[Planetlab-devel] Raw sockets on 2.6.22 + vserver

[Faiyaz Ahmed]

I'm seeing this now too.

Same vserver, but i rebooted the machine.  Now I see the following

[pl_sirius at houston-vm2 ~]$ ping 172.20.0.1
ping: icmp open socket: Operation not permitted

172.20.0.1 is the host running the node using qemu.  The node is 172.20.0.2.

...This is weird...


F

Andy Bavier wrote:
> I can ping the local IP address but nothing else:
> 
> [pl_acb1 at vici-02 ~]$ uname -a
> Linux vici-02.cs.princeton.edu 2.6.22.14-vs2.3.0.29.1.planetlab #1 SMP
> Wed Jan 23 02:30:04 EST 2008 i686 i686 i386 GNU/Linux
> 
> [pl_acb1 at vici-02 ~]$ ping -c 1 vici-02
> PING vici-02.cs.princeton.edu (128.112.139.25) 56(84) bytes of data.
> 64 bytes from vici-02.cs.princeton.edu (128.112.139.25): icmp_seq=1
> ttl=64 time=0.014 ms
> --- vici-02.cs.princeton.edu ping statistics ---
> 1 packets transmitted, 1 received, 0% packet loss, time 0ms
> rtt min/avg/max/mdev = 0.014/0.014/0.014/0.000 ms
> 
> [pl_acb1 at vici-02 ~]$ ping localhost
> connect: Operation not permitted
> 
> [pl_acb1 at vici-02 ~]$ ping -c 1 www.cs.princeton.edu
> PING coreweb.cs.princeton.edu (128.112.136.35) 56(84) bytes of data.
> --- coreweb.cs.princeton.edu ping statistics ---
> 1 packets transmitted, 0 received, 100% packet loss, time 0ms
> 
> Andy
> 
> On Jan 28, 2008 4:59 PM, Faiyaz Ahmed <faiyaza at cs.princeton.edu> wrote:
>> Actually, ICMP works for me.
>>
>> [pl_sirius at houston-vm2 ~]$ ping 172.20.0.1
>> PING 172.20.0.1 (172.20.0.1) 56(84) bytes of data.
>> 64 bytes from 172.20.0.1: icmp_seq=1 ttl=64 time=0.787 ms
>> 64 bytes from 172.20.0.1: icmp_seq=4 ttl=64 time=1.32 ms
>>
>>
>> Faiyaz
>>
>>
>> Faiyaz Ahmed wrote:
>>> Hi,
>>>
>>> I was wondering what the status of the kernel patches were with respect
>>> to raw sockets.
>>>
>>> It seemed the old vnet would allow slices to use raw sockets only if the
>>> packet could be demuxed back to the slice.  As such, it seems like TCP,
>>> UDP, and ICMP were supported where the first two would cause vnet to
>>> auto bind the socket to the slice if the port was already owned by the
>>> slice.
>>>
>>> This seems like an important detail and Marc was under the impression
>>> that this was discussed and addressed in some set of patches that may or
>>> may not have been applied to the kernel.
>>>
>>> Does anyone know if raw sockets (used to create certain packets) are
>>> supported by the current kernel?  Is there some extra setup procedure
>>> that we currently don't do which is causing ICMP from a slice to not work?
>>>
>>> Thanks in advance.
>>>
>>>
>>> Faiyaz
>>>
>>> _______________________________________________
>>> Devel mailing list
>>> Devel at lists.planet-lab.org
>>> https://lists.planet-lab.org/mailman/listinfo/devel
>> _______________________________________________
>> Devel mailing list
>> Devel at lists.planet-lab.org
>> https://lists.planet-lab.org/mailman/listinfo/devel
>>
> 
> _______________________________________________
> Devel mailing list
> Devel at lists.planet-lab.org
> https://lists.planet-lab.org/mailman/listinfo/devel