1408407175 M * Bertl off to bed now .. have a good one everyone!
1408407191 N * Bertl Bertl_zZ
1408423848 J * webhat ~quassel@31.25.99.5
1408423849 Q * redhat Read error: Connection reset by peer
1408428963 J * fisted_ ~fisted@xdsl-87-78-80-193.netcologne.de
1408429411 Q * fisted Ping timeout: 480 seconds
1408429411 N * fisted_ fisted
1408430897 J * Ghislain ~aqueos@adsl1.aqueos.com
1408432759 J * kleene ~epebara@192.176.1.90
1408432762 M * kleene hi
1408432839 M * kleene I'm building my vserver with the debootstrap method, but when debootstrap tries to finish, it errors out, because it can't mount /proc. what do you recommend to make it work?
1408434308 M * distemper kleene: are you able to chroot into the vserver dir manually? this mostly happens if one tries to install a 64bit guest running a 32bit kernel or sth. like that (so chrooting doesn't work)
1408434408 M * kleene distemper: I think it's because vserver doesn't allow mounting inside a chroot
1408434480 M * kleene sorry, I mean grsec
1408434638 M * distemper ok, that should appear in the syslog... but i'm not a debian expert, so can't tell you how to outsmart debootstrop, sorry
1408434685 M * kleene I will probably just disable the chroot mount restriction in grsec while installing a new guest
1408434699 M * kleene that's probably the cleanest way
1408434729 M * kleene then I'll clean up the guest, so it doesn't mount anything
1408435833 T * * http://linux-vserver.org/ |stable 3.6.x-vs2.3.x|util-vserver-0.30.216-pre3054| He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the Wiki, and we forget about the minute.
1408435833 T * ChanServ -
1408435925 J * marbug ~marbug@94.139.49.164
1408438771 J * thierryp ~thierry@zebra.inria.fr
1408442841 M * Guy- Bertl_zZ: I was able to reproduce the jenkins slave.jar hang I mentioned the other day
1408442849 M * Guy- Bertl_zZ: strace looks like this: http://paste.linux-vserver.org/235487
1408442856 M * Guy- I'll try to produce a full trace
1408443580 M * Guy- I now have a full-ish trace obtained using strace -ttttTffF -s200 -e trace='!futex,gettimeofday,clock_gettime' -o /tmp/jenkins.log
1408443588 M * Guy- it doesn't seem to contain anything untoward
1408443973 Q * kleene Quit: ircII EPIC5-1.1.2 -- Are we there yet?
1408444589 N * ensc Guest1211
1408444599 J * ensc ~irc-ensc@p54ADD4C9.dip0.t-ipconnect.de
1408445009 Q * Guest1211 Ping timeout: 480 seconds
1408445101 M * Guy- and yes, it's definitely hanging, no calls other than futex, gettimeofday or clock_gettime in the last 30 minutes
1408445567 M * Guy- Bertl_zZ: here is the full trace: http://paste.linux-vserver.org/235516
1408445957 Q * Aiken Remote host closed the connection
1408446489 M * marbug I'm trying to install a centos6 from the distributions template, it wasn't there so I tool scientificlinux, I addapted the repo to CentOS-Base.repo from a standard centos installation, but I'm always getting an error of $releasever not being set:
1408446494 M * marbug YumRepo Error: All mirror URLs are not using ftp, http[s] or file.
1408446494 M * marbug  Eg. $releasever is not a valid release or hasnt been released yet/
1408447491 J * thierryp_ ~thierry@zebra.inria.fr
1408447492 Q * thierryp Read error: Connection reset by peer
1408451453 Q * hijacker Ping timeout: 480 seconds
1408451488 M * Guy- this looks like a relevant difference between the bad and the good case with jenkins:
1408451491 M * Guy- -open("/usr/lib/jvm/java-1.6.0-openjdk-1.6.0/jre/lib/i386/client/tls/i686/sse2/libpthread.so.0", O_RDONLY) = -1 ENOENT (No such file or directory)
1408451494 M * Guy- +open("/usr/lib/jvm/java-1.6.0-openjdk-1.6.0/jre/lib/i386/server/tls/i686/sse2/libpthread.so.0", O_RDONLY) = -1 ENOENT (No such file or directory)
