def install_rook(self):
if not self._rook_built:
raise Exception("Rook must be built before being installed")
# TODO(jhesketh): We may want to provide ways for tests to override
# these
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'common.yaml'))
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'operator.yaml'))
# TODO(jhesketh): Check if sleeping is necessary
time.sleep(10)
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'cluster.yaml'))
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'toolbox.yaml'))
time.sleep(10)
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'csi/rbd/storageclass.yaml'))
logger.info("Wait for OSD prepare to complete "
"(this may take a while...)")
pattern = re.compile(r'.*rook-ceph-osd-prepare.*Completed')
common.wait_for_result(self.kubernetes.kubectl,
"--namespace rook-ceph get pods",
matcher=common.regex_count_matcher(pattern, 3),
attempts=90,
interval=10)
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'filesystem.yaml'))
logger.info("Wait for 2 mdses to start")
pattern = re.compile(r'.*rook-ceph-mds-myfs.*Running')
common.wait_for_result(self.kubernetes.kubectl,
"--namespace rook-ceph get pods",
matcher=common.regex_count_matcher(pattern, 2),
attempts=20,
interval=5)
logger.info("Wait for myfs to be active")
pattern = re.compile(r'.*active')
common.wait_for_result(self.execute_in_ceph_toolbox,
"ceph fs status myfs",
matcher=common.regex_matcher(pattern),
attempts=20,
interval=5)
def install(self):
self.kubernetes.kubectl("create namespace rook-ceph")
self._install_operator()
# reduce wait time to discover devices
self.kubernetes.kubectl(
"-n rook-ceph set env "
"deployment/rook-ceph-operator ROOK_DISCOVER_DEVICES_INTERVAL=2m")
# Load cluster.yaml as object
# Modify livenessProbe
# Save yaml somewhere
# apply new yaml
self.kubernetes.kubectl_apply(
self._modify_liveness(os.path.join(self.ceph_dir, 'cluster.yaml')))
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'toolbox.yaml'))
logger.info("Wait for OSD prepare to complete "
"(this may take a while...)")
pattern = re.compile(r'.*rook-ceph-osd-prepare.*Completed')
common.wait_for_result(self.kubernetes.kubectl, "-n rook-ceph get pods"
" -l app=rook-ceph-osd-prepare",
matcher=common.regex_count_matcher(pattern, 3),
attempts=120,
interval=15)
logger.info("Wait for rook-ceph-tools running")
pattern = re.compile(r'.*rook-ceph-tools.*Running')
common.wait_for_result(self.kubernetes.kubectl,
"-n rook-ceph get pods",
matcher=common.regex_count_matcher(pattern, 1),
attempts=30,
interval=10)
# As of Nautilus v14.2.20 and Octopus v15.2.11, clusters intentionally
# come up in HEALTH_WARN (AUTH_INSECURE_GLOBAL_ID_RECLAIM) and to
# achieve HEALTH_OK we have to disallow this insecure feature.
self.execute_in_ceph_toolbox(
"ceph config set mon auth_allow_insecure_global_id_reclaim false"
" || true")
logger.info("Wait for Ceph HEALTH_OK")
pattern = re.compile(r'.*HEALTH_OK')
common.wait_for_result(self.execute_in_ceph_toolbox,
"ceph status",
matcher=common.regex_matcher(pattern),
attempts=60,
interval=10)
logger.info("Rook successfully installed and ready!")
def install(self):
self.kubernetes.kubectl("create namespace rook-ceph")
self._install_operator()
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'cluster.yaml'))
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'toolbox.yaml'))
# reduce wait time to discover devices
self.kubernetes.kubectl(
"-n rook-ceph set env "
"deployment/rook-ceph-operator ROOK_DISCOVER_DEVICES_INTERVAL=2m")
logger.info("Wait for rook-ceph-operator running")
pattern = re.compile(r'.*rook-ceph-operator.*Running')
common.wait_for_result(self.kubernetes.kubectl,
"-n rook-ceph get pods",
matcher=common.regex_count_matcher(pattern, 1),
attempts=30,
interval=10)
logger.info("Wait for OSD prepare to complete "
"(this may take a while...)")
