Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | 0.99.12-5+jenner16vyatta/3.4.0/i386debian/0.99.12-5+jenner16 | Stephen Hemminger | 2009-07-20 | 1 | -0/+7 |
| | |||||
* | Zebra: don't use BGP routes for recursive routes | Stephen Hemminger | 2009-07-20 | 2 | -16/+8 |
| | | | | | | | This reverts change in Jenner development to allow BGP routes to be used as next hop candidates for recursive routes. The problem is that BGP route table can be large, and rib_update requires a full scan for each route added; leading to N^2 behavior. | ||||
* | BGP: update MD5 key on correct listener socket | Stephen Hemminger | 2009-07-20 | 1 | -7/+7 |
| | | | | | Only pass MD5 key to socket with same address family. Bug 4496 | ||||
* | 0.99.12-5+jenner15debian/0.99.12-5+jenner15 | Stephen Hemminger | 2009-07-17 | 1 | -0/+6 |
| | |||||
* | Only do RIB scan if route being updated is INTERNAL | Stephen Hemminger | 2009-07-17 | 1 | -7/+14 |
| | | | | | | | RIB scan can be expensive, so it slows down BGP updates; therefore only rescan RIB if route being updated is marked INTERNAL. Bug 4741 | ||||
* | 0.99.12-5+jenner14debian/0.99.12-5+jenner14 | Stephen Hemminger | 2009-07-16 | 1 | -0/+6 |
| | |||||
* | Force RIB update after routing protocols change routes | Stephen Hemminger | 2009-07-16 | 1 | -0/+6 |
| | | | | | | | | When a routing protocol communicates a route change to Zebra, Zebra has to do a rescan of the RIB to check for any changes in recursive routes. Bug 4228 | ||||
* | 0.99.12-5+jenner13debian/0.99.12-5+jenner13 | Stephen Hemminger | 2009-07-16 | 1 | -0/+10 |
| | |||||
* | Use separate sockets for IPV4/IPV6 | Stephen Hemminger | 2009-07-15 | 4 | -81/+106 |
| | | | | | | | | New code to keep track of two listener threads, with common code for both cases. Revert "Use single socket for both IPV4 and IPV6" This reverts commit 3e36dfb3cf0c4a631f693da58e7b80dff9ccce06. | ||||
* | Revert "Remove code in md5sig that attempts to normalize IPV4 mapped socket" | Stephen Hemminger | 2009-07-15 | 1 | -3/+43 |
| | | | | This reverts commit 2585b03455c2c495ab3a0cfbe13c4e5d3ac70e11. | ||||
* | Revert "Copy full address to md5 key" | Stephen Hemminger | 2009-07-15 | 1 | -1/+1 |
| | | | | This reverts commit 5405c4db88bd0b439f41776c94452b4360de1509. | ||||
* | Copy full address to md5 key | Stephen Hemminger | 2009-07-15 | 1 | -1/+1 |
| | | | | May fix issues with IPV6 and MD5. | ||||
* | 0.99.12-5+jenner12debian/0.99.12-5+jenner12 | Stephen Hemminger | 2009-07-13 | 1 | -0/+11 |
| | |||||
* | Remove code in md5sig that attempts to normalize IPV4 mapped socket | Stephen Hemminger | 2009-07-13 | 1 | -43/+3 |
| | | | | | The socket address is already normalize during the accept handling, so the broken code in md5sig is unnecessary. | ||||
* | BGP: delay listening until first AS defined | Stephen Hemminger | 2009-07-11 | 4 | -7/+27 |
| | | | | | | | This allows BGP daemon to run without leaving listener port open until needed. Bug 3088 | ||||
* | bgp_socket : remove unused arg | Stephen Hemminger | 2009-07-11 | 2 | -14/+8 |
| | | | | | bgp_socket is always called with NULL as first argument. For clarity, just remove it and update the places it is used. | ||||
