Adds custom ocf for HA of DataSource Node

ocf file for DseNode that datasources run on. When user want HA for
DataSourceDriver, use this ocf for RA on Pacemaker.

Partial-Bug: #1602853
Partially implements blueprint: high-availability-design

Change-Id: I597797b29b40292f5a9542c521aabc577e5e9c2b
This commit is contained in:
Masahito Muroi 2016-07-14 19:50:16 +09:00
parent 1b40706e47
commit cfb7a624ca
1 changed files with 313 additions and 0 deletions

View File

@ -0,0 +1,313 @@
#!/bin/sh
#
#
# OpenStack Congress DataSource Node
#
# Description: Manages an OpenStack Congress DataSource Node as an HA resource
#
# Authors: Masahito Muroi
#
# Support: openstack-dev@lists.openstack.org
# License: Apache Software License (ASL) 2.0
#
#
# See usage() function below for more details ...
#
# OCF instance parameters:
# OCF_RESKEY_binary
# OCF_RESKEY_config
# OCF_RESKEY_pid
#######################################################################
# Initialization:
: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat}
. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs
#######################################################################
# Fill in some defaults if no values are specified
OCF_RESKEY_binary_default="/usr/local/bin/congress-server"
OCF_RESKEY_config_default="/etc/congress/congress.conf"
OCF_RESKEY_node_id_default="datasource-node"
OCF_RESKEY_pid_default="$HA_RSCTMP/$OCF_RESOURCE_INSTANCE.pid"
: ${OCF_RESKEY_binary=${OCF_RESKEY_binary_default}}
: ${OCF_RESKEY_config=${OCF_RESKEY_config_default}}
: ${OCF_RESKEY_node_id=${OCF_RESKEY_node_id_default}}
: ${OCF_RESKEY_pid=${OCF_RESKEY_pid_default}}
#######################################################################
usage() {
cat <<UEND
usage: $0 (start|stop|validate-all|meta-data|status|monitor)
$0 manages an openstack congress datasource node as an ha resource
The 'start' operation starts the datasource node
The 'stop' operation stops the datasource node
The 'validate-all' operation reports whether the parameters are valid
The 'meta-data' operation reports this RA's meta-data information
The 'status' operation reports whether the datasource node is running
The 'monitor' operation reports whether the datasource node seems to be working
UEND
}
meta_data() {
cat <<END
<?xml version="1.0"?>
<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
<resource-agent name="congress-datasource">
<version>1.0</version>
<longdesc lang="en">
Resource agent for the OpenStack Congress DataSource Node
May manage a dsenode that all datasources run on.
</longdesc>
<shortdesc lang="en">Manages the OpenStack DataSource Node</shortdesc>
<parameters>
<parameter name="binary" unique="0" required="0">
<longdesc lang="en">
Location of the OpenStack Congress binary
</longdesc>
<shortdesc lang="en">OpenStack Congress binary</shortdesc>
<content type="string" default="${OCF_RESKEY_binary_default}" />
</parameter>
<parameter name="config" unique="0" required="0">
<longdesc lang="en">
Location of the OpenStack Congress configuration file
</longdesc>
<shortdesc lang="en">OpenStack Congress config file</shortdesc>
<content type="string" default="${OCF_RESKEY_config_default}" />
</parameter>
<parameter name="node_id" unique="0" required="0">
<longdesc lang="en">
ID of the DataSource node
</longdesc>
<shortdesc lang="en">DataSource Node ID</shortdesc>
<content type="string" default="${OCF_RESKEY_node_id_default}" />
</parameter>
<parameter name="pid" unique="0" required="0">
<longdesc lang="en">
The pid file to use for this OpenStack Congress instance
</longdesc>
<shortdesc lang="en">OpenStack Congress pid file</shortdesc>
<content type="string" default="${OCF_RESKEY_pid_default}" />
</parameter>
<parameter name="additional_parameters" unique="0" required="0">
<longdesc lang="en">
Additional parameters to pass on to the OpenStack Congress
</longdesc>
