[bgp]: Tuning zebra nexthop-group keep parameter (#21135)

Why I did it Setting the nexthop-group keep parameter to 1. This will instruct zebra not to save nexthop group for more than 1 second after removal. Without this zebra will keep nexthop group in the system for 180 seconds. In scaled scenarios when this parameter is not set it resulted in the queue growing so big and crashing zebra due to OOM when there is test on link flapping. How I did it Update the zebra template and initialize nexthop-group keep as 1. How to verify it Running the scale test with link flapping and ensure no memory increase in zebra.
sonic-net · Dec 16, 2024 · 3eb0196 · 3eb0196
1 parent bce1619
commit 3eb0196
Show file tree

Hide file tree

Showing 11 changed files with 11 additions and 0 deletions.
diff --git a/dockers/docker-fpm-frr/frr/zebra/zebra.conf.j2 b/dockers/docker-fpm-frr/frr/zebra/zebra.conf.j2
@@ -18,6 +18,7 @@ no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
 {% endblock fpm %}
+zebra nexthop-group keep 1
 !
 {% include "common/daemons.common.conf.j2" %}
 !

diff --git a/platform/vs/docker-sonic-vs/frr/zebra.conf b/platform/vs/docker-sonic-vs/frr/zebra.conf
@@ -1,4 +1,5 @@
 no fpm use-next-hop-groups
 
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 
diff --git a/src/sonic-bgpcfgd/tests/data/sonic-cfggen/zebra/zebra.conf b/src/sonic-bgpcfgd/tests/data/sonic-cfggen/zebra/zebra.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py2/t2-chassis-fe-vni-zebra.conf b/src/sonic-config-engine/tests/sample_output/py2/t2-chassis-fe-vni-zebra.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py2/t2-chassis-fe-zebra.conf b/src/sonic-config-engine/tests/sample_output/py2/t2-chassis-fe-zebra.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py2/zebra_frr.conf b/src/sonic-config-engine/tests/sample_output/py2/zebra_frr.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py2/zebra_frr_dualtor.conf b/src/sonic-config-engine/tests/sample_output/py2/zebra_frr_dualtor.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py3/t2-chassis-fe-vni-zebra.conf b/src/sonic-config-engine/tests/sample_output/py3/t2-chassis-fe-vni-zebra.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py3/t2-chassis-fe-zebra.conf b/src/sonic-config-engine/tests/sample_output/py3/t2-chassis-fe-zebra.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py3/zebra_frr.conf b/src/sonic-config-engine/tests/sample_output/py3/zebra_frr.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !

diff --git a/src/sonic-config-engine/tests/sample_output/py3/zebra_frr_dualtor.conf b/src/sonic-config-engine/tests/sample_output/py3/zebra_frr_dualtor.conf
@@ -8,6 +8,7 @@
 no fpm use-next-hop-groups
 !
 fpm address 127.0.0.1
+zebra nexthop-group keep 1
 !
 ! template: common/daemons.common.conf.j2
 !