9+2.3.0.24 bad 1191094293 M * daniel_hozac yeah, i'm just saying it's possible 2.6.20 works too with a known-good Linux-VServer patch. 1191094308 M * Bertl what about 2.6.20 with 2.2.x? 1191094321 M * Bertl do we have something in this area? 1191094472 M * Bertl maybe we could do some kind of patch version bisection? 1191094479 M * daniel_hozac are you using infiniband? 1191094490 M * arekm no infiniband 1191094541 M * arekm Bertl: vserver is tracked with git? 1191094551 M * daniel_hozac no 1191094572 M * Bertl not yet :) 1191094677 M * arekm I could test 2.6.19+2.6.19.1-vs2.3.0.6 (non vanilla) 1191094710 M * Bertl okay, any info which narrows it down would help, IMHO 1191094736 M * daniel_hozac of course, stable patches wouldn't hurt... 1191094738 M * Bertl btw, which 2.1.x version for 2.6.17? 1191094819 M * arekm patch-2.6.17.11-vs2.1.1-rc31.diff 1191094907 M * arekm checking patch-2.6.22.2-vs2.2.0.3.diff 1191094918 M * daniel_hozac why not .6? 1191094948 M * daniel_hozac though for recent kernels, devel vs. stable doesn't matter much. 1191094959 M * Bertl maybe we should have a look at the config too, just to see if anything stands out 1191094965 M * daniel_hozac indeed. 1191094968 M * Bertl (which makes this machine/kernel special) 1191095039 M * arekm http://carme.pld-linux.org/~arekm/.config (no modules were loaded) 1191095280 M * Bertl hmm, but NUMA is enabled here ... 1191095328 M * arekm and runtime disabled since not found 1191095685 Q * arekm Quit: znów reboot 1191095710 J * arekm arekm@carme.pld-linux.org 1191095808 Q * arekm Remote host closed the connection 1191095952 J * arekm ~arekm@chello089076024040.chello.pl 1191095999 M * Bertl CONFIG_FRAME_POINTER and CONFIG_PRINTK_TIME would be interesting in the future 1191096050 M * Bertl daniel_hozac: did you see anything obvious? 1191096069 M * arekm testing 2.6.22.6-vs2.2.0.3 now 1191096148 M * daniel_hozac no, but then again, i don't know what might cause this. 1191096289 M * arekm 2.6.22.6-vs2.2.0.3 crashed http://pastebin.com/m65bd81b 1191096393 M * arekm so which version now? 1191096419 M * daniel_hozac 2.6.18? 1191096437 M * arekm don't see patch for .18 at http://ftp.linux-vserver.org/pub/kernel/vs2.2/ 1191096446 M * daniel_hozac check testing. 1191096460 M * arekm ok, will test http://ftp.linux-vserver.org/pub/kernel/vs2.2/testing/patch-2.6.18.5-vs2.2.0-pre5.diff 1191096720 M * arekm crap, oldconfig asks for everything ;/ 1191096736 M * daniel_hozac with your 2.6.17 based config? 1191096750 M * daniel_hozac should only ask for netfilter related things and few other things, IIRC. 1191096763 M * arekm no, 2.6.22 one. 2.6.17 config is fully modular 1191096947 J * Aiken ~james@ppp121-45-249-108.lns2.bne4.internode.on.net 1191097368 J * Piet_ ~piet@tor.noreply.org 1191097436 J * cshelling ~Perbabr@62.215.195.66 1191097588 P * cshelling 1191097623 J * besonen_mobile_ ~besonen_m@71-220-231-201.eugn.qwest.net 1191097747 Q * Piet Ping timeout: 480 seconds 1191097983 Q * besonen_mobile Read error: Operation timed out 1191098047 M * arekm daniel_hozac: http://pastebin.com/m4b818e1a 1191098051 M * arekm daniel_hozac: oopsed ;( 1191098289 M * arekm which ver now? 1191098331 M * Bertl let's head for something on 2.6.17 1191098331 M * daniel_hozac and the same config works on 2.6.17? 1191098374 M * arekm daniel_hozac: yes+default value on new things that appared in .18 (but 2.6.17 uses 2.1 vserver, not 2.2) 1191098394 M * Bertl or maybe 2.6.18 with 2.1? 1191098412 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.18.2-vs2.1.1.diff 1191098501 M * arekm ok, testing 1191098523 M * Bertl arekm: and thanks for your time, we appreciate it! 1191099532 M * arekm testing 2.6.18.5-vs2.1.1smp for a while and it doesn't want to crash 1191099607 M * Bertl what about http://vserver.13thfloor.at/Experimental/patch-2.6.18.2-vs2.0.2.2-rc6.diff ? 1191099620 M * daniel_hozac hmm, why not 2.0.3-rc1? 