Implement best practices for monitor function

- remove the variable which contain the binary
- check the netstat binary in the validation function
- check if the ports set by the user are numeric and valid with a
length of 4
This commit is contained in:
Sébastien Han 2012-09-06 00:05:59 +02:00
parent 75ea145a60
commit 9d24bbe6f4

View File

@ -17,11 +17,9 @@
# #
# OCF instance parameters: # OCF instance parameters:
# OCF_RESKEY_binary # OCF_RESKEY_binary
# OCF_RESKEY_dhcp_binary
# OCF_RESKEY_config # OCF_RESKEY_config
# OCF_RESKEY_user # OCF_RESKEY_user
# OCF_RESKEY_pid # OCF_RESKEY_pid
# OCF_RESKEY_monitor_binary
# OCF_RESKEY_database_server_port # OCF_RESKEY_database_server_port
# OCF_RESKEY_amqp_server_port # OCF_RESKEY_amqp_server_port
# OCF_RESKEY_zeromq # OCF_RESKEY_zeromq
@ -37,21 +35,17 @@
# Fill in some defaults if no values are specified # Fill in some defaults if no values are specified
OCF_RESKEY_binary_default="nova-network" OCF_RESKEY_binary_default="nova-network"
OCF_RESKEY_dhcp_binary_default="dnsmasq"
OCF_RESKEY_config_default="/etc/nova/nova.conf" OCF_RESKEY_config_default="/etc/nova/nova.conf"
OCF_RESKEY_user_default="nova" OCF_RESKEY_user_default="nova"
OCF_RESKEY_pid_default="$HA_RSCTMP/$OCF_RESOURCE_INSTANCE.pid" OCF_RESKEY_pid_default="$HA_RSCTMP/$OCF_RESOURCE_INSTANCE.pid"
OCF_RESKEY_monitor_binary_default="netstat"
OCF_RESKEY_database_server_port_default="3306" OCF_RESKEY_database_server_port_default="3306"
OCF_RESKEY_amqp_server_port_default="5672" OCF_RESKEY_amqp_server_port_default="5672"
OCF_RESKEY_zeromq_default="false" OCF_RESKEY_zeromq_default="false"
: ${OCF_RESKEY_binary=${OCF_RESKEY_binary_default}} : ${OCF_RESKEY_binary=${OCF_RESKEY_binary_default}}
: ${OCF_RESKEY_dhcp_binary=${OCF_RESKEY_dhcp_binary_default}}
: ${OCF_RESKEY_config=${OCF_RESKEY_config_default}} : ${OCF_RESKEY_config=${OCF_RESKEY_config_default}}
: ${OCF_RESKEY_user=${OCF_RESKEY_user_default}} : ${OCF_RESKEY_user=${OCF_RESKEY_user_default}}
: ${OCF_RESKEY_pid=${OCF_RESKEY_pid_default}} : ${OCF_RESKEY_pid=${OCF_RESKEY_pid_default}}
: ${OCF_RESKEY_monitor_binary=${OCF_RESKEY_monitor_binary_default}}
: ${OCF_RESKEY_database_server_port=${OCF_RESKEY_database_server_port_default}} : ${OCF_RESKEY_database_server_port=${OCF_RESKEY_database_server_port_default}}
: ${OCF_RESKEY_amqp_server_port=${OCF_RESKEY_amqp_server_port_default}} : ${OCF_RESKEY_amqp_server_port=${OCF_RESKEY_amqp_server_port_default}}
: ${OCF_RESKEY_zeromq=${OCF_RESKEY_zeromq_default}} : ${OCF_RESKEY_zeromq=${OCF_RESKEY_zeromq_default}}
@ -174,6 +168,8 @@ nova_network_validate() {
local rc local rc
check_binary $OCF_RESKEY_binary check_binary $OCF_RESKEY_binary
check_binary netstat
check_binary dsmask
# A config file on shared storage that is not available # A config file on shared storage that is not available
# during probes is OK. # during probes is OK.
@ -218,9 +214,12 @@ nova_network_status() {
nova_network_monitor() { nova_network_monitor() {
local rc local rc
local token local pid
local rc_database local rc_db
local rc_amqp local rc_amqp
local network_db_check
local network_amqp_check
nova_network_status nova_network_status
rc=$? rc=$?
@ -230,32 +229,28 @@ nova_network_monitor() {
return $rc return $rc
fi fi