<shortdesc lang="en">Additional parameters for Congress</shortdesc>
<content type="string" />
</parameter>
</parameters>
<actions>
<action name="start" timeout="10" />
<action name="stop" timeout="10" />
<action name="status" timeout="10" />
<action name="monitor" timeout="5" interval="10" />
<action name="validate-all" timeout="5" />
<action name="meta-data" timeout="5" />
</actions>
</resource-agent>
END
}
#######################################################################
# Functions invoked by resource manager actions
congress_validate() {
local rc
check_binary $OCF_RESKEY_binary
if [ ! -f $OCF_RESKEY_config ]; then
if ! ocf_is_probe; then
ocf_log err "Config $OCF_RESKEY_config doesn't exist"
return $OCF_ERR_INSTALLED
fi
ocf_log_warn "Config $OCF_RESKEY_config not available during a probe"
fi
true
}
congress_status() {
local pid
local rc
if [ ! -f $OCF_RESKEY_pid ]; then
ocf_log info "OpenStack Congress is not running"
return $OCF_NOT_RUNNING
else
pid=`cat $OCF_RESKEY_pid`
fi
ocf_run -warn kill -s 0 $pid
rc=$?
if [ $rc -eq 0 ]; then
return $OCF_SUCCESS
else
ocf_log info "Old PID file found, but OpenStack Congress" \
"is not running"
return $OCF_NOT_RUNNING
fi
}
congress_monitor() {
local rc
local pid
congress_status
rc=$?
# If status returned anything but success, return that immediately
if [ $rc -ne $OCF_SUCCESS ]; then
return $rc
fi
ocf_pidfile_status $OCF_RESKEY_pid
rc=$?
if [ $rc -ne 0 ]; then
pid=`cat $OCF_RESKEY_pid`
ocf_log error "This pid: $pid from Congress is not running."
return $OCF_NOT_RUNNING
fi
ocf_log debug "OpenStack Congress monitor succeeded"
return $OCF_SUCCESS
}
congress_start() {
local rc
congress_status
rc=$?
if [ $rc -eq $OCF_SUCCESS ]; then
ocf_log info "OpenStack Congress already running"
return $OCF_SUCCESS
fi
su ${OCF_RESKEY_user} -s /bin/sh -c "${OCF_RESKEY_binary} \
--config-file=$OCF_RESKEY_config --datasources \
--node_id=$OCF_RESKEY_node_id \
$OCF_RESKEY_additional_parameters"' >> /dev/null 2>&1 & echo $!' \
> $OCF_RESKEY_pid
# Spin waiting for the server to come up.
# Let the CRM/LRM time us out if required
while true; do
congress_monitor
rc=$?
[ $rc -eq $OCF_SUCCESS ] && break
if [ $rc -ne $OCF_NOT_RUNNING ]; then
ocf_log err "OpenStack Congress start failed"
exit $OCF_ERR_GENERIC
fi
sleep 1
done
ocf_log info "OpenStack Congress started"
return $OCF_SUCCESS
}
congress_stop() {
local rc
local pid
congress_status
rc=$?
if [ $rc -eq $OCF_NOT_RUNNING ]; then
ocf_log info "OpenStack Congress already stopped"
return $OCF_SUCCESS
fi
# Try SIGTERM
pid=`cat $OCF_RESKEY_pid`
ocf_run kill -s TERM $pid
rc=$?
if [ $rc -ne 0 ]; then
ocf_log err "OpenStack Congress couldn't be stopped"
exit $OCF_ERR_GENERIC
fi
# stop waiting
shutdown_timeout=15
if [ -n "$OCF_RESKEY_CRM_meta_timeout" ]; then
shutdown_timeout=$((($OCF_RESKEY_CRM_meta_timeout/1000)-5))
fi
count=0
while [ $count -lt $shutdown_timeout ]; do
congress_status
rc=$?
if [ $rc -eq $OCF_NOT_RUNNING ]; then
break
fi
count=`expr $count + 1`
sleep 1
ocf_log debug "OpenStack Congress still hasn't stopped yet. Waiting ..."
done
congress_status
rc=$?
if [ $rc -ne $OCF_NOT_RUNNING ]; then
# SIGTERM didn't help either, try SIGKILL
ocf_log info "OpenStack Congress failed to stop" \
"after ${shutdown_timeout}s using SIGTERM. Trying SIGKILL ..."
ocf_run kill -s KILL $pid
fi
ocf_log info "OpenStack Congress stopped"
rm -f $OCF_RESKEY_pid
return $OCF_SUCCESS
}
#######################################################################
case "$1" in
meta-data)
meta_data
exit $OCF_SUCCESS
;;
usage|help)
usage
exit $OCF_SUCCESS
;;
esac
# Anything except meta-data and help must pass validation
congress_validate || exit $?
# What kind of method was invoked?
case "$1" in
start)
congress_start
;;
stop)
congress_stop
;;
status)
congress_status
;;
monitor)
congress_monitor
;;
validate-all)
;;
*)
usage
exit $OCF_ERR_UNIMPLEMENTED
;;
esac