We usually disable zebra. Something like a route server. Not sure if this hits other folks as a default behavior, but... we get thousands of warning messages like:
sendmsg_nexthop: Can't send NH register, Zebra client not established Recommended approach on this one? John Kemp --- bgpd/bgp_nht.c: 387 static void 388 sendmsg_nexthop (struct bgp_nexthop_cache *bnc, int command) 389 { 390 struct stream *s; 391 struct prefix *p; 392 int ret; 393 394 /* Check socket. */ 395 if (!zclient || zclient->sock < 0) 396 { 397 zlog_debug("%s: Can't send NH register, Zebra client not established", 398 __FUNCTION__); 399 return; 400 } On 10/18/16 7:50 AM, Paul Jakma wrote: > Quagga 1.1.0 has been released, available from the usual place: > > https://download.savannah.gnu.org/releases/quagga/ > > This is a release with a number of new features, and many bug fixes. > Notably: > > * Greatly improved nexthop resolution for recursive routes. (Cumulus) > * Event driven nexthop resolution for BGP (Cumulus) > * Route tags support (Piotr Chytła, Packet Consulting) > * Transport of TE related metrics over OSPF, IS-IS (Olivier Dugeon, Orange) > * IPv6 Multipath for zebra and BGP (Ayan Banerjee, Cumulus) > > This release also changed the default of 'link-detect' state, > controlling whether zebra will respond to link-state events and > consider an interface to be down when link is down. To retain current > the behavior save your config before updating, otherwise remove the > 'link-detect' flag from your config prior to updating. There is also a > new global 'default link-detect (on|off)' flag to configure the global > default. > > This release includes the security fixes of the 1.0.20161017 release. > > Thanks to all the contributors, who include: > > Andrej Ota <and...@ota.si> > Avneesh Sachdev <avne...@sproute.com> > Ayan Banerjee <a...@cumulusnetworks.com> > Balaji <balaji...@gmail.com> > Boian Bonev <bbo...@ipacct.com> > boris yakubov <borisyaku...@ruggedcom.com> > Christian Franke <ch...@opensourcerouting.org> > Christian Franke <nob...@nowhere.ws> > Colin Petrie <cpet...@ripe.net> > Daniel Walton <dwal...@cumulusnetworks.com> > David Lamparter <equi...@opensourcerouting.org> > Denil Vira <de...@cumulusnetworks.com> > Dinesh Dutt <dd...@cumulusnetworks.com> > Donald Sharp <sha...@cumulusnetworks.com> > Evgeny Uskov <e...@qrator.net> > Igor Ryzhov <iryz...@nfware.com> > Jafar Al-Gharaibeh <ja...@atcorp.com> > James Li <j...@cumulusnetworks.com> > Jonathan Hart <j...@onlab.us> > kitty <khiruthigai.balasubraman...@hpe.com> > Lou Berger <lber...@labn.net> > Matthieu Boutier <bout...@pps.univ-paris-diderot.fr> > Olivier Dugeon <olivier.dug...@orange.com> > Paul Jakma <paul.ja...@hpe.com> > Paul Jakma <p...@jakma.org> > Paul Jakma <p...@opensourcerouting.org> > Pawel Wieczorkiewicz <pwieczorkiew...@suse.de> > Philippe Guibert <philippe.guib...@6wind.com> > Piotr Chytła <p...@packetconsulting.pl> > Pradosh Mohapatra <pmoha...@cumulusnetworks.com> > Roman Hoog Antink <r...@open.ch> > Stas Nichiporovich <s...@iptel.by> > Timo Teräs <timo.te...@iki.fi> > Udaya Shankara KS <shankara.k....@gmail.com> > Vipin Kumar <vi...@cumulusnetworks.com> > Vivek Venkatraman <vi...@cumulusnetworks.com> > > See the full changelog available at the above URL for further details. > A summary of the changes are as follows: > > bgp: add "debug bgp allow-martians" next hops and related code/commands > bgp: bgp_nexthop init/free AFI_ETHER related NH tables > bgpd: Add a null check in bgp_address_del() function when connected > addresses are removed. > bgpd: add aspath_aggregate_mpath that preserves path length > bgpd: Add [bestpath|multipath] option to 'show ip bgp x.x.x.x' > bgpd: Add clear command to force a bestpath recalculation and > re-advertisement of a prefix > bgpd: Add flag to not change e{u,g}id on startup and run as > unprivileged user > bgpd: Addition of bgp dampening configuration commands under > IPv4/multicast > bgpd: Addition of dampening show commands under v4 unicast/multicast tree > bgpd: Add new configuration cli for graceful restart. > bgpd: Alow gracefull shutdown of peers > bgpd: atomic-aggregate is lost when we aggregate another aggregate that > has atomic-aggregate > bgpd: bgp_nexthop_cache not deleted with peers > bgpd: Correct a few fuzz failures in BGP > bgpd: crash if attributes alone consume > 4096 bytes > bgpd: Display BGP paths with unreachable nexthops as invalid > bgpd: ditch unused bgp_node_*() functions > bgpd: divorce router-id logic from CLI & zebra > bgpd, doc: Allow route-map policy modifications to also affect route > reflectors. > bgpd, doc, lib, zebra: nexthop-tracking in zebra > bgpd: don't count a route with an unreachable nexthop in PfxRcd > bgpd: Enable support for BGP IPV6 multipath. > bgpd: Fix aggregation issues found via ANVL > bgpd: Fix BGP_INFO_ATTR_CHANGED being cleared incorrectly > bgpd: fix BGP IPv6 route map error > bgpd: Fix buffer overflow error in bgp_dump_routes_func > bgpd: Fix code path that leads to uninitialized variables > bgpd: Fix crash in 'show ip bgp nexthop' > bgpd: fix MRT table dumps for locally-originated routes > bgpd: Fix NHT race with Connect leading to test tool issues > bgpd: fix off-by-one in attribute flags handling > bgpd: Fix 'show bgp ipv4 vpnv4 statistics' cli. > bgpd: Fix use after free in aspath_prepend with confeds > bgpd: fix wrong help strings of "match peer" and "no match peer" commands > bgpd: flag paths for multipath if we RX link-local and global nexthops > bgpd: iBGP multipath is broken if 'bgp deterministic-med' is enabled. > bgpd: implement admin distance > bgpd: Improve peer scaling > bgpd: IPv6 Multipath broken with nexthop global address > bgpd, lib: Clarify the different permutations of soft clearing a peer > bgpd, lib, ospfd, zebra: Add ability to read/write tag value > bgpd: Make aspath_highest 4-byte compatible for private AS > bgpd: Make "no redistribute" always remove the redistribute statement > bgpd: Make source interface selection in BGP for nexthop determination > more robust > bgpd: Make the private AS number check 4 byte compatible. > bgpd: minor header/API clean up > bgpd, ospfd, ospf6d, ripd, ripngd, zebra: 'set metric -12' is broken in > the parser > bgpd: Removed unused variable from 'struct attr_extra' > bgpd: Remove unused and leaking code > bgpd, ripngd, zebra: Remove duplicate PSIZE define > bgpd: Show more meaningful outq value in 'show ip bgp summary' output. > bgpd: Squash spurious "unknown afi" log messages > bgpd: Support matching on local preference in route-map > bgpd, vtysh: Add support for route tags > bgpd, zebra: Use next hop tracking for connected routes too > bgp: fix warning in bgp_nht.c > bgp: ignore NHT when bgpd has never connected zebra > bgp: restore missing check from original ignore NHT change > build: detect and create AM_SILENT_RULES macro if needed > build/pimd: remove special shell scripts > build: remove werror from buildtest.sh for now > build: support for "development build" > build: Test for features.h and move include to lib/zebra.h > build: turn off automake portability warnings > config: Remove unused library check > doc: add blurb on use of protobuf with FPM > doc: Add isisd.texi to quagga_TEXINFOS so it gets in dist tarball > doc: Update documentation for Traffic Engineering > fpm: Add protobuf support for FPM. > isisd: Add support of Traffic Engineering > isisd: API: area (L1), domain (L2) passwords > isisd: API: basic area config > isisd: API: basic circuit config > isisd: API: circuit password > isisd: API: LSP-MTU & area level > isisd: API: timers (IIH, CSNP, PSNP) > isisd: API: timers (LSP, SPF) > isisd: cleanup isis_circuit_lookup, errant from TE changes? > isisd: drop unused per-type metric values > isisd: fix a crash due to an lsp-mtu issue > isisd: fix an error that was probably a result of copypasting > isisd: fix assert warning > isisd: fix isis_circuit_af_set() on fresh circuit > isisd: fix isis_circuit_create() > isisd: fix is_type_set > isisd: fix network-type configuration > isisd: Fix size of malloc > isisd: fold up isis_circuit_is_type_set() > isisd: ignore unknown interfaces when adjusting IS-IS mtu > isisd: make sure that all interface addresses are advertised > isisd: Segmentation fault on isis daemon fixes > isisd: warn if there is an MTU issue on circuits > isisd: work around route table asserts for deleting node with info > *: Consolidate all double VIEW_NODE and ENABLE_NODE's > *: Fix duplicate commands from view/enable node consolidation > *: get rid of "MTYPE 0" > *: Remove C99ism, CI system runs builds with CFLAGS and compilers that > barf > *: Remove some for statement declarations > *: Widen width of Zserv routing tag field. > lib: Abstract fd set operations > lib: add AF_ETHERNET/AFI_ETHER > lib: add api method to read current config path > lib: add event_counter > lib: Add new if_link_params structure > lib: Allow zclient do-over of connect on initial attempt > lib: change command logging to be off by default > lib: consolidate ntohf/htonf from ospfd/isisd TE to lib/network > lib: Consolidate VIEW_NODE to be ENABLE_NODE as well > lib: dump memory stats on core > lib: fix memory leak in zprivs_caps_init > lib: fix MIN/MAX macros to not double-eval > lib: Fix priviledge