path: root/heartbeat/podman

#!/bin/sh
#
# The podman HA resource agent creates and launches a podman container
# based off a supplied podman image. Containers managed by this agent
# are both created and removed upon the agent's start and stop actions.
#
# Copyright (c) 2014 David Vossel <davidvossel@gmail.com>
#                    Michele Baldessari <michele@acksyn.org>
#                    All Rights Reserved.
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of version 2 of the GNU General Public License as
# published by the Free Software Foundation.
#
# This program is distributed in the hope that it would be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
#
# Further, this software is distributed without any warranty that it is
# free of the rightful claim of any third person regarding infringement
# or the like.  Any license provided herein, whether implied or
# otherwise, applies only to this software file.  Patent licenses, if
# any, provided herein do not apply to combinations of this program with
# other software, or any other product whatsoever.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write the Free Software Foundation,
# Inc., 59 Temple Place - Suite 330, Boston MA 02111-1307, USA.
#

#######################################################################
# Initialization:

: ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat}
. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs

# Parameter defaults

OCF_RESKEY_reuse_default="0"

: ${OCF_RESKEY_reuse=${OCF_RESKEY_reuse_default}}

#######################################################################

meta_data()
{
	cat <<END
<?xml version="1.0"?>
<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
<resource-agent name="podman" version="1.0">
<version>1.0</version>

<longdesc lang="en">
The podman HA resource agent creates and launches a podman container
based off a supplied podman image. Containers managed by this agent
are both created and removed upon the agent's start and stop actions.
</longdesc>
<shortdesc lang="en">Podman container resource agent.</shortdesc>

<parameters>
<parameter name="image" required="1" unique="0">
<longdesc lang="en">
The podman image to base this container off of.
</longdesc>
<shortdesc lang="en">podman image</shortdesc>
<content type="string"/>
</parameter>

<parameter name="name" required="0" unique="0">
<longdesc lang="en">
The name to give the created container. By default this will
be that resource's instance name.
</longdesc>
<shortdesc lang="en">podman container name</shortdesc>
<content type="string"/>
</parameter>

<parameter name="allow_pull" unique="0">
<longdesc lang="en">
Allow the image to be pulled from the configured podman registry when
the image does not exist locally. NOTE, this can drastically increase
the time required to start the container if the image repository is
pulled over the network.
</longdesc>
<shortdesc lang="en">Allow pulling non-local images</shortdesc>
<content type="boolean"/>
</parameter>

<parameter name="run_opts" required="0" unique="0">
<longdesc lang="en">
Add options to be appended to the 'podman run' command which is used
when creating the container during the start action. This option allows
users to do things such as setting a custom entry point and injecting
environment variables into the newly created container. Note the '-d'
option is supplied regardless of this value to force containers to run
in the background.

NOTE: Do not explicitly specify the --name argument in the run_opts. This
agent will set --name using either the resource's instance or the name
provided in the 'name' argument of this agent.

</longdesc>
<shortdesc lang="en">run options</shortdesc>
<content type="string"/>
</parameter>

<parameter name="run_cmd" required="0" unique="0">
<longdesc lang="en">
Specify a command to launch within the container once
it has initialized.
</longdesc>
<shortdesc lang="en">run command</shortdesc>
<content type="string"/>
</parameter>

<parameter name="mount_points" required="0" unique="0">
<longdesc lang="en">
A comma separated list of directories that the container is expecting to use.
The agent will ensure they exist by running 'mkdir -p'
</longdesc>
<shortdesc lang="en">Required mount points</shortdesc>
<content type="string"/>
</parameter>

<parameter name="monitor_cmd" required="0" unique="0">
<longdesc lang="en">
Specify the full path of a command to launch within the container to check
the health of the container. This command must return 0 to indicate that
the container is healthy. A non-zero return code will indicate that the
container has failed and should be recovered.

