[Xorp-users] FATAL xorp_fea ERROR
Vikram KAUL
vkaul at research.telcordia.com
Tue Jul 18 09:07:57 PDT 2006
Thanks for the replies. My comments interspersed..
pavlin>The MFC entry addition is probably bogus (see my reply to your
pavlin>earlier email). The real issue is why the PIM-SM Join message wasn't
pavlin>generated when the receiver joined.
I have sent a detailed email, along with log and xorpsh outputs detailing
that issue. Thanks again for the pointer on the bogus entry.
pavlin>> Why does the fea error occur ?
pavlin>First, could you double-check that HAVE_IF_INDEXTONAME inside
pavlin>xorp/config.h is defined:
pavlin>/* Define to 1 if you have the `if_indextoname' function. */
pavlin>#define HAVE_IF_INDEXTONAME 1
Yes. It is set to 1
pavlin>
pavlin>
pavlin>If yes, could you send the output of the following programs (after
pavlin>XORP is started):
pavlin>
pavlin>UNIX: cat /proc/net/ip_mr_vif
pavlin>Xorpsh: show interfaces
pavlin> show mfea interface
pavlin> show pim interface
[root at mobile17 ~]# cat /proc/net/ip_mr_vif
Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote
0 vf0 480 15 0 0 00000 1164A8C0 00000000
1 pimreg 0 0 0 0 00004 1164A8C0 00000000
[root at mobile17 ~]# cat /proc/net/ip_mr_cache
Group Origin Iif Pkts Bytes Wrong Oifs
root at mobile17> show interfaces
vf0/vf0: Flags:<ENABLED,BROADCAST,MULTICAST> mtu 1500
inet6 fe80::4c52:4cff:fe63:6a91 prefixlen 64
inet 192.168.100.17 subnet 192.168.100.0/24 broadcast 192.168.100.255
physical index 21
ether 4e:52:4c:63:6a:91
root at mobile17> show pim interface
Interface State Mode V PIMstate Priority DRaddr Neighbors
vf0 UP Sparse 2 NotDR 1 192.168.100.21 1
register_vif UP Sparse 2 DR 1 192.168.100.17 0
root at mobile17> show mfea interface
Interface State Vif/PifIndex Addr Flags
vf0 UP 0/21 192.168.100.17 MULTICAST BROADCAST KERN_UP
register_vif UP 1/21 192.168.100.17 PIM_REGISTER KERN_UP
pavlin>
pavlin>Also, do you always get this XLOG_FATAL() message right after the
pavlin>PIM Register message is transmitted:
Nope. After a bunch of PIM REGISTERS. Probably, they were running for 10
minutes, before the error occured
pavlin>For some reason, XORP probably wanted to send Prune messages
pavlin>toward the RP and toward S, but it couldn't because there was no
pavlin>upstream PIM-SM neighor router. Given that those messages were
pavlin>generated right after the FEA crashed, all bets are off about which
pavlin>particular event triggered them.
Alright. Let me approach these problems one at a time. :-)
pavlin>
pavlin>> Why does joining the group not create an immediate PIM Join when the
pavlin>> router/node has been configured to act on only IGMP messages from itself and
pavlin>> each router/node is a DR ? This should ordinarily have worked in sending a PIM
pavlin>> join towards the next upstream router in direction of the RP ?
pavlin>
pavlin>This is the question we need to answer. Hopefully, the debug info I
pavlin>asked about in my earlier email will help us finding the answer.
I have sent that one. Thanks for the help
regards
Vikram
More information about the Xorp-users
mailing list