pattern = re.compile(r'.*rook-ceph-osd-prepare.*Completed')
common.wait_for_result(self.kubernetes.kubectl, "-n rook-ceph get pods"
" -l app=rook-ceph-osd-prepare",
matcher=common.regex_count_matcher(pattern, 3),
attempts=120,
interval=15)
logger.info("Wait for rook-ceph-tools running")
pattern = re.compile(r'.*rook-ceph-tools.*Running')
common.wait_for_result(self.kubernetes.kubectl,
"-n rook-ceph get pods",
matcher=common.regex_count_matcher(pattern, 1),
attempts=30,
interval=10)
logger.info("Wait for Ceph HEALTH_OK")
pattern = re.compile(r'.*HEALTH_OK')
common.wait_for_result(self.execute_in_ceph_toolbox,
"ceph status",
matcher=common.regex_matcher(pattern),
attempts=60,
interval=10)
logger.info("Rook successfully installed and ready!")
def wait_for_pods_by_app_label(self, label, count=1, sleep=5, attempts=120,
namespace="rook-ceph"):
pattern = re.compile(r'.*Running')
common.wait_for_result(
self.kubectl,
f'-n {namespace} get pod -l app="{label}" --no-headers',
matcher=common.regex_count_matcher(pattern, count),
attempts=attempts, interval=sleep)
def install_kubernetes(self):
super().install_kubernetes()
self.join(self.hardware.workers)
logger.info("Wait for all nodes to be in 'Ready' state"
"(this may take a while...)")
pattern = re.compile(r' +Ready.*')
common.wait_for_result(self.workspace.execute,
f"{self._skuba} cluster status",
matcher=common.regex_count_matcher(
pattern, len(self.hardware.nodes)),
attempts=60,
interval=10,
chdir=self._clusterpath,
capture=True)
# Give skupa-update/zypper some time to release lock
time.sleep(5)
def deploy_filesystem(self):
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'filesystem.yaml'))
logger.info("Wait for 2 mdses to start")
pattern = re.compile(r'.*rook-ceph-mds-myfs.*Running')
common.wait_for_result(
self.kubernetes.kubectl, "-n rook-ceph get pods",
log_stdout=False,
matcher=common.regex_count_matcher(pattern, 2),
attempts=120, interval=10)
logger.info("Wait for myfs to be active")
pattern = re.compile(r'.*active')
common.wait_for_result(
self.execute_in_ceph_toolbox, "ceph fs status myfs",
log_stdout=False,
matcher=common.regex_matcher(pattern),
attempts=120, interval=10)
logger.info("Ceph FS successfully installed and ready!")
def _install_operator(self):
"""
Install operator using either kubectl of helm
"""
if settings.OPERATOR_INSTALLER == "helm":
logger.info('Deploying rook operator - using Helm')
self._install_operator_helm()
else:
logger.info('Deploying rook operator - using kubectl apply ...')
self._install_operator_kubectl()
logger.info("Wait for rook-ceph-operator running")
pattern = re.compile(r'.*rook-ceph-operator.*Running')
common.wait_for_result(self.kubernetes.kubectl,
"-n rook-ceph get pods",
matcher=common.regex_count_matcher(pattern, 1),
attempts=30,
interval=10)
# set operator log level
self.kubernetes.kubectl(
"--namespace rook-ceph set env "
"deployment/rook-ceph-operator ROOK_LOG_LEVEL=DEBUG")
def install(self):
# TODO(jhesketh): We may want to provide ways for tests to override
# these
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'common.yaml'))
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'operator.yaml'))
# TODO(jhesketh): Check if sleeping is necessary
time.sleep(10)
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'cluster.yaml'))
self.kubernetes.kubectl_apply(
os.path.join(self.ceph_dir, 'toolbox.yaml'))
logger.info("Wait for OSD prepare to complete "
"(this may take a while...)")
pattern = re.compile(r'.*rook-ceph-osd-prepare.*Completed')
common.wait_for_result(self.kubernetes.kubectl,
"--namespace rook-ceph get pods",
matcher=common.regex_count_matcher(pattern, 3),
attempts=90,
interval=10)
logger.info("Wait for Ceph HEALTH_OK")
pattern = re.compile(r'.*HEALTH_OK')
common.wait_for_result(self.execute_in_ceph_toolbox,
"ceph status",
matcher=common.regex_matcher(pattern),
attempts=20,
interval=5)
logger.info("Rook successfully installed and ready!")
