Commit 442259d6 authored by David Johnson's avatar David Johnson

Ensure database.max_overflow == DEFAULT.api_workers (num cpus).

This causes bugs in Neutron if not set like this, under heavy parallel
load, but the root cause is unknown on the bug tracker.
(api_workers defaults to the number of CPUs if unset.)
parent f1580b8b
Pipeline #1912 passed with stage
in 2 seconds
......@@ -93,8 +93,13 @@ mysql neutron < $OURDIR/capnet/networking-capnet/networking_capnet/db/create.sql
## Finally, modify the neutron config a bit to help it scale to big
## instance workloads better.
##
crudini --set /etc/neutron/neutron.conf database max_overflow 32
crudini --set /etc/neutron/neutron.conf database max_pool_size 32
## Whoops, these are supposed to be not just a large amount; they should
## be set to the same number as api_workers, and that defaults to the
## number of CPUs in the system.
##
ncpus=`cat /proc/cpuinfo | grep -i 'processor.*:' | wc -l`
crudini --set /etc/neutron/neutron.conf database max_overflow $ncpus
crudini --set /etc/neutron/neutron.conf database max_pool_size $ncpus
crudini --set /etc/neutron/neutron.conf database pool_timeout 30
##
......@@ -106,8 +111,8 @@ service_restart neutron-server
## Finally, modify the nova config a bit to help it scale to big
## instance workloads better.
##
crudini --set /etc/nova/nova.conf api_database max_overflow 32
crudini --set /etc/nova/nova.conf api_database max_pool_size 32
crudini --set /etc/nova/nova.conf api_database max_overflow $ncpus
crudini --set /etc/nova/nova.conf api_database max_pool_size $ncpus
crudini --set /etc/nova/nova.conf api_database pool_timeout 30
service nova-api restart
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment