[Xorp-users] xorp clearing routing table upon startup

Fri Oct 10 14:20:52 PDT 2008

Stegen Smith <stegen at owns.com> wrote:

> Hi Pavlin,
> 
> Yeah, it seems to occur when things sit idly.  Ultimately something gums up the works and I get
> those error messages, so it's not only when I modified the routing table myself.  It should be
> easily recreated.  Odd... So with that, I'll get one submitted right away with more detailed
> information.

I just saw the entry in Bugzilla. Thanks!

The missing routes might be related to the XRL errors.
The "life timer expired" indicates that the XRL communication has
timed-out, and after that all bets are off.
BTW, this particular timeout shouldn't happen under (very relaxed)
normal circumstances, so there is something else happening.
This probably also explains the shutdown problem with the leftover
XORP processes.

Re. the shutdown question: yes, if you are running XORP in
foreground, Ctrl-C should be the way to stop it.

I (or somebody else) should look at this problem very soon, because
it looks critical.

Pavlin

> Another note...  I've got bgp peering setup and I see the following:
> xorp at at-ig1.sv.atomz.com> show bgp routes
> Status Codes: * valid route, > best route
> Origin Codes: i IGP, e EGP, ? incomplete
> 
>    Prefix                Nexthop                    Peer            AS Path
>    ------                -------                    ----            -------
> *  10.3.48.0/20          10.3.14.39                 10.3.14.99    65000 i
> *  10.5.49.0/24          10.3.14.39                 10.3.14.99    65000 i
> *  10.3.48.0/20          10.3.14.7                  10.3.14.99    65000 i
> *  10.5.49.0/24          10.3.14.7                  10.3.14.99    65000 i
> 
> yet no routes are inserted on the system:
> root1paat-ig1.sv~ netstat -nr|grep 10.3.48
> root1paat-ig1.sv~
> 
> I'm guessing that this is another bug, or perhaps related to the first issue I've noticed.  I'll
> submit a bug for the first one, and if you would like me to do the same for this issue I'm seeing,
> let me know. =)
> 
> And one more side note... what's the cleanest way to shutdown the routing software?  I've been
> letting xorp_rtrmgr run in the foreground while I do testing, and ^C when I want to quit out.
> Probably not the best way to do it, as it ends up crashing (cores) and leaving other xorp processes
> up and running:
> 
> ^C[ 2008/10/10 11:13:26  INFO xorp_rtrmgr:8398 RTRMGR +1024 task.cc shutdown ] Shutting down module: bgp
> [ 2008/10/10 11:13:26  INFO xorp_bgp:18890 BGP +451 peer.cc send_notification ] Sending:
> Notification Packet: Cease(6)
> [ 2008/10/10 11:13:26  TRACE xorp_bgp:18890 BGP +456 peer.cc send_notification ] Peer
> {10.3.14.40(179) 10.3.14.39(179)}: Send: Notification Packet: Cease(6)
> 
> <cleared some cruft for email brevity>
> 
> [ 2008/10/10 11:13:28  INFO xorp_rtrmgr:8398 XRL +432 xrl_router.cc get_sender ] Sender died
> (protocol = "stcp", address = "127.0.0.1:11810")
> [ 2008/10/10 11:13:28  ERROR xorp_rtrmgr:8398 LIBCOMM +610 comm_sock.c comm_sock_connect4 ] Error
> connecting socket (family = 2, remote_addr = 127.0.0.1, remote_port = 11810): Connection refused
> [ 2008/10/10 11:13:28  ERROR xorp_rtrmgr:8398 XRL +63 xrl_pf_factory.cc create_sender ]
> XrlPFSenderFactory::create failed: XrlPFConstructorError from line 668 of xrl_pf_stcp.cc: Could not
> connect to 127.0.0.1:11810
> 
> [ 2008/10/10 11:13:28  ERROR xorp_rtrmgr:8398 XRL +451 xrl_router.cc get_sender ] Could not create
> XrlPFSender for protocol = "stcp" address = "127.0.0.1:11810"
> [1]    8398 segmentation fault (core dumped)  ./bin/xorp_rtrmgr
> root139p9at-ig1.sv/usr/local/xorp
> root139p9at-ig1.sv/usr/local/xorp ps auxww|grep xorp
> xorp     20544  0.0  0.1  3516  5484 p1  I+     5:21PM    0:00.35 /usr/local/xorp/bin/xorpsh
> root      2811  0.0  0.2  5076  6476 p9  S     11:09AM    0:00.75 xorp_fea
> root     17533  0.0  0.1  2812  4452 p9  S     11:10AM    0:00.22 xorp_rib
> root     16731  0.0  0.1  2188  3872 p9  I     11:10AM    0:00.18 xorp_fib2mrib
> root      9111  0.0  0.1  2728  4268 p9  S     11:10AM    0:00.20 xorp_igmp
> root     32406  0.0  0.1  4752  6028 p9  S     11:10AM    0:00.50 xorp_pimsm4
> root      4331  0.0  0.0   848   692 p9  R+/1  11:16AM    0:00.00 grep xorp (zsh)
> root0p9at-ig1.sv/usr/local/xorp
> 
> I'm just working hard at keeping you guys busy ;)
> 
> thanks,
> stegen
> 
> Pavlin Radoslavov wrote:
> > Stegen Smith <stegen at owns.com> wrote:
> > 
> >> Also an interesting thing to note is that if a routing entry is added manually, I see the following happening:
> >>
> >> [ 2008/10/09 16:31:35  ERROR xorp_fea:15980 XRL +379 xrl_pf_stcp.cc die ] STCPRequestHandler died: life timer expired
> >> [ 2008/10/09 16:31:35  ERROR xorp_fea:15980 XRL +379 xrl_pf_stcp.cc die ] STCPRequestHandler died: life timer expired
> > 
> > This issue is probably independent from the routing table cleanup,
> > hence please submit a Bugzilla entry.
> > 
> > Thanks,
> > Pavlin
> > 
> >> And also, I forgot to give the following information... I'm using the latest xorp build from cvs, and this is openbsd 4.3.
> > 
> 
> _______________________________________________
> Xorp-users mailing list
> Xorp-users at xorp.org
> http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users