1382155580 M * Bertl off to bed now ... have a good one everyone! 1382155594 N * Bertl Bertl_zZ 1382172520 J * Ghislain ~aqueos@adsl1.aqueos.com 1382180363 Q * ircuser-1 Ping timeout: 480 seconds 1382181127 N * Bertl_zZ Bertl 1382181141 M * Bertl morning folks! 1382182714 M * Bertl fback: ping? 1382182944 J * ircuser-1 ~ircuser-1@35.222-62-69.ftth.swbr.surewest.net 1382184026 M * Bertl off for now ... bbl 1382184030 N * Bertl Bertl_oO 1382186047 J * huser ~term@c-68-33-148-66.hsd1.md.comcast.net 1382186134 M * huser hi.. I'm trying to get hadoop to work inside a vserver, and the jobtracker (despite my configuring everything with the vserver's IP) somehow keeps managing to get the host's IP, and then fails to work because I can't re-route it to the local one. does anyone know what I can change so the vserver can only get the vserver's IP? 1382186222 M * Bertl_oO when the kernel is working correctly (what kernel/patch/util-vserver version do you use?) and the guest is configured properly (no host IPs assigned) then there should be no way to get a host IP inside a guest 1382186326 M * huser I'm using 3.6.11 with 2.3.4.6 1382186360 M * huser and none of the host IPs are assigned within the vserver, or do you mean something more? 1382186373 M * Bertl_oO so what does 'ip a l 1382186381 M * Bertl_oO inside the guest show for example? 1382186436 M * huser hm don't think I have iproute installed, ifconfig shows (cutting it down): 1382186442 M * huser dummy0: flags=195 mtu 1500 inet 10.1.1.3 netmask 255.255.255.0 broadcast 10.1.1.255 1382186445 M * huser lo: flags=73 mtu 16436 inet 127.0.0.1 netmask 255.0.0.0 1382186483 M * Bertl_oO well, ifconfig doesn't see everything, so you probably need to get ip or collect the information from /proc yourself 1382186494 M * huser okay let me install iproute 1382186761 M * huser 1: lo: mtu 16436 qdisc noqueue state UNKNOWN link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo 1382186764 M * huser 5: dummy0: mtu 1500 qdisc noqueue state UNKNOWN link/ether 1a:40:dd:7d:89:9f brd ff:ff:ff:ff:ff:ff inet 10.1.1.3/24 brd 10.1.1.255 scope global dummy0 1382186821 M * Bertl_oO so that looks fine for a start, can you trace (strace) hadoop and check where and how it gets the host IP? 1382186837 M * huser it's a java program 1382186853 M * huser I've been thumbing through some of the source code, but I'm doubting I can figure out what the jvm is calling 1382186862 M * Bertl_oO still it has to use the kernel interface :) 1382186874 M * huser I'll give it a shot 1382186902 Q * Aiken Remote host closed the connection 1382186929 M * Bertl_oO I'm not so much interested in where or why hadoop figures this out but more in how it gets the information from the kernel (if it does get it from the kernel at all) 1382186948 M * Bertl_oO it might simply be in some config file or handed out by a dns lookup or similar 1382186975 M * huser understood 1382186984 M * Bertl_oO (strace -fF should help) 1382188893 M * huser strace didn't yield a lot, trying ltrace 1382188902 M * huser I see an accept call with the host IP, though 1382189856 M * huser ltrace won't cooperate.. not real sure what's going on 1382189888 M * huser but I don't see it looking up the host IP 1382190538 M * huser well, all of the services seem to have the host IP only in the accept call 1382196250 Q * Ghislain Quit: Leaving. 1382196290 Q * Walex Ping timeout: 480 seconds 1382198484 J * Walex ~Walex@cl-203.lon-02.gb.sixxs.net 1382199459 N * l0kit Guest2859 1382199464 J * l0kit ~1oxT@0001b54e.user.oftc.net 1382199861 Q * Guest2859 Ping timeout: 480 seconds 1382202225 J * bonbons ~bonbons@2001:a18:224:e01:71ff:a6ca:b4dd:83ce 1382204216 M * Bertl_oO huser: maybe the host IP is sent from the connecting end point? 1382212305 J * treaki adbbcf03e2@p4FDF6CDB.dip0.t-ipconnect.de 1382212649 Q * treaki__ Ping timeout: 480 seconds 1382213794 J * Aiken ~Aiken@2001:44b8:2168:1000:21f:d0ff:fed6:d63f 1382215609 M * huser Bertl_oO: it's all between services within the guest 1382215642 M * Bertl_oO so the service inside the guest creates a socket on a guest IP 1382215653 M * huser yes 1382215658 M * Bertl_oO but then accepts a connection with a host IP which is not coming from the host? 1382215670 M * huser it seems that way according to strace 1382215687 M * Bertl_oO okay, could you upload the relvant strace portions somewhere? 