iptables sporadic "sendmsg: operation not permitted" problem and packet loss

iptables sporadic "sendmsg: operation not permitted" problem and packet loss

[Dennis Jacobfeuerborn]

Hi,

I'm running into a problem on a machine that right now acts as a simple 
gateway but is supposed to become a firewall too. When I start iptables 
using "/etc/init.d/iptables start" on the Centos 5.2 machine first 
everything works fine but after about 30 seconds I'm seeing packet loss and 
running a ping outputs "sendmsg: operation not permitted" sporadically.
The moment I stop iptables again everything returns to normal. What is 
consufing to me is that I don't even have any rules defined so far. This is 
what my "/etc/sysconfig/iptables" file looks like:

# Generated by iptables-save v1.3.5 on Thu Mar  5 17:40:28 2009
*filter
:INPUT ACCEPT [26715202:4750206096]
:FORWARD ACCEPT [1382646771:1563210213960]
:OUTPUT ACCEPT [22930985:6256734041]
COMMIT

iptables -L says:

Chain INPUT (policy ACCEPT)
target     prot opt source               destination

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination

Does anyone have an idea why that would have such a severe impact on the 
traffic? The fact that it takes a moment for the problems to show up makes 
me suspect some kind of buffer issue so that the packet loss only begins to 
occur after some buffer begins to overflow. That just a guess though and I 
have no idea what buffer that could be.

Regards,
   Dennis

Re: iptables sporadic "sendmsg: operation not permitted" problem and packet loss

[Gáspár Lajos]

Dennis Jacobfeuerborn írta:
> Hi,
>
> I'm running into a problem on a machine that right now acts as a 
> simple gateway but is supposed to become a firewall too. When I start 
> iptables using "/etc/init.d/iptables start" on the Centos 5.2 machine 
> first everything works fine but after about 30 seconds I'm seeing 
> packet loss and running a ping outputs "sendmsg: operation not 
> permitted" sporadically.
> The moment I stop iptables again everything returns to normal. What is 
> consufing to me is that I don't even have any rules defined so far. 
> This is what my "/etc/sysconfig/iptables" file looks like:
>
> # Generated by iptables-save v1.3.5 on Thu Mar  5 17:40:28 2009
> *filter
> :INPUT ACCEPT [26715202:4750206096]
> :FORWARD ACCEPT [1382646771:1563210213960]
> :OUTPUT ACCEPT [22930985:6256734041]
> COMMIT
>
> iptables -L says:
>
> Chain INPUT (policy ACCEPT)
> target     prot opt source               destination
>
> Chain FORWARD (policy ACCEPT)
> target     prot opt source               destination
>
> Chain OUTPUT (policy ACCEPT)
> target     prot opt source               destination
>
> Does anyone have an idea why that would have such a severe impact on 
> the traffic? The fact that it takes a moment for the problems to show 
> up makes me suspect some kind of buffer issue so that the packet loss 
> only begins to occur after some buffer begins to overflow. That just a 
> guess though and I have no idea what buffer that could be.
>
> Regards,
>   Dennis
>

Hi Dennis,

What about the other tables?

iptables -vnL -t raw
iptables -vnL -t mangle
iptables -vnL -t nat

Swifty

Re: iptables sporadic "sendmsg: operation not permitted" problem and packet loss

[Dennis J.]

On 03/09/2009 02:33 PM, Gáspár Lajos wrote:
> Dennis Jacobfeuerborn írta:
>> Hi,
>>
>> I'm running into a problem on a machine that right now acts as a
>> simple gateway but is supposed to become a firewall too. When I start
>> iptables using "/etc/init.d/iptables start" on the Centos 5.2 machine
>> first everything works fine but after about 30 seconds I'm seeing
>> packet loss and running a ping outputs "sendmsg: operation not
>> permitted" sporadically.
>> The moment I stop iptables again everything returns to normal. What is
>> consufing to me is that I don't even have any rules defined so far.
>> This is what my "/etc/sysconfig/iptables" file looks like:
>>
>> # Generated by iptables-save v1.3.5 on Thu Mar 5 17:40:28 2009
>> *filter
>> :INPUT ACCEPT [26715202:4750206096]
>> :FORWARD ACCEPT [1382646771:1563210213960]
>> :OUTPUT ACCEPT [22930985:6256734041]
>> COMMIT
>>
>> iptables -L says:
>>
>> Chain INPUT (policy ACCEPT)
>> target prot opt source destination
>>
>> Chain FORWARD (policy ACCEPT)
>> target prot opt source destination
>>
>> Chain OUTPUT (policy ACCEPT)
>> target prot opt source destination
>>
>> Does anyone have an idea why that would have such a severe impact on
>> the traffic? The fact that it takes a moment for the problems to show
>> up makes me suspect some kind of buffer issue so that the packet loss
>> only begins to occur after some buffer begins to overflow. That just a
>> guess though and I have no idea what buffer that could be.
>>
>> Regards,
>> Dennis
>>
>
> Hi Dennis,
>
> What about the other tables?
>
> iptables -vnL -t raw
> iptables -vnL -t mangle
> iptables -vnL -t nat