Note: Using this method for monitoring processes inside a container
is not recommended, as containerd tries to track processes running
inside the container and does not deal well with many short-lived
processes being spawned. Ensure that your container monitors its
own processes and terminates on fatal error rather than invoking
a command from the outside.
</longdesc>
<shortdesc lang="en">monitor command</shortdesc>
<content type="string"/>
</parameter>

<parameter name="force_kill" required="0" unique="0">
<longdesc lang="en">
Kill a container immediately rather than waiting for it to gracefully
shutdown
</longdesc>
<shortdesc lang="en">force kill</shortdesc>
<content type="boolean"/>
</parameter>

<parameter name="reuse" required="0" unique="0">
<longdesc lang="en">
Allow the container to be reused once it is stopped.  By default,
containers get removed once they are stopped.  Enable this option
to have the particular one persist when this happens.
</longdesc>
<shortdesc lang="en">reuse container</shortdesc>
<content type="boolean" default="${OCF_RESKEY_reuse_default}"/>
</parameter>

<parameter name="drop_in_dependency" required="0" unique="0">
<longdesc lang="en">
Use transient drop-in files to add extra dependencies to the systemd
scopes associated to the container. During reboot, this prevents systemd
to stop the container before pacemaker.
</longdesc>
<shortdesc lang="en">drop-in dependency</shortdesc>
<content type="boolean"/>
</parameter>
</parameters>

<actions>
<action name="start"        timeout="90s" />
<action name="stop"         timeout="90s" />
<action name="monitor"      timeout="30s" interval="30s" depth="0" />
<action name="meta-data"    timeout="5s" />
<action name="validate-all"   timeout="30s" />
</actions>
</resource-agent>
END
}

#######################################################################
REQUIRE_IMAGE_PULL=0

podman_usage()
{
	cat <<END
usage: $0 {start|stop|monitor|validate-all|meta-data}

Expects to have a fully populated OCF RA-compliant environment set.
END
}


monitor_cmd_exec()
{
	local rc=$OCF_SUCCESS
	local out

	out=$(podman exec ${CONTAINER} $OCF_RESKEY_monitor_cmd 2>&1)
	rc=$?
	# 125: no container with name or ID ${CONTAINER} found
	# 126: container state improper (not running)
	# 127: any other error
	# 255: podman 2+: container not running
	case "$rc" in
		125|126|255)
			rc=$OCF_NOT_RUNNING
			;;
		0)
			ocf_log debug "monitor cmd passed: exit code = $rc"
			;;
		*)
			ocf_exit_reason "monitor cmd failed (rc=$rc), output: $out"
			rc=$OCF_ERR_GENERIC
			;;
	esac

	return $rc
}

container_exists()
{
	local rc
	local out

	out=$(podman exec ${CONTAINER} $OCF_RESKEY_monitor_cmd 2>&1)
	rc=$?
	# 125: no container with name or ID ${CONTAINER} found
	if [ $rc -ne 125 ]; then
		return 0
	fi
	return 1
}

remove_container()
{
	local rc
	local execids

	if ocf_is_true "$OCF_RESKEY_reuse"; then
		# never remove the container if we have reuse enabled.
		return 0
	fi

	container_exists
	if [ $? -ne 0 ]; then
		# don't attempt to remove a container that doesn't exist
		return 0
	fi
	ocf_log notice "Cleaning up inactive container, ${CONTAINER}."
	ocf_run podman rm -v $CONTAINER
	rc=$?
	if [ $rc -ne 0 ]; then
		# due to a podman bug (rhbz#1841485), sometimes a stopped
		# container can still be associated with Exec sessions, in
		# which case the "podman rm" has to be forced
		execids=$(podman inspect $CONTAINER --format '{{len .ExecIDs}}')
		if [ "$execids" -ne "0" ]; then
			ocf_log warn "Inactive container ${CONTAINER} has lingering exec sessions. Force-remove it."
			ocf_run podman rm -f $CONTAINER
			rc=$?
		fi
	fi
	return $rc
}

podman_simple_status()
{
	local rc

	# simple status is implemented via podman exec
	# everything besides success is considered "not running"
	monitor_cmd_exec
	rc=$?
	if [ $rc -ne $OCF_SUCCESS ]; then
		rc=$OCF_NOT_RUNNING;
	fi
	return $rc
}

