HADOOP-13673. Update scripts to be smarter when running with privilege

Signed-off-by: Andrew Wang <wang@apache.org>
Signed-off-by: Ravi Prakash <raviprak@apache.org>
This commit is contained in:
Allen Wittenauer 2016-08-04 18:08:23 -07:00
parent acda1a5fee
commit 0eb4b513b7
18 changed files with 504 additions and 96 deletions

View File

@ -183,13 +183,24 @@ else
exit 1 exit 1
fi fi
if [ $# = 0 ]; then # now that we have support code, let's abs MYNAME so we can use it later
MYNAME=$(hadoop_abs "${MYNAME}")
if [[ $# = 0 ]]; then
hadoop_exit_with_usage 1 hadoop_exit_with_usage 1
fi fi
HADOOP_SUBCMD=$1 HADOOP_SUBCMD=$1
shift shift
if hadoop_need_reexec hadoop "${HADOOP_SUBCMD}"; then
hadoop_uservar_su hadoop "${HADOOP_SUBCMD}" \
"${MYNAME}" \
"--reexec" \
"${HADOOP_USER_PARAMS[@]}"
exit $?
fi
hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}" hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}"
HADOOP_SUBCMD_ARGS=("$@") HADOOP_SUBCMD_ARGS=("$@")

View File

@ -41,6 +41,42 @@ function hadoop_debug
fi fi
} }
## @description Given a filename or dir, return the absolute version of it
## @description This works as an alternative to readlink, which isn't
## @description portable.
## @audience public
## @stability stable
## @param fsobj
## @replaceable no
## @return 0 success
## @return 1 failure
## @return stdout abspath
function hadoop_abs
{
declare obj=$1
declare dir
declare fn
declare dirret
if [[ ! -e ${obj} ]]; then
return 1
elif [[ -d ${obj} ]]; then
dir=${obj}
else
dir=$(dirname -- "${obj}")
fn=$(basename -- "${obj}")
fn="/${fn}"
fi
dir=$(cd -P -- "${dir}" >/dev/null 2>/dev/null && pwd -P)
dirret=$?
if [[ ${dirret} = 0 ]]; then
echo "${dir}${fn}"
return 0
fi
return 1
}
## @description Given variable $1 delete $2 from it ## @description Given variable $1 delete $2 from it
## @audience public ## @audience public
## @stability stable ## @stability stable
@ -79,6 +115,101 @@ function hadoop_verify_entry
[[ ${!1} =~ \ ${2}\ ]] [[ ${!1} =~ \ ${2}\ ]]
} }
## @description Check if we are running with privilege
## @description by default, this implementation looks for
## @description EUID=0. For OSes that have true privilege
## @description separation, this should be something more complex
## @audience private
## @stability evolving
## @replaceable yes
## @return 1 = no priv
## @return 0 = priv
function hadoop_privilege_check
{
[[ "${EUID}" = 0 ]]
}
## @description Execute a command via su when running as root
## @description if the given user is found or exit with
## @description failure if not.
## @description otherwise just run it. (This is intended to
## @description be used by the start-*/stop-* scripts.)
## @audience private
## @stability evolving
## @replaceable yes
## @param user
## @param commandstring
## @return exitstatus
function hadoop_su
{
declare user=$1
shift
declare idret
if hadoop_privilege_check; then
id -u "${user}" >/dev/null 2>&1
idret=$?
if [[ ${idret} != 0 ]]; then
hadoop_error "ERROR: Refusing to run as root: ${user} account is not found. Aborting."
return 1
else
su -l "${user}" -- "$@"
fi
else
"$@"
fi
}
## @description Execute a command via su when running as root
## @description with extra support for commands that might
## @description legitimately start as root (e.g., datanode)
## @description (This is intended to
## @description be used by the start-*/stop-* scripts.)
## @audience private
## @stability evolving
## @replaceable no
## @param user
## @param commandstring
## @return exitstatus
function hadoop_uservar_su
{
## startup matrix:
#
# if $EUID != 0, then exec
# if $EUID =0 then
# if hdfs_subcmd_user is defined, call hadoop_su to exec
# if hdfs_subcmd_user is not defined, error
#
# For secure daemons, this means both the secure and insecure env vars need to be
# defined. e.g., HDFS_DATANODE_USER=root HDFS_DATANODE_SECURE_USER=hdfs
# This function will pick up the "normal" var, switch to that user, then
# execute the command which will then pick up the "secure" version.
#
declare program=$1
declare command=$2
shift 2
declare uprogram
declare ucommand
declare uvar
if hadoop_privilege_check; then
uvar=$(hadoop_get_verify_uservar "${program}" "${command}")
if [[ -n "${!uvar}" ]]; then
hadoop_su "${!uvar}" "$@"
else
hadoop_error "ERROR: Attempting to launch ${program} ${command} as root"
hadoop_error "ERROR: but there is no ${uvar} defined. Aborting launch."
return 1
fi
else
"$@"
fi
}
## @description Add a subcommand to the usage output ## @description Add a subcommand to the usage output
## @audience private ## @audience private
## @stability evolving ## @stability evolving
@ -343,6 +474,9 @@ function hadoop_bootstrap
