def run_ocs_upgrade(operation=None, *operation_args, **operation_kwargs):
"""
Run upgrade procedure of OCS cluster
Args:
operation: (function): Function to run
operation_args: (iterable): Function's arguments
operation_kwargs: (map): Function's keyword arguments
"""
ceph_cluster = CephCluster()
original_ocs_version = config.ENV_DATA.get("ocs_version")
upgrade_in_current_source = config.UPGRADE.get("upgrade_in_current_source",
False)
upgrade_ocs = OCSUpgrade(
namespace=config.ENV_DATA["cluster_namespace"],
version_before_upgrade=original_ocs_version,
ocs_registry_image=config.UPGRADE.get("upgrade_ocs_registry_image"),
upgrade_in_current_source=upgrade_in_current_source,
)
upgrade_version = upgrade_ocs.get_upgrade_version()
assert (
upgrade_ocs.get_parsed_versions()[1] >=
upgrade_ocs.get_parsed_versions()[0]), (
f"Version you would like to upgrade to: {upgrade_version} "
f"is not higher or equal to the version you currently running: "
f"{upgrade_ocs.version_before_upgrade}")
# For external cluster , create the secrets if upgraded version is 4.8
if (config.DEPLOYMENT["external_mode"] and original_ocs_version == "4.7"
and upgrade_version == "4.8"):
access_key = config.EXTERNAL_MODE.get("access_key_rgw-admin-ops-user",
"")
secret_key = config.EXTERNAL_MODE.get("secret_key_rgw-admin-ops-user",
"")
if not (access_key and secret_key):
raise ExternalClusterRGWAdminOpsUserException(
"Access and secret key for rgw-admin-ops-user not found")
cmd = (
f'oc create secret generic --type="kubernetes.io/rook"'
f' "rgw-admin-ops-user" --from-literal=accessKey={access_key} --from-literal=secretKey={secret_key}'
)
exec_cmd(cmd)
csv_name_pre_upgrade = upgrade_ocs.get_csv_name_pre_upgrade()
pre_upgrade_images = upgrade_ocs.get_pre_upgrade_image(
csv_name_pre_upgrade)
upgrade_ocs.load_version_config_file(upgrade_version)
if config.DEPLOYMENT.get("disconnected"):
upgrade_ocs.ocs_registry_image = prepare_disconnected_ocs_deployment(
upgrade=True)
log.info(
f"Disconnected upgrade - new image: {upgrade_ocs.ocs_registry_image}"
)
with CephHealthMonitor(ceph_cluster):
channel = upgrade_ocs.set_upgrade_channel()
upgrade_ocs.set_upgrade_images()
upgrade_ocs.update_subscription(channel)
if original_ocs_version == "4.8" and upgrade_version == "4.9":
deployment = Deployment()
deployment.subscribe_ocs()
if (config.ENV_DATA["platform"] == constants.IBMCLOUD_PLATFORM
) and not (upgrade_in_current_source):
for attempt in range(2):
# We need to do it twice, because some of the SA are updated
# after the first load of OCS pod after upgrade. So we need to
# link updated SA again.
log.info(f"Sleep 1 minute before attempt: {attempt+1}/2 "
"of linking secret/SAs")
time.sleep(60)
link_all_sa_and_secret_and_delete_pods(
constants.OCS_SECRET, config.ENV_DATA["cluster_namespace"])
if operation:
log.info(f"Calling test function: {operation}")
_ = operation(*operation_args, **operation_kwargs)
# Workaround for issue #2531
time.sleep(30)
# End of workaround
for sample in TimeoutSampler(
timeout=725,
sleep=5,
func=upgrade_ocs.check_if_upgrade_completed,
channel=channel,
csv_name_pre_upgrade=csv_name_pre_upgrade,
):
try:
if sample:
log.info("Upgrade success!")
break
except TimeoutException:
raise TimeoutException("No new CSV found after upgrade!")
