proxy-arp & mpd

[Damien Deville]

Hi,

we are facing a similar issue with arp blocked in sbwait state.

Here is a way to reproduce it:
- add a bunch of arp entries in your arp table (best is around 255 entries).
- launch two arp -a -d in parallel ('arp -a -d & arp -a -d &')

Both processes will be in concurence to access the table. One process 
will successfully nuke all entries of the arp table, the other one will 
be blocked in rtmsg function on the read while executing a RTM_GET or 
RTM_DELETE command after some time. By instrumenting arp we noticed that 
it happened when both process access to the same entry.

Here is a backtrace of the blocked arp on FreeBSD 7.0

(gdb) bt
#0  0x28158f81 in read () from /lib/libc.so.7
#1  0x08049091 in rtmsg ()
#2  0x08049b44 in delete ()
#3  0x0804a1fd in nuke_entry ()
#4  0x08049a77 in search ()
#5  0x08049e75 in main ()

I can reproduce this on FreeBSD 4.11, 6.2 and 6.3, and FreeBSD 7.0.

Damien Deville

Oleksandr Samoylyk wrote:
> Oleksandr Samoylyk пишет:
>> Dear Community,
>>
>> I'm using proxy-arp for public ips for our clients in order to give 
>> them internet access using pptp-tunnels with mpd:
>>
>> # cat /usr/local/etc/mpd5/mpd.conf | grep arp
>>         set iface enable proxy-arp
>>
>> # uname -a
>> FreeBSD xxx.xxxxxx.xxx 7.0-STABLE FreeBSD 7.0-STABLE #0: Tue May 13 
>> 06:54:41 EEST 2008     root at xxx.xxxxxx.xxx:/usr/obj/usr/src/sys/XXX  
>> amd64
>>
>> Sometimes I see hanged arps:
>>
>> root   72148  0.0  0.0  2564   788  ??  S     9:33PM   0:20.69 
>> /usr/sbin/arp -S xx.xxx.xxx.xx 0:e:c:70:c6:4c pub
>>
>> where xx.xxx.xxx.xx is public ip address
> 
> as well:
> 
> root   58831  0.0  0.0  2564   788  ??  S    11:58PM   0:01.04 
> /usr/sbin/arp -d xx.xxx.xxx.xx
> 
> 
> 
>> In such moments clients can't connect to pptp-server (e.g., Error 800 
>> for Windowz clients).
>>
>> Why does that happen? How can I prevent such situations?
>>
>>
>> Thank you!
>>
> 
> 

-- 
Damien Deville
R&D engineer
damien.deville at netasq.com
http://www.netasq.com