All clear:

[root@gw ~]# iptables -vnL -t raw
Chain PREROUTING (policy ACCEPT 132 packets, 77370 bytes)
  pkts bytes target     prot opt in     out     source 
destination

Chain OUTPUT (policy ACCEPT 0 packets, 0 bytes)
  pkts bytes target     prot opt in     out     source 
destination

[root@gw ~]# iptables -vnL -t mangle
Chain PREROUTING (policy ACCEPT 2688M packets, 1487G bytes)
  pkts bytes target     prot opt in     out     source 
destination

Chain INPUT (policy ACCEPT 2048K packets, 368M bytes)
  pkts bytes target     prot opt in     out     source 
destination

Chain FORWARD (policy ACCEPT 2686M packets, 1486G bytes)
  pkts bytes target     prot opt in     out     source 
destination

Chain OUTPUT (policy ACCEPT 1961K packets, 997M bytes)
  pkts bytes target     prot opt in     out     source 
destination

Chain POSTROUTING (policy ACCEPT 2688M packets, 1487G bytes)
  pkts bytes target     prot opt in     out     source 
destination

[root@gw ~]# iptables -vnL -t nat
Chain PREROUTING (policy ACCEPT 178 packets, 40798 bytes)
  pkts bytes target     prot opt in     out     source 
destination

Chain POSTROUTING (policy ACCEPT 121 packets, 26962 bytes)
  pkts bytes target     prot opt in     out     source 
destination

Chain OUTPUT (policy ACCEPT 0 packets, 0 bytes)
  pkts bytes target     prot opt in     out     source 
destination

Regards,
   Dennis

Re: iptables sporadic "sendmsg: operation not permitted" problem and packet loss

[Mart Frauenlob]

netfilter-owner@vger.kernel.org wrote:
> Hi,
>
> I'm running into a problem on a machine that right now acts as a 
> simple gateway but is supposed to become a firewall too. When I start 
> iptables using "/etc/init.d/iptables start" on the Centos 5.2 machine 
> first everything works fine but after about 30 seconds I'm seeing 
> packet loss and running a ping outputs "sendmsg: operation not 
> permitted" sporadically.
> The moment I stop iptables again everything returns to normal. What is 
> consufing to me is that I don't even have any rules defined so far. 
> This is what my "/etc/sysconfig/iptables" file looks like:
>
> # Generated by iptables-save v1.3.5 on Thu Mar  5 17:40:28 2009
> *filter
> :INPUT ACCEPT [26715202:4750206096]
> :FORWARD ACCEPT [1382646771:1563210213960]
> :OUTPUT ACCEPT [22930985:6256734041]
> COMMIT
>
> iptables -L says:
>
> Chain INPUT (policy ACCEPT)
> target     prot opt source               destination
>
> Chain FORWARD (policy ACCEPT)
> target     prot opt source               destination
>
> Chain OUTPUT (policy ACCEPT)
> target     prot opt source               destination
>
> Does anyone have an idea why that would have such a severe impact on 
> the traffic? The fact that it takes a moment for the problems to show 
> up makes me suspect some kind of buffer issue so that the packet loss 
> only begins to occur after some buffer begins to overflow. That just a 
> guess though and I have no idea what buffer that could be.
>
> Regards,
>   Dennis
>
Hello,

sounds like something that happend to me, when I accidentally  did set 
ip_conntrack_max to `1'.
If you don't know where it's located do some like: find /proc/sys/net 
-name ip_conntrack_max

Maybe it's that issue, maybe not - just a guess.

Greets

Mart