old_image = upgrade_ocs.get_images_post_upgrade(
channel, pre_upgrade_images, upgrade_version)
verify_image_versions(
old_image,
upgrade_ocs.get_parsed_versions()[1],
upgrade_ocs.version_before_upgrade,
)
ocs_install_verification(
timeout=600,
skip_osd_distribution_check=True,
ocs_registry_image=upgrade_ocs.ocs_registry_image,
post_upgrade_verification=True,
version_before_upgrade=upgrade_ocs.version_before_upgrade,
)
def run_ocs_upgrade(operation=None, *operation_args, **operation_kwargs):
"""
Run upgrade procedure of OCS cluster
Args:
operation: (function): Function to run
operation_args: (iterable): Function's arguments
operation_kwargs: (map): Function's keyword arguments
"""
ceph_cluster = CephCluster()
original_ocs_version = config.ENV_DATA.get("ocs_version")
upgrade_in_current_source = config.UPGRADE.get("upgrade_in_current_source",
False)
upgrade_ocs = OCSUpgrade(
namespace=config.ENV_DATA["cluster_namespace"],
version_before_upgrade=original_ocs_version,
ocs_registry_image=config.UPGRADE.get("upgrade_ocs_registry_image"),
upgrade_in_current_source=upgrade_in_current_source,
)
upgrade_version = upgrade_ocs.get_upgrade_version()
assert (
upgrade_ocs.get_parsed_versions()[1] >=
upgrade_ocs.get_parsed_versions()[0]), (
f"Version you would like to upgrade to: {upgrade_version} "
f"is not higher or equal to the version you currently running: "
f"{upgrade_ocs.version_before_upgrade}")
# create external cluster object
if config.DEPLOYMENT["external_mode"]:
host, user, password = get_external_cluster_client()
external_cluster = ExternalCluster(host, user, password)
# For external cluster , create the secrets if upgraded version is 4.8
if (config.DEPLOYMENT["external_mode"] and original_ocs_version == "4.7"
and upgrade_version == "4.8"):
external_cluster.create_object_store_user()
access_key = config.EXTERNAL_MODE.get("access_key_rgw-admin-ops-user",
"")
secret_key = config.EXTERNAL_MODE.get("secret_key_rgw-admin-ops-user",
"")
if not (access_key and secret_key):
raise ExternalClusterRGWAdminOpsUserException(
"Access and secret key for rgw-admin-ops-user not found")
cmd = (
f'oc create secret generic --type="kubernetes.io/rook"'
f' "rgw-admin-ops-user" --from-literal=accessKey={access_key} --from-literal=secretKey={secret_key}'
)
exec_cmd(cmd)
csv_name_pre_upgrade = upgrade_ocs.get_csv_name_pre_upgrade()
pre_upgrade_images = upgrade_ocs.get_pre_upgrade_image(
csv_name_pre_upgrade)
upgrade_ocs.load_version_config_file(upgrade_version)
if config.DEPLOYMENT.get("disconnected") and not config.DEPLOYMENT.get(
"disconnected_env_skip_image_mirroring"):
upgrade_ocs.ocs_registry_image = prepare_disconnected_ocs_deployment(
upgrade=True)
log.info(
f"Disconnected upgrade - new image: {upgrade_ocs.ocs_registry_image}"
)
with CephHealthMonitor(ceph_cluster):
channel = upgrade_ocs.set_upgrade_channel()
upgrade_ocs.set_upgrade_images()
live_deployment = config.DEPLOYMENT["live_deployment"]
disable_addon = config.DEPLOYMENT.get("ibmcloud_disable_addon")
if (config.ENV_DATA["platform"] == constants.IBMCLOUD_PLATFORM
and live_deployment and not disable_addon):
clustername = config.ENV_DATA.get("cluster_name")
cmd = f"ibmcloud ks cluster addon disable openshift-data-foundation --cluster {clustername} -f"
run_ibmcloud_cmd(cmd)
time.sleep(120)
cmd = (
f"ibmcloud ks cluster addon enable openshift-data-foundation --cluster {clustername} -f --version "
f"{upgrade_version}.0 --param ocsUpgrade=true")
run_ibmcloud_cmd(cmd)
time.sleep(120)
else:
ui_upgrade_supported = False
if config.UPGRADE.get("ui_upgrade"):
if (version.get_semantic_ocp_version_from_config()
== version.VERSION_4_9
and original_ocs_version == "4.8"
and upgrade_version == "4.9"):
ui_upgrade_supported = True
else:
log.warning(
"UI upgrade combination is not supported. It will fallback to CLI upgrade"
)
if ui_upgrade_supported:
ocs_odf_upgrade_ui()
else:
if (config.ENV_DATA["platform"] == constants.IBMCLOUD_PLATFORM
) and not (upgrade_in_current_source):
create_ocs_secret(config.ENV_DATA["cluster_namespace"])
if upgrade_version != "4.9":