# daemonization # daemonization
HADOOP_SUBCMD_SUPPORTDAEMONIZATION=false HADOOP_SUBCMD_SUPPORTDAEMONIZATION=false
# by default, we have not been self-re-execed
HADOOP_REEXECED_CMD=false
# shellcheck disable=SC2034 # shellcheck disable=SC2034
HADOOP_SUBCMD_SECURESERVICE=false HADOOP_SUBCMD_SECURESERVICE=false
@ -624,9 +758,10 @@ function hadoop_basic_init
fi fi
# if for some reason the shell doesn't have $USER defined # if for some reason the shell doesn't have $USER defined
# let's define it as 'hadoop' # (e.g., ssh'd in to execute a command)
# let's get the effective username and use that
USER=${USER:-$(id -nu)}
HADOOP_IDENT_STRING=${HADOOP_IDENT_STRING:-$USER} HADOOP_IDENT_STRING=${HADOOP_IDENT_STRING:-$USER}
HADOOP_IDENT_STRING=${HADOOP_IDENT_STRING:-hadoop}
HADOOP_LOG_DIR=${HADOOP_LOG_DIR:-"${HADOOP_HOME}/logs"} HADOOP_LOG_DIR=${HADOOP_LOG_DIR:-"${HADOOP_HOME}/logs"}
HADOOP_LOGFILE=${HADOOP_LOGFILE:-hadoop.log} HADOOP_LOGFILE=${HADOOP_LOGFILE:-hadoop.log}
HADOOP_LOGLEVEL=${HADOOP_LOGLEVEL:-INFO} HADOOP_LOGLEVEL=${HADOOP_LOGLEVEL:-INFO}
@ -1400,8 +1535,7 @@ function hadoop_verify_secure_prereq
# and you are using pfexec, you'll probably want to change # and you are using pfexec, you'll probably want to change
# this. # this.
# ${EUID} comes from the shell itself! if ! hadoop_privilege_check && [[ -z "${HADOOP_SECURE_COMMAND}" ]]; then
if [[ "${EUID}" -ne 0 ]] && [[ -z "${HADOOP_SECURE_COMMAND}" ]]; then
hadoop_error "ERROR: You must be a privileged user in order to run a secure service." hadoop_error "ERROR: You must be a privileged user in order to run a secure service."
exit 1 exit 1
else else
@ -1994,20 +2128,18 @@ function hadoop_secure_daemon_handler
esac esac
} }
## @description Verify that ${USER} is allowed to execute the ## @description Get the environment variable used to validate users
## @description given subcommand.
## @audience public ## @audience public
## @stability stable ## @stability stable
## @replaceable yes ## @replaceable yes
## @param subcommand ## @param subcommand
## @return will exit on failure conditions ## @return string
function hadoop_verify_user function hadoop_get_verify_uservar
{ {
declare program=$1 declare program=$1
declare command=$2 declare command=$2
declare uprogram declare uprogram
declare ucommand declare ucommand
declare uvar
if [[ -z "${BASH_VERSINFO[0]}" ]] \ if [[ -z "${BASH_VERSINFO[0]}" ]] \
|| [[ "${BASH_VERSINFO[0]}" -lt 4 ]]; then || [[ "${BASH_VERSINFO[0]}" -lt 4 ]]; then
@ -2018,7 +2150,25 @@ function hadoop_verify_user
ucommand=${command^^} ucommand=${command^^}
fi fi
uvar="${uprogram}_${ucommand}_USER" echo "${uprogram}_${ucommand}_USER"
}
## @description Verify that ${USER} is allowed to execute the
## @description given subcommand.
## @audience public
## @stability stable
## @replaceable yes
## @param command
## @param subcommand
## @return return 0 on success
## @return exit 1 on failure
function hadoop_verify_user
{
declare program=$1
declare command=$2
declare uvar
uvar=$(hadoop_get_verify_uservar "${program}" "${command}")
if [[ -n ${!uvar} ]]; then if [[ -n ${!uvar} ]]; then
if [[ ${!uvar} != "${USER}" ]]; then if [[ ${!uvar} != "${USER}" ]]; then
@ -2026,6 +2176,42 @@ function hadoop_verify_user
exit 1 exit 1
fi fi
fi fi
return 0
}
## @description Verify that ${USER} is allowed to execute the
## @description given subcommand.
## @audience public
## @stability stable
## @replaceable yes
## @param subcommand
## @return 1 on no re-exec needed
## @return 0 on need to re-exec
function hadoop_need_reexec
{
declare program=$1
declare command=$2
declare uvar
# we've already been re-execed, bail
if [[ "${HADOOP_REEXECED_CMD}" = true ]]; then
return 1
fi
# if we have privilege, and the _USER is defined, and _USER is
# set to someone who isn't us, then yes, we should re-exec.
# otherwise no, don't re-exec and let the system deal with it.
if hadoop_privilege_check; then
uvar=$(hadoop_get_verify_uservar "${program}" "${command}")
if [[ -n ${!uvar} ]]; then
if [[ ${!uvar} != "${USER}" ]]; then
return 0
fi
fi
fi
return 1
} }
## @description Add custom (program)_(command)_OPTS to HADOOP_OPTS. ## @description Add custom (program)_(command)_OPTS to HADOOP_OPTS.
@ -2228,6 +2414,15 @@ function hadoop_parse_args
shift shift
((HADOOP_PARSE_COUNTER=HADOOP_PARSE_COUNTER+2)) ((HADOOP_PARSE_COUNTER=HADOOP_PARSE_COUNTER+2))
;; ;;
--reexec)
shift
if [[ "${HADOOP_REEXECED_CMD}" = true ]]; then
hadoop_error "ERROR: re-exec fork bomb prevention: --reexec already called"
exit 1
fi
HADOOP_REEXECED_CMD=true
((HADOOP_PARSE_COUNTER=HADOOP_PARSE_COUNTER+1))
;;
--workers) --workers)
shift shift
# shellcheck disable=SC2034 # shellcheck disable=SC2034

