def get_hadoop_conf_dir():
"""
Return the hadoop shared conf directory which should be used for the command's component. The
directory including the component's version is tried first, but if that doesn't exist,
this will fallback to using "current".
"""
stack_root = Script.get_stack_root()
stack_version = Script.get_stack_version()
hadoop_conf_dir = os.path.join(os.path.sep, "etc", "hadoop", "conf")
if check_stack_feature(StackFeature.CONFIG_VERSIONING, stack_version):
# read the desired version from the component map and use that for building the hadoop home
version = component_version.get_component_repository_version()
if version is None:
version = default("/commandParams/version", None)
hadoop_conf_dir = os.path.join(stack_root, str(version), "hadoop",
"conf")
if version is None or sudo.path_isdir(hadoop_conf_dir) is False:
hadoop_conf_dir = os.path.join(stack_root, "current",
"hadoop-client", "conf")
Logger.info("Using hadoop conf dir: {0}".format(hadoop_conf_dir))
return hadoop_conf_dir
def get_hadoop_dir(target):
"""
Return the hadoop shared directory which should be used for the command's component. The
directory including the component's version is tried first, but if that doesn't exist,
this will fallback to using "current".
:target: the target directory
"""
stack_root = Script.get_stack_root()
stack_version = Script.get_stack_version()
if not target in HADOOP_DIR_DEFAULTS:
raise Fail("Target {0} not defined".format(target))
hadoop_dir = HADOOP_DIR_DEFAULTS[target]
formatted_stack_version = format_stack_version(stack_version)
if stack_features.check_stack_feature(StackFeature.ROLLING_UPGRADE, formatted_stack_version):
# read the desired version from the component map and use that for building the hadoop home
version = component_version.get_component_repository_version()
if version is None:
version = default("/commandParams/version", None)
# home uses a different template
if target == "home":
hadoop_dir = HADOOP_HOME_DIR_TEMPLATE.format(stack_root, version, "hadoop")
if version is None or sudo.path_isdir(hadoop_dir) is False:
hadoop_dir = HADOOP_HOME_DIR_TEMPLATE.format(stack_root, "current", "hadoop-client")
else:
hadoop_dir = HADOOP_DIR_TEMPLATE.format(stack_root, version, "hadoop", target)
if version is None or sudo.path_isdir(hadoop_dir) is False:
hadoop_dir = HADOOP_DIR_TEMPLATE.format(stack_root, "current", "hadoop-client", target)
return hadoop_dir
def get_hadoop_dir(target, force_latest_on_upgrade=False):
"""
Return the hadoop shared directory in the following override order
1. Use default for 2.1 and lower
2. If 2.2 and higher, use <stack-root>/current/hadoop-client/{target}
3. If 2.2 and higher AND for an upgrade, use <stack-root>/<version>/hadoop/{target}.
However, if the upgrade has not yet invoked <stack-selector-tool>, return the current
version of the component.
:target: the target directory
:force_latest_on_upgrade: if True, then this will return the "current" directory
without the stack version built into the path, such as <stack-root>/current/hadoop-client
"""
stack_root = Script.get_stack_root()
stack_version = Script.get_stack_version()
if not target in HADOOP_DIR_DEFAULTS:
raise Fail("Target {0} not defined".format(target))
hadoop_dir = HADOOP_DIR_DEFAULTS[target]
formatted_stack_version = format_stack_version(stack_version)
if formatted_stack_version and check_stack_feature(
StackFeature.ROLLING_UPGRADE, formatted_stack_version):
# home uses a different template
if target == "home":
hadoop_dir = HADOOP_HOME_DIR_TEMPLATE.format(
stack_root, "current", "hadoop-client")
else:
hadoop_dir = HADOOP_DIR_TEMPLATE.format(stack_root, "current",
"hadoop-client", target)
# if we are not forcing "current" for HDP 2.2, then attempt to determine
# if the exact version needs to be returned in the directory
if not force_latest_on_upgrade:
stack_info = _get_upgrade_stack()
if stack_info is not None:
stack_version = stack_info[1]
# determine if <stack-selector-tool> has been run and if not, then use the current
# hdp version until this component is upgraded
current_stack_version = get_role_component_current_stack_version(
)
if current_stack_version is not None and stack_version != current_stack_version:
stack_version = current_stack_version
if target == "home":
# home uses a different template
hadoop_dir = HADOOP_HOME_DIR_TEMPLATE.format(
stack_root, stack_version, "hadoop")
else:
hadoop_dir = HADOOP_DIR_TEMPLATE.format(
stack_root, stack_version, "hadoop", target)
return hadoop_dir
def get_hadoop_dir(target, force_latest_on_upgrade=False):
"""
Return the hadoop shared directory in the following override order
1. Use default for 2.1 and lower
2. If 2.2 and higher, use <stack-root>/current/hadoop-client/{target}
3. If 2.2 and higher AND for an upgrade, use <stack-root>/<version>/hadoop/{target}.
However, if the upgrade has not yet invoked <stack-selector-tool>, return the current
version of the component.