1408451515 M * Guy- when it's bad, it uses the "client" flavour and when it's good it uses "server"... but why?
1408451789 M * Guy- well, in the vserver it sees a 32bit personality, that might explain it
1408451795 J * hijacker ~hijacker@213.91.163.5
1408452012 M * Guy- but losing the personality and starting with -server explicitly also doesn't solve the problem
1408452398 M * Guy- now I'm at a point where it seemingly always works when straced from the start, but not if I don't strace it
1408452452 N * Bertl_zZ Bertl
1408452461 M * Bertl morning folks!
1408452474 M * Bertl Guy-: java is sometimes strange ...
1408452705 M * Guy- Bertl: no shit :)
1408452719 M * Guy- still, I only get the hanging behaviour in the vserver
1408452725 M * Guy- if I run it in a chroot, it never hangs
1408452739 M * Guy- Bertl: any ideas? workarounds?
1408453951 M * Bertl remove any memory limits, personality changes, etc
1408453970 M * Bertl make sure that the guest mounts the 'expected' tmpfs entries
1408454095 M * Guy- Bertl: the guest doesn't have any limits and now also no personality
1408454153 M * Guy- (I think removing the personality was what resulted in the current state, i.e. it works when straced but not when not)
1408454331 M * Guy- hmmm, now it works without strace (after a vserver restart)
1408454432 M * Guy- but it crashed on a SIGSEGV relatively quickly
1408454592 M * Bertl that is probably due to a mixed guest environment
1408454612 M * Guy- mixed how?
1408454621 M * Bertl i.e. your guest obviously has a different binary architecture than the host
1408454622 M * Guy- x86_64 arch but i386 binaries?
1408454628 M * Bertl yes
1408454629 M * Guy- yes, that's the case
1408454637 M * Guy- but in a chroot it doesn't crash
1408454640 M * Bertl so the userspace behaves differently
1408454918 M * Guy- why does it behave differently in the vserver than in the chroot?
1408454920 M * renihs maybe stuff leaks in from the chroot that makes a diff?
1408455046 M * Bertl most likely, like limits, cgroups, certain mounts, etc
1408455101 M * Bertl for example, unless you mount /proc in the chroot, it won't be visible there (just to give an example)
1408455143 M * Bertl but there are many other subtle differences between chroot and a complete guest system
1408455168 M * Guy- well, rlimits would certainly leak in
1408455177 M * Guy- I do mount /proc in the chroot, also /sys
1408455178 M * Bertl btw, the past you sent me doesn't show any indication of 'waiting'?
1408455183 M * Bertl *paste
1408455203 M * Bertl but maybe it was cut off
1408455216 M * Guy- Bertl: which paste do you mean?
1408455222 M * Guy- the first one only contained the "waiting" bit
1408455227 M * Bertl http://paste.linux-vserver.org/235516
1408455272 M * Bertl it ends with a read, which IMHO should not 'hang'
1408455274 M * Guy- well, at this point the java process only executes futex, gettimeofday and clock_gettime in an endless loop
1408455291 M * Guy- but those are excluded from the trace to keep the size sane
1408455345 M * Bertl you are aware that futex are userspace semaphores?
1408455356 P * undefined 
1408455361 M * Guy- yes
1408455372 M * Bertl so running futex, and gettime in a loop looks like a proper main thread synchronizing other threads
1408455380 M * Bertl i.e. nothing unusual there either
1408455387 M * Guy- of course
1408455395 M * Guy- like I said, "normal java background crap"
1408455404 M * Guy- which is why I felt safe excluding it from the strace
1408455418 M * Guy- _but_ other than doing this, it doesn't make any progress
1408455445 J * undefined ~undefined@00011a48.user.oftc.net
1408455450 M * Bertl but that sounds more like a userspace problem, maybe turn on java logging/debugging?
1408455459 M * Guy- haha, nice one :)
1408455484 M * Guy- that's what I tried first, but either I'm stupid (possible) or this thing has no useful tracing
1408455525 M * Guy- I'm fairly certain it's a userspace problem, but it somehow only manifests when running in a vserver guest
