I just tested this out again on a newly installed machine. I have the following in my vz.conf:

IPTABLES="ipt_REJECT ipt_tos ipt_limit ipt_multiport iptable_filter iptable_mangle ipt_TCPMSS ipt_tcpmss ipt_ttl ipt_length ipt_LOG ipt_conntrack ipt_state xt_connlimit ipt_connlimit"

IPTABLES_MODULES="$IPTABLES"

I have tried just "ipt_connlimit", just "xt_connlimit, and both in the IPTABLES list. I even rebooted the machine before I started each test. I get the same results. The connlimit rule is not tripped by multiple SSH connections as it should be.

Dan

On 06/21/2013 05:55 PM, Kir Kolyshkin wrote:


On 21 June 2013 08:09, Dan Bassett <[email protected] <mailto:[email protected]>> wrote:

    I am running 2.6.32-042stab076.8 on a CentOS6 HN with CentOS6
    containers, vzctl-4.2-1.x86_64. I have two containers running on
    the same HN. Each CT has a veth device. Both CT's veth device is
    connected to the same bridge device on the HN. They are assigned
    the addresses 192.168.0.1/24 <http://192.168.0.1/24> and
    192.168.0.2/24 <http://192.168.0.2/24> and can ping each other.

    On the HN I am able to modprobe the connlimit module (xt_connlimit).


You are supposed to load it before loading VZ modules (and this is the reason of having it listed in vz.conf so vz initscript takes care of loading it.

If you loaded it after loading vz modules it might not work. So, to make sure you'd do a reboot (or reloading of all vz modules -- /etc/init.d/vz stop, then make sure vz modules such as vznet are not loaded, then /etc/init.d/vz start).

    I can then use iptables rules on the hostnode such as:

    iptables -A INPUT -p tcp --syn --dport 22 -m connlimit
    --connlimit-above 1 --connlimit-mask=32 -j REJECT

    If I then attempt to initiate more than one ssh session to this HN
    from the same host, the second (and all subsequent) ssh
    connections are rejected.  This leads me to believe that the
    connlimit module is working properly with the OpenVZ kernel.

    I am able to successfully insert the same rule into the iptables
    configuration in either of my containers, so it seems that the
    xt_connlimit module is properly loaded (I get errors if I try to
    use an iptables module that has not been loaded on the hostnode).
     However, I can create more than one successful ssh session,
    meaning that the iptables rule in the CT is not being matched for
    some reason. I have both "xt_limit" and "xt_connlimit" in the list
    of iptables modules to load in vz.conf. When I am running these
    tests, I have no iptables rules loaded on the HN, and I have no
    other iptables rules loaded in the container.

    One interesting thing I have noticed is that when I enter a
    container on this hostnode, I get the following errors:

    [root@hn0 ~]# vzctl enter 2033892
    Warning: Unknown iptable module: xt_connlimit, skipped
    Warning: Unknown iptable module: xt_limit, skipped

    It seems like this might be related to my issue, but it might just
    be a red herring.



This warning is harmless, it is not a bug, details are here:
http://git.openvz.org/?p=vzctl;a=commit;h=d284c8a7

Or see vz.conf(5) man page, description of IPTABLES and IPTABLES_MODULES.


    Am I missing something obvious here? Or is this a bug with OpenVZ?

    Thanks,
    Dan
    _______________________________________________
    Users mailing list
    [email protected] <mailto:[email protected]>
    https://lists.openvz.org/mailman/listinfo/users




_______________________________________________
Users mailing list
[email protected]
https://lists.openvz.org/mailman/listinfo/users
_______________________________________________
Users mailing list
[email protected]
https://lists.openvz.org/mailman/listinfo/users

Reply via email to