:target: the target directory
:force_latest_on_upgrade: if True, then this will return the "current" directory
without the stack version built into the path, such as <stack-root>/current/hadoop-client
"""
stack_root = Script.get_stack_root()
stack_version = Script.get_stack_version()
if not target in HADOOP_DIR_DEFAULTS:
raise Fail("Target {0} not defined".format(target))
hadoop_dir = HADOOP_DIR_DEFAULTS[target]
formatted_stack_version = format_stack_version(stack_version)
if formatted_stack_version and check_stack_feature(StackFeature.ROLLING_UPGRADE, formatted_stack_version):
# home uses a different template
if target == "home":
hadoop_dir = HADOOP_HOME_DIR_TEMPLATE.format(stack_root, "current", "hadoop-client")
else:
hadoop_dir = HADOOP_DIR_TEMPLATE.format(stack_root, "current", "hadoop-client", target)
# if we are not forcing "current" for HDP 2.2, then attempt to determine
# if the exact version needs to be returned in the directory
if not force_latest_on_upgrade:
stack_info = _get_upgrade_stack()
if stack_info is not None:
stack_version = stack_info[1]
# determine if <stack-selector-tool> has been run and if not, then use the current
# hdp version until this component is upgraded
current_stack_version = get_role_component_current_stack_version()
if current_stack_version is not None and stack_version != current_stack_version:
stack_version = current_stack_version
if target == "home":
# home uses a different template
hadoop_dir = HADOOP_HOME_DIR_TEMPLATE.format(stack_root, stack_version, "hadoop")
else:
hadoop_dir = HADOOP_DIR_TEMPLATE.format(stack_root, stack_version, "hadoop", target)
return hadoop_dir
def get_hadoop_conf_dir(force_latest_on_upgrade=False):
"""
Gets the shared hadoop conf directory using:
1. Start with /etc/hadoop/conf
2. When the stack is greater than HDP-2.2, use <stack-root>/current/hadoop-client/conf
3. Only when doing a RU and HDP-2.3 or higher, use the value as computed
by <conf-selector-tool>. This is in the form <stack-root>/VERSION/hadoop/conf to make sure
the configs are written in the correct place. However, if the component itself has
not yet been upgraded, it should use the hadoop configs from the prior version.
This will perform an <stack-selector-tool> status to determine which version to use.
:param force_latest_on_upgrade: if True, then force the returned path to always
be that of the upgrade target version, even if <stack-selector-tool> has not been called. This
is primarily used by hooks like before-ANY to ensure that hadoop environment
configurations are written to the correct location since they are written out
before the <stack-selector-tool>/<conf-selector-tool> would have been called.
"""
hadoop_conf_dir = "/etc/hadoop/conf"
stack_name = None
stack_root = Script.get_stack_root()
stack_version = Script.get_stack_version()
version = None
allow_setting_conf_select_symlink = False
if not Script.in_stack_upgrade():
# During normal operation, the HDP stack must be 2.3 or higher
if stack_version and check_stack_feature(StackFeature.ROLLING_UPGRADE,
stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current",
"hadoop-client", "conf")
if stack_version and check_stack_feature(
StackFeature.CONFIG_VERSIONING, stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current",
"hadoop-client", "conf")
stack_name = default("/hostLevelParams/stack_name", None)
version = default("/commandParams/version", None)
if not os.path.islink(hadoop_conf_dir) and stack_name and version:
version = str(version)
allow_setting_conf_select_symlink = True
else:
# During an upgrade/downgrade, which can be a Rolling or Express Upgrade, need to calculate it based on the version
'''
Whenever upgrading to HDP 2.2, or downgrading back to 2.2, need to use /etc/hadoop/conf
Whenever upgrading to HDP 2.3, or downgrading back to 2.3, need to use a versioned hadoop conf dir
Type__|_Source_|_Target_|_Direction_____________|_Comment_____________________________________________________________
Normal| | 2.2 | | Use /etc/hadoop/conf
Normal| | 2.3 | | Use /etc/hadoop/conf, which should be a symlink to <stack-root>/current/hadoop-client/conf
EU | 2.1 | 2.3 | Upgrade | Use versioned <stack-root>/current/hadoop-client/conf
| | | No Downgrade Allowed | Invalid
EU/RU | 2.2 | 2.2.* | Any | Use <stack-root>/current/hadoop-client/conf
EU/RU | 2.2 | 2.3 | Upgrade | Use <stack-root>/$version/hadoop/conf, which should be a symlink destination
| | | Downgrade | Use <stack-root>/current/hadoop-client/conf
EU/RU | 2.3 | 2.3.* | Any | Use <stack-root>/$version/hadoop/conf, which should be a symlink destination
'''
