Такой вот пинг:
64 bytes from 10.10.7.1: icmp_seq=39 ttl=255 time=0.026 ms
64 bytes from 10.10.7.1: icmp_seq=40 ttl=255 time=0.027 ms
64 bytes from 10.10.7.1: icmp_seq=41 ttl=255 time=500.188 ms
64 bytes from 10.10.7.1: icmp_seq=42 ttl=255 time=0.027 ms
64 bytes from 10.10.7.1: icmp_seq=43 ttl=255 time=1548.026 ms

netstat -m
1034/1666/2700 mbufs in use (current/cache/total)
1032/800/1832/25600 mbuf clusters in use (current/cache/total/max)
1032/760 mbuf+clusters out of packet secondary zone in use (current/cache)
0/0/0/0 4k (page size) jumbo clusters in use (current/cache/total/max)
0/0/0/0 9k jumbo clusters in use (current/cache/total/max)
0/0/0/0 16k jumbo clusters in use (current/cache/total/max)
2322K/2016K/4339K bytes allocated to network (current/cache/total)
0/0/0 requests for mbufs denied (mbufs/clusters/mbuf+clusters)
0/0/0 requests for jumbo clusters denied (4k/9k/16k)
0/4/6656 sfbufs in use (current/peak/max)
0 requests for sfbufs denied
0 requests for sfbufs delayed
0 requests for I/O initiated by sendfile
0 calls to protocol drain routines

# netstat -I em1 -w 1
            input          (em1)           output
   packets  errs      bytes    packets  errs      bytes colls
      4148     0    1305953       9491     0    5602805     0
      3710     0    1216925       8281     0    4805680     0
      3935     0    1367981       8311     0    4696024     0

# netstat -iI em1
Name    Mtu Network       Address              Ipkts Ierrs    Opkts Oerrs  Coll
em1    1500 <Link#4>      00:11:0a:5a:5e:5b  8829685    99 18078164     0     0
em1    1500 10.10.7/24    10.10.7.1          4974078     - 12704744     -     -

Ядро собрано с smp, 2 процессора.
Замирание пинга проходит одновременно на всех 4 сетевых картах (2 em и 2 bge).
Загрузка процессора не более 20%. Сетевые в пиках не более 50-60 мбит, в среднем 20 мбит.

Включен polling, без него так же все.
Чего может быть такого?