[Xorp-users] kernel msg "mroute: pending queue full, dropping entries."

[Štefan Gula]

Hi,

I am currently also receiving that message. I am not sure what cause
this problem but I am sure it is done byt XORP. I am running five XORP
instances with more than 3000 users behind them so it is really
problematic to find out what can cause this.

fw1 mld6igmp # netstat -alnp | grep raw
raw   109500      0 0.0.0.0:2               0.0.0.0:*               7
        20285/xorp_fea
raw        0      0 0.0.0.0:2               0.0.0.0:*               7
        20285/xorp_fea
raw   109340      0 0.0.0.0:89              0.0.0.0:*               7
        20285/xorp_fea
raw        0      0 0.0.0.0:89              0.0.0.0:*               7
        20285/xorp_fea
raw   109508      0 0.0.0.0:103             0.0.0.0:*               7
        20285/xorp_fea
raw        0      0 0.0.0.0:103             0.0.0.0:*               7
        20285/xorp_fea
raw6  109276      0 :::89                   :::*                    7
        20285/xorp_fea
raw6       0      0 :::89                   :::*                    7
        20285/xorp_fea

According to output above, there is relatively enough packets in
queues. I have found on other multicast implementation the same bug
and it was fixed by simply periodical reading from given sockets
(http://www.cschill.de/smcroute/) and discarding the packets. I was
trying to create a patch but as I am not familiar with XORP code, its
really problematic to find out.

Any ideas how to fix it?

kernel version: 2.6.25-hardened-r9
XORP version: 1.6

-- 
Stefan Gula
CCNP, CCIP