def install_administrative_scripts():
scripts_dir = helpers.scripts_dir()
logs_dir = helpers.logs_dir()
helpers.makedirs(scripts_dir, mode=0o755)
# The database backup script. Most of this is redundant now.
source = os.path.join(hookenv.charm_dir(), 'scripts', 'pgbackup.py')
destination = os.path.join(scripts_dir, 'dump-pg-db')
with open(source, 'r') as f:
helpers.write(destination, f.read(), mode=0o755)
backups_dir = helpers.backups_dir()
helpers.makedirs(backups_dir, mode=0o750,
user='******', group='postgres')
# Generate a wrapper that invokes the backup script for each
# database.
data = dict(logs_dir=logs_dir,
scripts_dir=scripts_dir,
# backups_dir probably should be deprecated in favour of
# a juju storage mount.
backups_dir=backups_dir)
destination = os.path.join(helpers.scripts_dir(), 'pg_backup_job')
templating.render('pg_backup_job.tmpl', destination, data,
owner='root', group='postgres', perms=0o755)
# Install the reaper scripts.
script = 'pgkillidle.py'
source = os.path.join(hookenv.charm_dir(), 'scripts', script)
if (reactive.helpers.any_file_changed([source]) or
not os.path.exists(source)):
destination = os.path.join(scripts_dir, script)
with open(source, 'r') as f:
helpers.write(destination, f.read(), mode=0o755)
if not os.path.exists(logs_dir):
helpers.makedirs(logs_dir, mode=0o755, user='******',
group='postgres')
# Create the backups.log file used by the backup wrapper if it
# does not exist, in order to trigger spurious alerts when a
# unit is installed, per Bug #1329816.
helpers.write(helpers.backups_log_path(), '', mode=0o644,
user='******', group='postgres')
def update_nrpe_config():
update_nagios_pgpass()
nrpe = NRPE()
user = nagios_username()
port = postgresql.port()
nrpe.add_check(shortname="pgsql", description="Check pgsql", check_cmd="check_pgsql -P {} -l {}".format(port, user))
if reactive.is_state("postgresql.replication.is_master"):
# TODO: These should be calcualted from the backup schedule,
# which is difficult since that is specified in crontab format.
warn_age = 172800
crit_age = 194400
backups_log = helpers.backups_log_path()
nrpe.add_check(
shortname="pgsql_backups",
description="Check pgsql backups",
check_cmd=("check_file_age -w {} -c {} -f {}" "".format(warn_age, crit_age, backups_log)),
)
else:
# Standbys don't do backups. We still generate a check though,
# to ensure alerts get through to monitoring after a failover.
nrpe.add_check(
shortname="pgsql_backups",
description="Check pgsql backups",
check_cmd=r"check_dummy 0 standby_does_not_backup",
)
nrpe.write()
reactive.remove_state("postgresql.nagios.needs_update")
def install_administrative_scripts():
scripts_dir = helpers.scripts_dir()
logs_dir = helpers.logs_dir()
helpers.makedirs(scripts_dir, mode=0o755)
# The database backup script. Most of this is redundant now.
source = os.path.join(hookenv.charm_dir(), "scripts", "pgbackup.py")
destination = os.path.join(scripts_dir, "dump-pg-db")
with open(source, "r") as f:
helpers.write(destination, f.read(), mode=0o755)
backups_dir = helpers.backups_dir()
helpers.makedirs(backups_dir, mode=0o750, user="******", group="postgres")
# Generate a wrapper that invokes the backup script for each
# database.
data = dict(
logs_dir=logs_dir,
scripts_dir=scripts_dir,
# backups_dir probably should be deprecated in favour of
# a juju storage mount.
backups_dir=backups_dir,
)
destination = os.path.join(helpers.scripts_dir(), "pg_backup_job")
templating.render(
"pg_backup_job.tmpl",
destination,
data,
owner="root",
group="postgres",
perms=0o755,
)
# Install the reaper scripts.
script = "pgkillidle.py"
source = os.path.join(hookenv.charm_dir(), "scripts", script)
destination = os.path.join(scripts_dir, script)
if reactive.helpers.any_file_changed([source]) or not os.path.exists(destination):
with open(source, "r") as f:
helpers.write(destination, f.read(), mode=0o755)
if not os.path.exists(logs_dir):
helpers.makedirs(logs_dir, mode=0o755, user="******", group="postgres")
# Create the backups.log file used by the backup wrapper if it
# does not exist, in order to trigger spurious alerts when a
# unit is installed, per Bug #1329816.
helpers.write(
helpers.backups_log_path(),
"",
mode=0o644,
user="******",
group="postgres",
)
reactive.set_state("postgresql.cluster.support-scripts")
def update_nrpe_config():
update_nagios_pgpass()
nrpe = NRPE()
user = nagios_username()
port = postgresql.port()
nrpe.add_check(
shortname="pgsql",
description="Check pgsql",
check_cmd="check_pgsql -P {} -l {}".format(port, user),
)
# copy the check script which will run cronned as postgres user
with open("scripts/find_latest_ready_wal.py") as fh:
check_script = fh.read()
check_script_path = "{}/{}".format(helpers.scripts_dir(), "find_latest_ready_wal.py")
helpers.write(check_script_path, check_script, mode=0o755)
# create an (empty) file with appropriate permissions for the above
check_output_path = "/var/lib/nagios/postgres-wal-max-age.txt"
if not os.path.exists(check_output_path):
helpers.write(check_output_path, b"0\n", mode=0o644, user="******", group="postgres")
# retrieve the threshold values from the charm config
config = hookenv.config()
check_warn_threshold = config["wal_archive_warn_threshold"] or 0
check_crit_threshold = config["wal_archive_crit_threshold"] or 0
check_cron_path = "/etc/cron.d/postgres-wal-archive-check"
if check_warn_threshold and check_crit_threshold:
# create the cron job to run the above
check_cron = "*/2 * * * * postgres {}".format(check_script_path)
helpers.write(check_cron_path, check_cron, mode=0o644)
# copy the nagios plugin which will check the cronned output
with open("scripts/check_latest_ready_wal.py") as fh:
check_script = fh.read()
check_script_path = "{}/{}".format("/usr/local/lib/nagios/plugins", "check_latest_ready_wal.py")
helpers.write(check_script_path, check_script, mode=0o755)
# write the nagios check definition
nrpe.add_check(
shortname="pgsql_stale_wal",
description="Check for stale WAL backups",
check_cmd="{} {} {}".format(check_script_path, check_warn_threshold, check_crit_threshold),
)
if reactive.is_state("postgresql.replication.is_master"):
# TODO: These should be calculated from the backup schedule,
# which is difficult since that is specified in crontab format.
warn_age = 172800
crit_age = 194400
backups_log = helpers.backups_log_path()
nrpe.add_check(
shortname="pgsql_backups",
description="Check pgsql backups",
check_cmd=("check_file_age -w {} -c {} -f {}" "".format(warn_age, crit_age, backups_log)),
)
else:
# Standbys don't do backups. We still generate a check though,
# to ensure alerts get through to monitoring after a failover.
nrpe.add_check(
shortname="pgsql_backups",
description="Check pgsql backups",
check_cmd=r"check_dummy 0 standby_does_not_backup",
)
nrpe.write()
reactive.remove_state("postgresql.nagios.needs_update")