podman_monitor()
{
	# We rely on running podman exec to monitor the container
	# state because that command seems to be less prone to
	# performance issue under IO load.
	#
	# For probes to work, we expect cmd_exec to be able to report
	# when a container is not running. Here, we're not interested
	# in distinguishing whether it's stopped or non existing
	# (there's function container_exists for that)
	monitor_cmd_exec
	return $?
}

podman_create_mounts() {
	oldIFS="$IFS"
	IFS=","
	for directory in $OCF_RESKEY_mount_points; do
		mkdir -p "$directory"
	done
	IFS="$oldIFS"
}

podman_container_id()
{
	# Retrieve the container ID by doing a "podman ps" rather than
	# a "podman inspect", because the latter has performance issues
	# under IO load.
	# We could have run "podman start $CONTAINER" to get the ID back
	# but if the container is stopped, the command will return a
	# name instead of a container ID. This would break us.
	podman ps --no-trunc --format '{{.ID}} {{.Names}}' | grep -F -w -m1 "$CONTAINER" | cut -d' ' -f1
}


create_transient_drop_in_dependency()
{
	local cid=$1
	local rc=$OCF_SUCCESS

	if [ -z "$cid" ]; then
		ocf_exit_reason "Container ID not found for \"$CONTAINER\". Not creating drop-in dependency"
		return $OCF_ERR_GENERIC
	fi

	ocf_log info "Creating drop-in dependency for \"$CONTAINER\" ($cid)"
	for scope in "libpod-$cid.scope.d" "libpod-conmon-$cid.scope.d"; do
		if [ $rc -eq $OCF_SUCCESS ] && [ ! -d /run/systemd/transient/"$scope" ]; then
			mkdir -p /run/systemd/transient/"$scope" && \
			printf "[Unit]\nBefore=pacemaker.service" > /run/systemd/transient/"$scope"/dep.conf && \
			chmod ago+r /run/systemd/transient/"$scope" /run/systemd/transient/"$scope"/dep.conf
			rc=$?
		fi
	done

	if [ $rc -ne $OCF_SUCCESS ]; then
		ocf_log error "Could not create drop-in dependency for \"$CONTAINER\" ($cid)"
	else
		systemctl daemon-reload
		rc=$?
		if [ $rc -ne $OCF_SUCCESS ]; then
			ocf_log error "Could not refresh service definition after creating drop-in for \"$CONTAINER\""
		fi
	fi

	return $rc
}


run_new_container()
{
	local opts=$1
	local image=$2
	local cmd=$3
	local rc
	
	ocf_log info "running container $CONTAINER for the first time"
	out=$(podman run $opts $image $cmd 2>&1)
	rc=$?

	if [ -n "$out" ]; then
		out="$(echo "$out" | tr -s ' \t\r\n' ' ')"
		if [ $rc -eq 0 ]; then
			ocf_log info "$out"
		else
			ocf_log err "$out"
		fi
	fi

	if [ $rc -eq 125 ]; then
		# If an internal podman error occurred, it might be because
		# the internal storage layer still references an old container
		# with the same name, even though podman itself thinks there
		# is no such container. If so, purge the storage layer to try
		# to clean the corruption and try again.
		if echo "$out" | grep -q "unknown.*flag"; then
			ocf_exit_reason "$out"
			return $rc
		fi

		ocf_log warn "Internal podman error while creating new container $CONTAINER. Retrying."
		ocf_run podman rm --storage $CONTAINER
		ocf_run podman run $opts $image $cmd
		rc=$?
	elif [ $rc -eq 127 ]; then
		# rhbz#1972209: podman 3.0.x seems to be hit by a race
		# where the cgroup is not yet set up properly when the OCI
		# runtime configures the container. If that happens, recreate
		# the container as long as we get the same error code or
		# until start timeout preempts us.
		while [ $rc -eq 127 ] && (echo "$out" | grep -q "cgroup.*scope not found") ; do
			ocf_log warn "Internal podman error while assigning cgroup. Retrying."
			# Arbitrary sleep to prevent consuming all CPU while looping
			sleep 1
			podman rm -f "$CONTAINER"
			out=$(podman run $opts $image $cmd 2>&1)
			rc=$?
		done
		# Log the created container ID if it succeeded
		if  [ $rc -eq 0 ]; then
			ocf_log info "$out"
		fi
	fi
	