1408455536 M * Guy- (did I mention I hate java?)
1408455641 M * Bertl I'm with you on the hate part ...
1408455921 M * Bertl maybe try to play around a little with the memory options for java
1408455936 M * Guy- tried already, made no observable difference
1408455941 M * Ghislain look also for /tmp filling just in case
1408455960 M * Ghislain default one run in small ram and can be quickly filled
1408455974 M * Ghislain especialy by java who take gigabyte for hello world
1408456010 M * Guy- my /tmp is a real fs
1408456020 M * Guy- also, an fs filling up would have showed up in strace
1408456037 M * Ghislain ok so one mark on the list, go one line down 
1408456056 M * Guy- which list?
1408456133 M * Ghislain list of thing to check ;p
1408456327 M * Guy- Ghislain: OK, what would be the next item on the list? :)
1408456385 Q * marbug Ping timeout: 480 seconds
1408456946 J * marbug ~marbug@94.139.49.164
1408458390 Q * Jb_boin Read error: Connection reset by peer
1408458431 J * Jb_boin ~dedior@proxad.eu
1408459268 J * Wermwud ~Wermwud@69-29-150-18.stat.centurytel.net
1408460013 M * marbug I've did a yum update and the kernel was updated for the linux vserver, but now I'm getting "can not change context: migrate kernel feature missing and 'compat' API disabled: Function not implemented" there isn't any update left, is vserver screwed for centos or what is that? :/
1408460041 M * Bertl most likely you got a kernel without linux-vserver support
1408460085 M * marbug yes that was what I thought, and it installed a new without the linux-vserver support, I'm not getting why it did that :/
1408460154 M * marbug and I don't get it why, but really I only have problems with vserver inside centos :(
1408460168 M * Bertl well, you have to ask the centos folks about that, but as Linux-VServer is not officially supported, I presume they will tell you that it is your fault or so :)
1408460188 M * marbug nope the'll send me to here ;-)
1408460203 M * marbug maybe it was better that I installed a gentoo host too :)
1408460287 M * Bertl personally I never cared about distro support, because compiling the kernel and tools from sources is usually trivial and gives a well tailored and performant system
1408460463 M * marbug yes, but I'm already tampering since yesterday, got problems with ip and now even ip-ports I never had those problems
1408460747 M * marbug the thing is, I setted up a vserver guest with an ssh service bound to another port. From the host, I can connect to the ssh with that port, but not from my laptop :/ (note: the server has 2 routers attached, but the sshd is bound to the right network)
1408460795 M * marbug I have the feeling there is something really wrong with the ethernet thing in the vserver kernel :/
1408464385 J * bonbons ~bonbons@2001:a18:207:101:f1a8:f269:3931:51f1
1408464992 J * zerick ~eocrospom@190.187.21.53
1408466121 Q * marbug Read error: Operation timed out
1408466582 Q * thierryp_ Remote host closed the connection
1408469515 M * Bertl off for a nap ... bbl
1408469530 N * Bertl Bertl_zZ
1408470427 J * thierryp ~thierry@home.parmentelat.net
1408471655 Q * thierryp Remote host closed the connection
1408475587 J * fisted_ ~fisted@xdsl-84-44-226-172.netcologne.de
1408475650 N * Bertl_zZ Bertl
1408475657 M * Bertl back now ...
1408476036 Q * fisted Ping timeout: 480 seconds
1408476038 N * fisted_ fisted
1408476248 J * Aiken ~Aiken@d63f.h.jbmb.net
1408479611 Q * Romster Ping timeout: 480 seconds
1408479778 J * Romster ~Romster@202.168.100.149.dynamic.rev.eftel.com
1408481290 Q * bonbons Quit: Leaving
1408485883 Q * disposab1e Read error: Connection reset by peer
1408486391 Q * Wermwud Quit: Leaving (Please imagine me slamming the door on my way out)
1408486711 Q * Ghislain Quit: Leaving.
1408487207 J * thierryp ~thierry@home.parmentelat.net
1408487767 Q * thierryp Remote host closed the connection
1408487777 J * thierryp ~thierry@2a01:e35:2e2b:e2c0:e835:648d:445b:d45c
1408488263 Q * thierryp Ping timeout: 480 seconds
1408490773 J * thierryp ~thierry@home.parmentelat.net
1408491250 Q * thierryp Remote host closed the connection
1408491321 J * thierryp ~thierry@home.parmentelat.net
1408492569 Q * thierryp Remote host closed the connection