# The "stack_version" is the desired stack, e.g., 2.2 or 2.3
# In an RU, it is always the desired stack, and doesn't change even during the Downgrade!
# In an RU Downgrade from HDP 2.3 to 2.2, the first thing we do is
# rm /etc/[component]/conf and then mv /etc/[component]/conf.backup /etc/[component]/conf
if stack_version and check_stack_feature(StackFeature.ROLLING_UPGRADE,
stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current",
"hadoop-client", "conf")
# This contains the "version", including the build number, that is actually used during a stack upgrade and
# is the version upgrading/downgrading to.
stack_info = stack_select._get_upgrade_stack()
if stack_info is not None:
stack_name = stack_info[0]
version = stack_info[1]
else:
raise Fail("Unable to get parameter 'version'")
Logger.info(
"In the middle of a stack upgrade/downgrade for Stack {0} and destination version {1}, determining which hadoop conf dir to use."
.format(stack_name, version))
# This is the version either upgrading or downgrading to.
if version and check_stack_feature(StackFeature.CONFIG_VERSIONING,
version):
# Determine if <stack-selector-tool> has been run and if not, then use the current
# hdp version until this component is upgraded.
if not force_latest_on_upgrade:
current_stack_version = stack_select.get_role_component_current_stack_version(
)
if current_stack_version is not None and version != current_stack_version:
version = current_stack_version
stack_selector_name = stack_tools.get_stack_tool_name(
stack_tools.STACK_SELECTOR_NAME)
Logger.info(
"{0} has not yet been called to update the symlink for this component, "
"keep using version {1}".format(
stack_selector_name, current_stack_version))
# Only change the hadoop_conf_dir path, don't <conf-selector-tool> this older version
hadoop_conf_dir = os.path.join(stack_root, version, "hadoop",
"conf")
Logger.info("Hadoop conf dir: {0}".format(hadoop_conf_dir))
allow_setting_conf_select_symlink = True
if allow_setting_conf_select_symlink:
# If not in the middle of an upgrade and on HDP 2.3 or higher, or if
# upgrading stack to version 2.3.0.0 or higher (which may be upgrade or downgrade), then consider setting the
# symlink for /etc/hadoop/conf.
# If a host does not have any HDFS or YARN components (e.g., only ZK), then it will not contain /etc/hadoop/conf
# Therefore, any calls to <conf-selector-tool> will fail.
# For that reason, if the hadoop conf directory exists, then make sure it is set.
if os.path.exists(hadoop_conf_dir):
conf_selector_name = stack_tools.get_stack_tool_name(
stack_tools.CONF_SELECTOR_NAME)
Logger.info(
"The hadoop conf dir {0} exists, will call {1} on it for version {2}"
.format(hadoop_conf_dir, conf_selector_name, version))
select(stack_name, "hadoop", version)
Logger.info("Using hadoop conf dir: {0}".format(hadoop_conf_dir))
return hadoop_conf_dir
def get_hadoop_conf_dir(force_latest_on_upgrade=False):
"""
Gets the shared hadoop conf directory using:
1. Start with /etc/hadoop/conf
2. When the stack is greater than HDP-2.2, use <stack-root>/current/hadoop-client/conf
3. Only when doing a RU and HDP-2.3 or higher, use the value as computed
by <conf-selector-tool>. This is in the form <stack-root>/VERSION/hadoop/conf to make sure
the configs are written in the correct place. However, if the component itself has
not yet been upgraded, it should use the hadoop configs from the prior version.
This will perform an <stack-selector-tool> status to determine which version to use.