# In the case of upgrade to ODF 4.9, the ODF operator should upgrade
# OCS automatically.
upgrade_ocs.update_subscription(channel)
if original_ocs_version == "4.8" and upgrade_version == "4.9":
deployment = Deployment()
deployment.subscribe_ocs()
else:
# In the case upgrade is not from 4.8 to 4.9 and we have manual approval strategy
# we need to wait and approve install plan, otherwise it's approved in the
# subscribe_ocs method.
subscription_plan_approval = config.DEPLOYMENT.get(
"subscription_plan_approval")
if subscription_plan_approval == "Manual":
wait_for_install_plan_and_approve(
config.ENV_DATA["cluster_namespace"])
if (config.ENV_DATA["platform"] == constants.IBMCLOUD_PLATFORM
) and not (upgrade_in_current_source):
for attempt in range(2):
# We need to do it twice, because some of the SA are updated
# after the first load of OCS pod after upgrade. So we need to
# link updated SA again.
log.info(
f"Sleep 1 minute before attempt: {attempt + 1}/2 "
"of linking secret/SAs")
time.sleep(60)
link_all_sa_and_secret_and_delete_pods(
constants.OCS_SECRET,
config.ENV_DATA["cluster_namespace"])
if operation:
log.info(f"Calling test function: {operation}")
_ = operation(*operation_args, **operation_kwargs)
# Workaround for issue #2531
time.sleep(30)
# End of workaround
for sample in TimeoutSampler(
timeout=725,
sleep=5,
func=upgrade_ocs.check_if_upgrade_completed,
channel=channel,
csv_name_pre_upgrade=csv_name_pre_upgrade,
):
try:
if sample:
log.info("Upgrade success!")
break
except TimeoutException:
raise TimeoutException("No new CSV found after upgrade!")
old_image = upgrade_ocs.get_images_post_upgrade(
channel, pre_upgrade_images, upgrade_version)
verify_image_versions(
old_image,
upgrade_ocs.get_parsed_versions()[1],
upgrade_ocs.version_before_upgrade,
)
# update external secrets
if config.DEPLOYMENT["external_mode"]:
upgrade_version = version.get_semantic_version(upgrade_version, True)
if upgrade_version >= version.VERSION_4_10:
external_cluster.update_permission_caps()
else:
external_cluster.update_permission_caps(EXTERNAL_CLUSTER_USER)
external_cluster.get_external_cluster_details()
# update the external cluster details in secrets
log.info("updating external cluster secret")
external_cluster_details = NamedTemporaryFile(
mode="w+",
prefix="external-cluster-details-",
delete=False,
)
with open(external_cluster_details.name, "w") as fd:
decoded_external_cluster_details = decode(
config.EXTERNAL_MODE["external_cluster_details"])
fd.write(decoded_external_cluster_details)
cmd = (
f"oc set data secret/rook-ceph-external-cluster-details -n {constants.OPENSHIFT_STORAGE_NAMESPACE} "
f"--from-file=external_cluster_details={external_cluster_details.name}"
)
exec_cmd(cmd)
if config.ENV_DATA.get("mcg_only_deployment"):
mcg_only_install_verification(
ocs_registry_image=upgrade_ocs.ocs_registry_image)
else:
ocs_install_verification(
timeout=600,
skip_osd_distribution_check=True,
ocs_registry_image=upgrade_ocs.ocs_registry_image,
post_upgrade_verification=True,
version_before_upgrade=upgrade_ocs.version_before_upgrade,
)