1382215687 M * huser each of these services creates a listen socket and then they connect to one another 1382215702 M * huser hm where'd be good? 1382215732 M * huser I've been running them as root, BTW.. going to try it as a non-root user too 1382215788 M * Bertl_oO any pastebin would be fine I guess 1382215996 M * Bertl_oO root inside a guest shouldn't be a problem 1382216007 M * Bertl_oO btw, I didn't get what util-vserver version you are using 1382216023 M * daniel_hozac are you using NAT? 1382216061 M * Bertl_oO good point, it could be that the connections are nat-ed because of incorrect iptable rules 1382216065 M * huser daniel_hozac: yes 1382216073 M * huser hold on I'll post the rule 1382216137 M * huser util-vserver-0.30.216_pre3025 1382216159 M * huser SNAT all -- 10.1.1.0/24 0.0.0.0/0 to:192.168.1.1 1382216233 M * huser http://pastebin.com/mVv6QYLL 1382216281 M * Bertl_oO and 10.1.1.x are guest or host IPs? 1382216285 M * huser guest 1382216296 M * huser that's in the POSTROUTING chain 1382216370 M * huser whoops, posted the wrong clone.. but the idea is the same, it creates a child process 1382216454 M * Bertl_oO well, looks good then 1382216486 M * Bertl_oO i.e. the SNAT translates the source to the host IP 1382216497 M * Bertl_oO so the connection appears to come from the host 1382216507 M * huser hmm 1382216527 M * Bertl_oO try to restrict the SNAT to non guest IPs 1382216532 M * huser so the iptables rules on the host even apply to completely-within-the-guest activities? 1382216535 M * Bertl_oO or to a non lo interface 1382216543 M * huser okay 1382217214 M * huser that looks like the problem 1382217216 M * huser thanks! 1382217231 M * huser (now I have other one, but at least that one seems to be gone) 1382217285 J * JonB ~NoSuchUse@212-60-115-150.ip.cust.zensystems.net 1382217385 J * cuba33ci_ ~cuba33ci@114-36-246-117.dynamic.hinet.net 1382217390 M * Bertl_oO you're welcome! daniel_hozac saved the day :) 1382217446 Q * cuba33ci Read error: Connection reset by peer 1382217463 N * cuba33ci_ cuba33ci 1382218796 M * JonB he did? 1382218804 M * JonB maybe he can save my day? 1382218915 M * Bertl_oO I'm positive he can 1382218932 M * JonB I upgraded to home compiled linux-3.2.51 but I got this error message trying to start http://linux-vserver.org/Frequently_Asked_Questions#When_I_try_to_start_a_guest_i_get_this_message_.22vsched:_vc_set_sched.28.29:_Function_not_implemented.22. 1382218943 M * JonB trying to start clients 1382218973 M * Bertl_oO that is already weird ... 1382218974 M * JonB with util-vserver package from debian sid 1382218991 M * Bertl_oO you have upgraded and home compiled in the same line as linux 3.2.51 :) 1382218997 M * JonB I could not find any /etc/vservers/*/sched/ 1382219015 M * JonB well, the old kernel was Linux dkVserver 2.6.32-5-vserver-amd64 #1 SMP Mon Sep 23 23:03:09 UTC 2013 x86_64 GNU/Linux 1382219043 M * Bertl_oO yes, scheduler has been 'replaced' by cgroup magic 1382219054 M * JonB which I run again now so I can start some clients 1382219075 M * Bertl_oO what is your util-vserver version? 1382219091 M * JonB 0.30.216-pre2864-2.1 1382219110 M * Bertl_oO daniel_hozac: please go ahead and save his day too :) 1382219121 M * JonB no, not today 1382219132 M * JonB it is 23:45 and I have been at work for 12+ hours 1382219146 M * Bertl_oO guess that can be done in a few minutes 1382219153 M * JonB I hope so 1382219164 M * JonB but I would also have to boot into the newer kernel 1382219294 M * JonB i'll go make a sandwich, back in 5 1382219403 M * daniel_hozac upgrading your utils should fix that. 1382219487 M * JonB from where? 1382219501 M * JonB I tried with utils from debian sid 1382219503 Q * Walex 1382219598 M * JonB [ ] util-vserver-0.30.216-pre2864.tar.bz2 25-Nov-2009 13:53 712K Patches 1382219607 M * JonB wooow package in sid is oooold 1382219634 M * Bertl_oO try building from the source or grab one of the beng packages 1382219830 M * JonB beng packages? 1382220633 M * JonB time to go home, be well until my next visit 1382220651 Q * JonB Quit: This computer has gone to sleep 1382223696 J * treaki__ b0fd6bc3db@p4FDF6D3C.dip0.t-ipconnect.de 1382224118 Q * treaki Ping timeout: 480 seconds 1382227175 J * treaki_ 36f939d7f1@p4FF4BF44.dip0.t-ipconnect.de