:param force_latest_on_upgrade: if True, then force the returned path to always
be that of the upgrade target version, even if <stack-selector-tool> has not been called. This
is primarily used by hooks like before-ANY to ensure that hadoop environment
configurations are written to the correct location since they are written out
before the <stack-selector-tool>/<conf-selector-tool> would have been called.
"""
hadoop_conf_dir = "/etc/hadoop/conf"
stack_name = None
stack_root = Script.get_stack_root()
stack_version = Script.get_stack_version()
version = None
allow_setting_conf_select_symlink = False
if not Script.in_stack_upgrade():
# During normal operation, the HDP stack must be 2.3 or higher
if stack_version and check_stack_feature(StackFeature.ROLLING_UPGRADE, stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current", "hadoop-client", "conf")
if stack_version and check_stack_feature(StackFeature.CONFIG_VERSIONING, stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current", "hadoop-client", "conf")
stack_name = default("/hostLevelParams/stack_name", None)
version = default("/commandParams/version", None)
if stack_name and version:
version = str(version)
allow_setting_conf_select_symlink = True
else:
# During an upgrade/downgrade, which can be a Rolling or Express Upgrade, need to calculate it based on the version
'''
Whenever upgrading to HDP 2.2, or downgrading back to 2.2, need to use /etc/hadoop/conf
Whenever upgrading to HDP 2.3, or downgrading back to 2.3, need to use a versioned hadoop conf dir
Type__|_Source_|_Target_|_Direction_____________|_Comment_____________________________________________________________
Normal| | 2.2 | | Use /etc/hadoop/conf
Normal| | 2.3 | | Use /etc/hadoop/conf, which should be a symlink to <stack-root>/current/hadoop-client/conf
EU | 2.1 | 2.3 | Upgrade | Use versioned <stack-root>/current/hadoop-client/conf
| | | No Downgrade Allowed | Invalid
EU/RU | 2.2 | 2.2.* | Any | Use <stack-root>/current/hadoop-client/conf
EU/RU | 2.2 | 2.3 | Upgrade | Use <stack-root>/$version/hadoop/conf, which should be a symlink destination
| | | Downgrade | Use <stack-root>/current/hadoop-client/conf
EU/RU | 2.3 | 2.3.* | Any | Use <stack-root>/$version/hadoop/conf, which should be a symlink destination
'''
# The "stack_version" is the desired stack, e.g., 2.2 or 2.3
# In an RU, it is always the desired stack, and doesn't change even during the Downgrade!
# In an RU Downgrade from HDP 2.3 to 2.2, the first thing we do is
# rm /etc/[component]/conf and then mv /etc/[component]/conf.backup /etc/[component]/conf
if stack_version and check_stack_feature(StackFeature.ROLLING_UPGRADE, stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current", "hadoop-client", "conf")
# This contains the "version", including the build number, that is actually used during a stack upgrade and
# is the version upgrading/downgrading to.
stack_info = stack_select._get_upgrade_stack()
if stack_info is not None:
stack_name = stack_info[0]
version = stack_info[1]
else:
raise Fail("Unable to get parameter 'version'")
Logger.info("In the middle of a stack upgrade/downgrade for Stack {0} and destination version {1}, determining which hadoop conf dir to use.".format(stack_name, version))
# This is the version either upgrading or downgrading to.
if version and check_stack_feature(StackFeature.CONFIG_VERSIONING, version):
# Determine if <stack-selector-tool> has been run and if not, then use the current
# hdp version until this component is upgraded.
if not force_latest_on_upgrade:
current_stack_version = stack_select.get_role_component_current_stack_version()
if current_stack_version is not None and version != current_stack_version:
version = current_stack_version
stack_selector_name = stack_tools.get_stack_tool_name(stack_tools.STACK_SELECTOR_NAME)
Logger.info("{0} has not yet been called to update the symlink for this component, "
"keep using version {1}".format(stack_selector_name, current_stack_version))
# Only change the hadoop_conf_dir path, don't <conf-selector-tool> this older version
hadoop_conf_dir = os.path.join(stack_root, version, "hadoop", "conf")
Logger.info("Hadoop conf dir: {0}".format(hadoop_conf_dir))
allow_setting_conf_select_symlink = True
if allow_setting_conf_select_symlink:
# If not in the middle of an upgrade and on HDP 2.3 or higher, or if
# upgrading stack to version 2.3.0.0 or higher (which may be upgrade or downgrade), then consider setting the
# symlink for /etc/hadoop/conf.
# If a host does not have any HDFS or YARN components (e.g., only ZK), then it will not contain /etc/hadoop/conf
# Therefore, any calls to <conf-selector-tool> will fail.
# For that reason, if the hadoop conf directory exists, then make sure it is set.
if os.path.exists(hadoop_conf_dir):
conf_selector_name = stack_tools.get_stack_tool_name(stack_tools.CONF_SELECTOR_NAME)
Logger.info("The hadoop conf dir {0} exists, will call {1} on it for version {2}".format(
hadoop_conf_dir, conf_selector_name, version))
select(stack_name, "hadoop", version)
Logger.info("Using hadoop conf dir: {0}".format(hadoop_conf_dir))
return hadoop_conf_dir
def get_hadoop_conf_dir(force_latest_on_upgrade=False):
"""
Gets the shared hadoop conf directory using:
1. Start with /etc/hadoop/conf
2. When the stack is greater than HDP-2.2, use <stack-root>/current/hadoop-client/conf
3. Only when doing a RU and HDP-2.3 or higher, use the value as computed
by <conf-selector-tool>. This is in the form <stack-root>/VERSION/hadoop/conf to make sure
the configs are written in the correct place. However, if the component itself has
not yet been upgraded, it should use the hadoop configs from the prior version.
This will perform an <stack-selector-tool> status to determine which version to use.
:param force_latest_on_upgrade: if True, then force the returned path to always
be that of the upgrade target version, even if <stack-selector-tool> has not been called. This
is primarily used by hooks like before-ANY to ensure that hadoop environment
configurations are written to the correct location since they are written out
before the <stack-selector-tool>/<conf-selector-tool> would have been called.
"""
hadoop_conf_dir = "/etc/hadoop/conf"
stack_name = None
stack_root = Script.get_stack_root()
stack_version = Script.get_stack_version()
version = None
if not Script.in_stack_upgrade():
# During normal operation, the HDP stack must be 2.3 or higher
if stack_version and check_stack_feature(StackFeature.ROLLING_UPGRADE,
stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current",
"hadoop-client", "conf")
if stack_version and check_stack_feature(
StackFeature.CONFIG_VERSIONING, stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current",
"hadoop-client", "conf")
stack_name = default("/hostLevelParams/stack_name", None)
version = default("/commandParams/version", None)
if not os.path.islink(hadoop_conf_dir) and stack_name and version:
version = str(version)
else:
# The "stack_version" is the desired stack, e.g., 2.2 or 2.3
# In an RU, it is always the desired stack, and doesn't change even during the Downgrade!
# In an RU Downgrade from HDP 2.3 to 2.2, the first thing we do is
# rm /etc/[component]/conf and then mv /etc/[component]/conf.backup /etc/[component]/conf
if stack_version and check_stack_feature(StackFeature.ROLLING_UPGRADE,
stack_version):
hadoop_conf_dir = os.path.join(stack_root, "current",
"hadoop-client", "conf")
# This contains the "version", including the build number, that is actually used during a stack upgrade and
# is the version upgrading/downgrading to.
stack_info = stack_select._get_upgrade_stack()
if stack_info is None:
raise Fail(
"Unable to retrieve the upgrade/downgrade stack information from the request"
)
stack_name = stack_info[0]
version = stack_info[1]
Logger.info(
"An upgrade/downgrade for {0}-{1} is in progress, determining which hadoop conf dir to use."
.format(stack_name, version))
# This is the version either upgrading or downgrading to.
if version and check_stack_feature(StackFeature.CONFIG_VERSIONING,
version):
# Determine if <stack-selector-tool> has been run and if not, then use the current
# hdp version until this component is upgraded.
if not force_latest_on_upgrade:
current_stack_version = stack_select.get_role_component_current_stack_version(
)
if current_stack_version is not None and version != current_stack_version:
version = current_stack_version
stack_selector_name = stack_tools.get_stack_tool_name(
stack_tools.STACK_SELECTOR_NAME)
Logger.info(
"{0} has not yet been called to update the symlink for this component, "
"keep using version {1}".format(
stack_selector_name, current_stack_version))
# Only change the hadoop_conf_dir path, don't <conf-selector-tool> this older version
hadoop_conf_dir = os.path.join(stack_root, version, "hadoop",
"conf")
Logger.info("Hadoop conf dir: {0}".format(hadoop_conf_dir))
Logger.info("Using hadoop conf dir: {0}".format(hadoop_conf_dir))
return hadoop_conf_dir