* | Use single socket for both IPV4 and IPV6 | Stephen Hemminger | 2009-07-11 | 3 | -34/+14 |
| | | | | | The bug that required BGP to listen on two sockets (using IPV6_only) should now be fixed. | ||||
* | Fix bug in IPV6 normalized addresses | Stephen Hemminger | 2009-07-11 | 1 | -0/+1 |
| | | | | | This is probably why using IPV4 in IPV6 doesn't work in BGP. The code to normalize address was not copying port value! | ||||
* | Merge branch 'jenner' of suva.vyatta.com:/git/vyatta-quagga into jenner | Stephen Hemminger | 2009-07-11 | 1 | -0/+6 |
|\ | |||||
| * | 0.99.12-5+jenner11debian/0.99.12-5+jenner11 | Stephen Hemminger | 2009-07-09 | 1 | -0/+6 |
| | | |||||
* | | OSPF6: don't enable TCP vty port | Stephen Hemminger | 2009-07-10 | 1 | -2/+0 |
|/ | | | | | | Bug 4717 Don't open a TCP control port for ospf6, there is always a Unix domain socket opened. This makes ospf6d behave like other quagga daemons. | ||||
* | Start all routing daemons at boot time | Stephen Hemminger | 2009-07-09 | 1 | -8/+15 |
| | | | | Go back to starting all routing daemons at boot time. | ||||
* | 0.99.12-5+jenner10debian/0.99.12-5+jenner10 | Stephen Hemminger | 2009-07-08 | 1 | -0/+7 |
| | |||||
* | Allow RIB to store multiple kernel routes | Stephen Hemminger | 2009-07-08 | 1 | -0/+9 |
| | | | | | It is possible when using default routes to have two entries in FIB with different protocols/metrics so don't assume they are the same. | ||||
* | Allow iBGP to be used for recusive routes | Stephen Hemminger | 2009-07-08 | 1 | -4/+8 |
| | | | | | Bug 4228 This inclues iBGP routes in the active route calculation. | ||||
* | 0.99.12-5+jenner9debian/0.99.12-5+jenner9 | Stig Thormodsrud | 2009-06-24 | 1 | -0/+6 |
| | |||||
* | Revert "ospf: fix performance of ospf_lsa_maxage" | Stig Thormodsrud | 2009-06-24 | 2 | -10/+23 |
| | | | | | | This patch seems to cause bug 4627. This reverts commit ca5f52dcb72b6bc19037400ec27053d8fe2b5e6e. | ||||
* | 0.99.12-5+jenner8debian/0.99.12-5+jenner8 | Stephen Hemminger | 2009-06-19 | 1 | -0/+11 |
| | |||||
* | ospf: fix performance of ospf_lsa_maxage | Stephen Hemminger | 2009-06-19 | 2 | -23/+10 |
| | | | | | | | | When flushing lots of entries the performance of ospf_lsa_maxage is terrible because it looks up entry before deleting (N^2). Use a flag to mark whether entry is already in maxage list. Bug 4421 | ||||
* | Merge branch 'jenner' of suva.vyatta.com:/git/vyatta-quagga into jenner | Stephen Hemminger | 2009-06-18 | 1 | -0/+12 |
|\ | |||||
| * | 0.99.12-5+jenner7debian/0.99.12-5+jenner7 | Denis Ovsienko | 2009-06-18 | 1 | -0/+12 |
| | | |||||
| * | [bgpd] Fixed as-path prepend/exclude ASN handling | Denis Ovsienko | 2009-06-18 | 2 | -5/+5 |
| | | | | | | | | | | | | | | | | | | - aspath_gettoken(): use as_t to fix 4-byte ASN scanning (bug #484) - set_aspath_prepend_cmd(): use CMD_AS_RANGE in message to match actual range - no_set_aspath_prepend_val_cmd(): idem - set_aspath_exclude_cmd(): idem - no_set_aspath_exclude_val_cmd(): idem | ||||
* | | Merge branch 'jenner' of 192.168.100.1:git/vyatta-quagga into jenner | Stephen Hemminger | 2009-06-18 | 2 | -55/+8 |
|\ \ | |||||
| * | | ospfd: make some functions static | Stephen Hemminger | 2009-06-18 | 2 | -55/+8 |