1191099633 M * Bertl yeah, fine too 1191099670 M * arekm ok, http://vserver.13thfloor.at/Experimental/patch-2.6.18.5-vs2.0.3-rc1.diff 1191100393 M * arekm crashed http://pastebin.com/m6e68fc8 1191100450 M * Bertl okay, the 2.6.18.5-vs2.1.1 was self compiled too? 1191100484 M * arekm I'm testing everything by compiling some 2.6.22 tree 1191100507 M * arekm and on vs2.1.1 compilation finished, 2 times 1191100541 M * arekm + I was running 2.6.17+vs2.1 for weeks on heavily used machine so that version works for sure 1191100575 M * arekm on this (heavily) used machine (carme) 1191100616 M * Bertl what I'm trying to figure is, you did compile both, the 2.6.18.5-vs2.1.1 and the 2.6.18.5-vs2.0.3 in a similar/identical way, and with almost identical config, yes? 1191100619 Q * Julius Remote host closed the connection 1191100678 M * arekm Bertl: http://pastebin.com/mc9bd0f9 1191100867 J * dna_ ~dna@148-244-dsl.kielnet.net 1191100920 Q * dna Read error: Connection reset by peer 1191101170 M * arekm are there any versions between these to test? 1191101286 M * Bertl what are the two patches you actually used? 1191101305 M * Bertl the base kernel is identical, yes? 1191101312 M * arekm patch-2.6.18.2-vs2.1.1.diff and patch-2.6.18.5-vs2.0.3-rc1.diff 1191101324 M * arekm yes, I'm reverting one patch and applying the other 1191101388 M * Bertl okay, let me see what the actual differences are ... 1191101409 Q * dna_ Quit: Verlassend 1191101503 M * Bertl mainly __do_IRQ enter/leave changes 1191101753 M * arekm some CLONE_KTHREAD 1191101770 M * Bertl yep, but that one is only used to block kernel threads 1191101784 M * Bertl daniel_hozac: we replaced the enter/leave stuff with additional checks, right? 1191101803 M * Bertl daniel_hozac: could it be that we missed some cases for mm stuff? 1191101831 M * arekm btw. I'm running ntpd on host if that matters 1191101857 M * daniel_hozac yeah, the enter/leave stuff was made vx_check flags. 1191101890 M * daniel_hozac i guess it's possible... 1191102161 Q * hparker Quit: reboot 1191102456 Q * DavidS Quit: Leaving. 1191102608 J * hparker ~hparker@linux.homershut.net 1191102825 M * Bertl I don't see anything obvious atm ... but I'm pretty tired .... maybe we should continue tomorrow? 1191102928 M * arekm VX_IDENT is still to be used? 1191102964 M * arekm I see that it's replaced by VS_IDENT in many places 1191102975 M * arekm but not all 1191103019 M * Bertl should be replaced by now 1191103162 M * arekm ok 1191103174 J * hardwire` ~bip@rdbck-2318.palmer.mtaonline.net 1191103187 Q * hardwire Ping timeout: 480 seconds 1191103412 M * arekm going away then, will back tommorow 1191103421 M * Bertl okay, thanks! have a good night! 1191103441 M * Bertl and a good one to everyone too .. cya 1191103448 Q * arekm Quit: karamba, kraina deszczowcow etc 1191103448 N * Bertl Bertl_zZ 1191103768 J * arekm arekm@carme.pld-linux.org 1191104382 Q * Piet_ Quit: Piet_ 1191104642 Q * ensc Remote host closed the connection 1191105693 J * onox ~onox@kalfjeslab.demon.nl 1191105732 M * onox sometimes the netwerk connection in a vserver guest just goes away, ifconfig doesn't show anything. Is this a known issue? 1191105800 M * daniel_hozac do you have multiple guests sharing a network on which the host does not have an address? 1191105866 M * onox you mean the master host doesn't have an address? 1191105885 M * onox it does have one, i can fix the broken guest by restarting it 1191106484 Q * bonbons Quit: Leaving 1191107630 J * adrien-modulis ~adrien3@216.252.77.86 1191107632 P * adrien-modulis 1191107666 J * adrien-modulis ~adrien3@216.252.77.86 1191107694 P * adrien-modulis 1191107940 J * esa ~esa@ip-87-238-2-45.adsl.cheapnet.it 1191107949 Q * eSa| Ping timeout: 480 seconds 1191109516 M * daniel_hozac onox: did you down the host's address? 1191109548 M * daniel_hozac as long as you haven't given the guest too many caps, it is completely unable to affect the networking. 1191109745 M * onox no 1191109770 M * onox i was just emerging some program until wget complained it couldn't resolve some hostname