	return $rc
}


podman_start()
{
	local cid
	local rc

	podman_create_mounts
	local run_opts="-d --name=${CONTAINER}"
	# check to see if the container has already started
	podman_simple_status
	if [ $? -eq $OCF_SUCCESS ]; then
		return $OCF_SUCCESS
	fi

	if [ -n "$OCF_RESKEY_run_opts" ]; then
		run_opts="$run_opts $OCF_RESKEY_run_opts"
	fi

	if [ $REQUIRE_IMAGE_PULL -eq 1 ]; then
		ocf_log notice "Beginning pull of image, ${OCF_RESKEY_image}"
		podman pull "${OCF_RESKEY_image}"
		if [ $? -ne 0 ]; then
			ocf_exit_reason "failed to pull image ${OCF_RESKEY_image}"
			return $OCF_ERR_GENERIC
		fi
	fi

	if ocf_is_true "$OCF_RESKEY_reuse" && container_exists; then
		ocf_log info "starting existing container $CONTAINER."
		ocf_run podman start $CONTAINER
	else
		# make sure any previous container matching our container name is cleaned up first.
		# we already know at this point it wouldn't be running
		remove_container
		run_new_container "$run_opts" $OCF_RESKEY_image "$OCF_RESKEY_run_cmd"
		if [ $? -eq 125 ]; then
			return $OCF_ERR_GENERIC
		fi

	fi
	rc=$?

	# if the container was stopped or didn't exist before, systemd
	# removed the libpod* scopes. So always try to recreate the drop-ins
	if [ $rc -eq 0 ] && ocf_is_true "$OCF_RESKEY_drop_in_dependency"; then
		cid=$(podman_container_id)
		create_transient_drop_in_dependency "$cid"
		rc=$?
	fi

	if [ $rc -ne 0 ]; then
		ocf_exit_reason "podman failed to launch container (rc: $rc)"
		return $OCF_ERR_GENERIC
	fi


	# wait for monitor to pass before declaring that the container is started
	while true; do
		podman_simple_status
		if [ $? -ne $OCF_SUCCESS ]; then
			ocf_exit_reason "Newly created podman container exited after start"
			return $OCF_ERR_GENERIC
		fi

		monitor_cmd_exec
		if [ $? -eq $OCF_SUCCESS ]; then
			ocf_log notice "Container $CONTAINER  started successfully"
			return $OCF_SUCCESS
		fi

		ocf_exit_reason "waiting on monitor_cmd to pass after start"
		sleep 1
	done
}

podman_stop()
{
	local timeout=60
	local rc
	podman_simple_status
	if [ $? -eq  $OCF_NOT_RUNNING ]; then
		remove_container
		return $OCF_SUCCESS
	fi

	if [ -n "$OCF_RESKEY_CRM_meta_timeout" ]; then
		timeout=$((($OCF_RESKEY_CRM_meta_timeout/1000) -10 ))
		if [ $timeout -lt 10 ]; then
			timeout=10
		fi
	fi

	if ocf_is_true "$OCF_RESKEY_force_kill"; then
		ocf_run podman kill $CONTAINER
		rc=$?
	else
		ocf_log debug "waiting $timeout second[s] before killing container"
		ocf_run podman stop -t=$timeout $CONTAINER
		rc=$?
		# on stop, systemd will automatically delete any transient
		# drop-in conf that has been created earlier
	fi