# Check whether we are supposed to monitor by logging into nova-network # Check the connections according to the PID.
# and do it if that's the case. # We are sure to hit the scheduler process and not other nova process with the same connection behavior (for example nova-cert)
if ! check_binary $OCF_RESKEY_monitor_binary; then
ocf_log warn "$OCF_RESKEY_monitor_binary missing, can not monitor!"
else
if ocf_is_true "$OCF_RESKEY_zeromq"; then if ocf_is_true "$OCF_RESKEY_zeromq"; then
PID=`cat $OCF_RESKEY_pid` pid=`cat $OCF_RESKEY_pid`
SCHEDULER_DATABASE_CO_CHECK=`"$OCF_RESKEY_monitor_binary" -punt | grep -s "$OCF_RESKEY_database_server_port" | grep -s "$PID" | grep -qs "ESTABLISHED"` network_db_check=`netstat -punt | grep -s "$OCF_RESKEY_database_server_port" | grep -s "$pid" | grep -qs "ESTABLISHED"`
rc_database=$? rc_db=$?
if [ $rc_database -ne 0 ]; then if [ $rc_db -ne 0 ]; then
ocf_log err "Nova Network is not connected to the database server: $rc_database" ocf_log err "Nova Network is not connected to the database server: $rc_db"
return $OCF_NOT_RUNNING return $OCF_NOT_RUNNING
fi fi
else else
PID=`cat $OCF_RESKEY_pid` pid=`cat $OCF_RESKEY_pid`
# check the connections according to the PID # check the connections according to the PID
SCHEDULER_DATABASE_CO_CHECK=`"$OCF_RESKEY_monitor_binary" -punt | grep -s "$OCF_RESKEY_database_server_port" | grep -s "$PID" | grep -qs "ESTABLISHED"` network_db_check=`netstat -punt | grep -s "$OCF_RESKEY_database_server_port" | grep -s "$pid" | grep -qs "ESTABLISHED"`
rc_database=$? rc_db=$?
SCHEDULER_AMQP_CO_CHECK=`"$OCF_RESKEY_monitor_binary" -punt | grep -s "$OCF_RESKEY_amqp_server_port" | grep -s "$PID" | grep -qs "ESTABLISHED"` network_amqp_check=`netstat -punt | grep -s "$OCF_RESKEY_amqp_server_port" | grep -s "$pid" | grep -qs "ESTABLISHED"`
rc_amqp=$? rc_amqp=$?
if [ $rc_amqp -ne 0 ] || [ $rc_database -ne 0 ]; then if [ $rc_amqp -ne 0 ] || [ $rc_db -ne 0 ]; then
ocf_log err "Nova Network is not connected to the AMQP server and/or the database server: AMQP connection test returned $rc_amqp and database connection test returned $rc_database" ocf_log err "Nova Network is not connected to the AMQP server and/or the database server: AMQP connection test returned $rc_amqp and database connection test returned $rc_db"
return $OCF_NOT_RUNNING return $OCF_NOT_RUNNING
fi fi
fi fi
fi
ocf_log debug "OpenStack Nova Network (nova-network) monitor succeeded" ocf_log debug "OpenStack Nova Network (nova-network) monitor succeeded"
return $OCF_SUCCESS return $OCF_SUCCESS
@ -278,7 +273,6 @@ nova_network_start() {
# Spin waiting for the server to come up. # Spin waiting for the server to come up.
# Let the CRM/LRM time us out if required # Let the CRM/LRM time us out if required
sleep 1
while true; do while true; do
nova_network_monitor nova_network_monitor
rc=$? rc=$?
@ -308,7 +302,7 @@ nova_network_stop() {
# Try SIGTERM # Try SIGTERM
pid=`cat $OCF_RESKEY_pid` pid=`cat $OCF_RESKEY_pid`
ocf_run kill -s TERM $pid ocf_run kill -s TERM $pid
pkill $OCF_RESKEY_dhcp_binary pkill dnsmask
rc=$? rc=$?
if [ $rc -ne 0 ]; then if [ $rc -ne 0 ]; then
ocf_log err "OpenStack Nova Network (nova-network) couldn't be stopped" ocf_log err "OpenStack Nova Network (nova-network) couldn't be stopped"