| |/ | | | | | | | | | | | | | | | ospf_network_match_iface - make static so it can be inlined and remove old #ifdef'd out code that just confuses Other functions were defined static in prototype, but not declared static. | ||||
* | | Merge branch 'jenner' of 192.168.100.1:git/vyatta-quagga into jenner | Stephen Hemminger | 2009-06-11 | 2 | -5/+5 |
|\ \ | |||||
| * \ | Merge branch 'upstream' into jenner | Stephen Hemminger | 2009-06-11 | 2 | -5/+5 |
| |\ \ | | |/ | |/| | |||||
| | * | [bgpd] Fixed as-path prepend/exclude ASN handling | Denis Ovsienko | 2009-06-09 | 2 | -5/+5 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | - aspath_gettoken(): use as_t to fix 4-byte ASN scanning (bug #484) - set_aspath_prepend_cmd(): use CMD_AS_RANGE in message to match actual range - no_set_aspath_prepend_val_cmd(): idem - set_aspath_exclude_cmd(): idem - no_set_aspath_exclude_val_cmd(): idem | ||||
| * | | Merge branch 'jenner' of suva.vyatta.com:/git/vyatta-quagga into jenner | Stephen Hemminger | 2009-06-11 | 1 | -0/+8 |
| |\ \ | |||||
* | | | | Merge branch 'jenner' of 192.168.100.1:git/vyatta-quagga into jenner | Stephen Hemminger | 2009-06-11 | 1 | -0/+2 |
|\| | | | |/ / |/| | | |||||
| * | | Show no import-check | Stephen Hemminger | 2009-06-05 | 1 | -0/+2 |
| | | | | | | | | | | | | | | | Since BGPD now has flag to default to import-check on, show the current state of import-check in dump output. | ||||
* | | | 0.99.12-5+jenner6vyatta/3.3.99test3/i386vyatta/3.3.99test3/amd64vyatta/3.3.99test2/i386vyatta/3.3.99test1/i386debian/0.99.12-5+jenner6 | Stephen Hemminger | 2009-06-05 | 1 | -0/+8 |
|/ / | |||||
* | | Avoid N^2 behaviour on redistribute withdraw | Stephen Hemminger | 2009-06-05 | 1 | -1/+4 |
| | | | | | | | | | | The old algorithim looked up every node twice as it withdrew the entry. This may explain the slow performance in Bug 4421 | ||||
* | | OSPF vty cleanup | Stephen Hemminger | 2009-06-05 | 1 | -15/+2 |
| | | | | | | | | Make some strings static, and remove strings that are not used. | ||||
* | | Move route_lock_node to fix warnings | Stephen Hemminger | 2009-06-05 | 1 | -18/+18 |
| | | | | | | | | route_node_delete used by inline before defined. | ||||
* | | 0.99.12-5+jenner5debian/0.99.12-5+jenner5 | Stephen Hemminger | 2009-06-04 | 1 | -0/+42 |
| | | |||||
* | | Merge branch 'upstream' into jenner | Stephen Hemminger | 2009-06-04 | 20 | -45/+50 |
|\| | | | | | | | | | | | | | | | | | Conflicts: bgpd/bgp_aspath.c bgpd/bgpd.c lib/smux.c lib/workqueue.c ospfd/ospf_snmp.c | ||||
| * | Merge remote branch 'origin/master' | Paul Jakma | 2009-06-02 | 10 | -12/+12 |
| |\ | |||||
| | * | [trivia] Fixed typos in communities text (#334) | Denis Ovsienko | 2009-06-01 | 2 | -4/+4 |
| | | | |||||
| | * | [doc] Stick with quagga.net for consistency (#436) | Denis Ovsienko | 2009-06-01 | 7 | -7/+7 |
| | | | |||||
| | * | [bgpd] 64-bit bugfix in community_del_val by Jeremy Jackson <jerj@coplanar.net> | Andrew J. Schorr | 2009-05-29 | 1 | -1/+1 |
| | | | | | | | | | | | | | | | * bgpd/bgp_community.c: (community_del_val) Fix bug in memcpy that was using the wrong size on architectures where a pointer is not 32 bits. |