	if [ $rc -ne 0 ]; then
		# If the stop failed, it could be because the controlling conmon
		# process died unexpectedly. If so, a generic error code is returned
		# but the associated container exit code is -1. If that's the case,
		# assume there's no failure and continue with the rm as usual.
		if [ $rc -eq 125 ] && \
		   podman inspect --format '{{.State.Status}}:{{.State.ExitCode}}' $CONTAINER | grep -wq "stopped:-1"; then
			ocf_log warn "Container ${CONTAINER} had an unexpected stop outcome. Trying to remove it anyway."
		else
			ocf_exit_reason "Failed to stop container, ${CONTAINER}, based on image, ${OCF_RESKEY_image}."
			return $OCF_ERR_GENERIC
		fi
	fi

	remove_container
	if [ $? -ne 0 ]; then
		ocf_exit_reason "Failed to remove stopped container, ${CONTAINER}, based on image, ${OCF_RESKEY_image}."
		return $OCF_ERR_GENERIC
	fi

	return $OCF_SUCCESS
}

image_exists()
{
	podman image exists "${OCF_RESKEY_image}"
	if [ $? -eq 0 ]; then
		# image found
		return 0
	fi

	if ocf_is_true "$OCF_RESKEY_allow_pull"; then
		REQUIRE_IMAGE_PULL=1
		ocf_log notice "Image (${OCF_RESKEY_image}) does not exist locally but will be pulled during start"
		return 0
	fi
	# image not found.
	return 1
}

podman_validate()
{
	check_binary podman
	if [ -z "$OCF_RESKEY_image" ]; then
		ocf_exit_reason "'image' option is required"
		exit $OCF_ERR_CONFIGURED
	fi

	image_exists
	if [ $? -ne 0 ]; then
		ocf_exit_reason "base image, ${OCF_RESKEY_image}, could not be found."
		exit $OCF_ERR_CONFIGURED
	fi

	return $OCF_SUCCESS
}

# TODO :
# When a user starts plural clones in a node in globally-unique, a user cannot appoint plural name parameters.
# When a user appoints reuse, the resource agent cannot connect plural clones with a container.

if ocf_is_true "$OCF_RESKEY_CRM_meta_globally_unique"; then
	if [ -n "$OCF_RESKEY_name" ]; then
		if [ -n "$OCF_RESKEY_CRM_meta_clone_node_max" ] && [ "$OCF_RESKEY_CRM_meta_clone_node_max" -ne 1 ]
		then
			ocf_exit_reason "Cannot make plural clones from the same name parameter."
			exit $OCF_ERR_CONFIGURED
		fi
		if [ -n "$OCF_RESKEY_CRM_meta_master_node_max" ] && [ "$OCF_RESKEY_CRM_meta_master_node_max" -ne 1 ]
		then
			ocf_exit_reason "Cannot make plural master from the same name parameter."
			exit $OCF_ERR_CONFIGURED
		fi
	fi
	: ${OCF_RESKEY_name=`echo ${OCF_RESOURCE_INSTANCE} | tr ':' '-'`}
else
	: ${OCF_RESKEY_name=${OCF_RESOURCE_INSTANCE}}
fi

CONTAINER=$OCF_RESKEY_name

# Note: we currently monitor podman containers by with the "podman exec"
# command, so make sure that invocation is always valid by enforcing the
# exec command to be non-empty
: ${OCF_RESKEY_monitor_cmd:=/bin/true}

# When OCF_RESKEY_drop_in_dependency is not populated, we
# look at another file-based way of enabling the option.
# Otherwise, consider it disabled.
if [ -z "$OCF_RESKEY_drop_in_dependency" ]; then
	if [ -f "/etc/sysconfig/podman_drop_in" ] || \
	   [ -f "/etc/default/podman_drop_in" ]; then
		OCF_RESKEY_drop_in_dependency=yes
	fi
fi

case $__OCF_ACTION in
meta-data) meta_data
		exit $OCF_SUCCESS;;
start)
	podman_validate
	podman_start;;
stop)		podman_stop;;
monitor)	podman_monitor;;
validate-all)	podman_validate;;
usage|help)	podman_usage
		exit $OCF_SUCCESS
		;;
*)		podman_usage
		exit $OCF_ERR_UNIMPLEMENTED
		;;
esac
rc=$?
ocf_log debug "${OCF_RESOURCE_INSTANCE} $__OCF_ACTION : $rc"
exit $rc