modification for vty group specified > lib: Fix regression in thread.c > lib: IEC559 tests are fragile, reduce to warning rather than error. > lib: keep hash of node's commands to detect duplicate installs > lib: Make distribute.c accepts both v4 and v6. > lib: Refactor read/write functionality > lib: Refactor thread_process_fd > lib: Remove unnecessary parentheses > lib: Replace lists with arrays to store read and write threads > lib: simplify distribute.c's code. > lib/stream: add float/double <-> IEEE-754 single/double mux/demuxers > lib: update Solaris multicast API (BZ#725) > lib: use constant to replace magic value for length of quagga_timestamp > opsf6d: Update router-LSA when nbr's interface-ID changes > ospf6d: Adding the initialization check in ospfv3_clean() > ospf6d: Add the missing ospf6 running check in show cmds > ospf6d: don't access nexthops out of bounds > ospf6d: Fix double increment of Sequence Number > ospf6d: Fixing a couple of issues with ospf6_route_remove () routine. > ospf6d: Fix loss of hello's on interface > ospf6d: fix off-by-one on display of spf reasons > ospf6d: implement admin distance > ospf6d: LA (local-address) bit related inter-op fix. > ospf6d, ripd, vtysh: Fix "no set metric" for ospf6 and RIP > ospf6d: Support for 'clear ipv6 ospf6 interface [ifname] > ospf6d: We should accept long form of "no redistribute" > osfd: Make OSPF compliant to the last sentence of this section in RFC 2328 > ospfd: Add support Router Capabilities support to OSPF > ospfd: Don't wait for state change to Exchange to start LSReq > ospfd: fix - correct neighbor index on changing/p2p/virtual links > ospfd: Fix no ip ospf hello-interval > ospfd: fix non-zero metric when describing loopback addresses > ospfd: Make destination of p2p to multicast for LS-ACKS > ospfd: "show ip ospf neighbor" header spacing > ospfd: Support for 'clear ip ospf interface [IFNAME]' > ospfd: Update Traffic Engineering support > ospfd, vtysh: Add support for Route tags > pimd: Fix hang when doing nexthop lookup from zebra > pimd: send pim prune via correct interface when rpf upstream interface > for a source changes > qpb: Add support for protobuf. > release: Bump version to 1.1.0 > ripd: code simplification for redistribution. > ripd: Fix Null pointer dereference > ripd, ripngd: add support for route tags > ripd: split-horizon default differed between rip_interface_new and _reset > ripngd: code simplification for redistribution. > tests: Fix testbgpmpattr and make check, broken by BGP NHT. > Use only the ISC license for .proto files. > vtysh: Add ability to only look at one processes work queue > vtysh: Add support of new link-params CLI > vtysh: auto-generated vtysh_cmd.c file should depend on its creator > vtysh: Make ipv6 unicast/multicast address-family work > vtysh: service integrated-vtysh-config not being written to file > zebra: *addattr should use size_t len args, quell signed v unsigned > warning > zebra: Add check to notice when an interface is unnumbered > zebra: Add command to configure default for link-state, and make it sticky > zebra: add developer test functions for FPM code > zebra: Add internal support for route tags > zebra: Add link parameters support to Zebra > zebra: add missing vty commands > zebra: add support for route tags > zebra: count iface up/down events and keep last time of their occurrence > zebra: Dissallow outside programs to delete Quagga routes > zebra: Enable fpm module to connect to remote fpm server > zebra: fix crash caused by using route-map with "set src" > zebra: Fix incorrectly flagged nexthop > zebra: Fix some warnings found during compile. > zebra: Gather and display detailed info about clients of Zebra > zebra: handle multihop nexthop changes properly > zebra: ipv6 multipath support > zebra: optionally use protobuf with FPM > zebra: Set link-detect on by default > zebra: stack overrun in IPv6 RA receive code (CVE-2016-1245) > zebra: Suppress displaying default config as part of running config > zebra: Update news with the status of link-detect and the global default > zserv: [pimd] fix - avoid dereferencing a NULL pointer > > regards, > -- Paul Jakma | p...@jakma.org | @pjakma | Key ID: 0xD86BF79464A2FF6A > Fortune: > Imbalance of power corrupts and monopoly of power corrupts absolutely. > -- Genji > > > > _______________________________________________ > Quagga-dev mailing list > Quagga-dev@lists.quagga.net > https://lists.quagga.net/mailman/listinfo/quagga-dev > _______________________________________________ Quagga-dev mailing list Quagga-dev@lists.quagga.net https://lists.quagga.net/mailman/listinfo/quagga-dev