View File

@ -15,10 +15,14 @@
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
# limitations under the License. # limitations under the License.
echo "This script is deprecated. Use start-dfs.sh and start-yarn.sh instead." ## @description catch the ctrl-c
## @audience private
## @stability evolving
## @replaceable no
function hadoop_abort_startall()
{
exit 1 exit 1
}
# let's locate libexec... # let's locate libexec...
if [[ -n "${HADOOP_HOME}" ]]; then if [[ -n "${HADOOP_HOME}" ]]; then
@ -38,6 +42,16 @@ else
echo "ERROR: Cannot execute ${HADOOP_LIBEXEC_DIR}/hadoop-config.sh." 2>&1 echo "ERROR: Cannot execute ${HADOOP_LIBEXEC_DIR}/hadoop-config.sh." 2>&1
exit 1 exit 1
fi fi
if ! hadoop_privilege_check; then
trap hadoop_abort_startall INT
hadoop_error "WARNING: Attempting to start all Apache Hadoop daemons as ${USER} in 10 seconds."
hadoop_error "WARNING: This is not a recommended production deployment configuration."
hadoop_error "WARNING: Use CTRL-C to abort."
sleep 10
trap - INT
fi
# start hdfs daemons if hdfs is present # start hdfs daemons if hdfs is present
if [[ -f "${HADOOP_HDFS_HOME}/sbin/start-dfs.sh" ]]; then if [[ -f "${HADOOP_HDFS_HOME}/sbin/start-dfs.sh" ]]; then
"${HADOOP_HDFS_HOME}/sbin/start-dfs.sh" --config "${HADOOP_CONF_DIR}" "${HADOOP_HDFS_HOME}/sbin/start-dfs.sh" --config "${HADOOP_CONF_DIR}"
@ -49,4 +63,3 @@ if [[ -f "${HADOOP_YARN_HOME}/sbin/start-yarn.sh" ]]; then
fi fi

View File

@ -15,12 +15,17 @@
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
# limitations under the License. # limitations under the License.
## @description catch the ctrl-c
## @audience private
## @stability evolving
## @replaceable no
function hadoop_abort_stopall()
{
exit 1
}
# Stop all hadoop daemons. Run this on master node. # Stop all hadoop daemons. Run this on master node.
echo "This script is deprecated. Use stop-dfs.sh and stop-yarn.sh instead."
exit 1
# let's locate libexec... # let's locate libexec...
if [[ -n "${HADOOP_HOME}" ]]; then if [[ -n "${HADOOP_HOME}" ]]; then
HADOOP_DEFAULT_LIBEXEC_DIR="${HADOOP_HOME}/libexec" HADOOP_DEFAULT_LIBEXEC_DIR="${HADOOP_HOME}/libexec"
@ -40,6 +45,14 @@ else
exit 1 exit 1
fi fi
if ! hadoop_privilege_check; then
trap hadoop_abort_stopall INT
hadoop_error "WARNING: Stopping all Apache Hadoop daemons as ${USER} in 10 seconds."
hadoop_error "WARNING: Use CTRL-C to abort."
sleep 10
trap - INT
fi
# stop hdfs daemons if hdfs is present # stop hdfs daemons if hdfs is present
if [[ -f "${HADOOP_HDFS_HOME}/sbin/stop-dfs.sh" ]]; then if [[ -f "${HADOOP_HDFS_HOME}/sbin/stop-dfs.sh" ]]; then
"${HADOOP_HDFS_HOME}/sbin/stop-dfs.sh" --config "${HADOOP_CONF_DIR}" "${HADOOP_HDFS_HOME}/sbin/stop-dfs.sh" --config "${HADOOP_CONF_DIR}"

View File

@ -103,6 +103,15 @@ In addition, daemons that run in an extra security mode also support `(command)_
Apache Hadoop provides a way to do a user check per-subcommand. While this method is easily circumvented and should not be considered a security-feature, it does provide a mechanism by which to prevent accidents. For example, setting `HDFS_NAMENODE_USER=hdfs` will make the `hdfs namenode` and `hdfs --daemon start namenode` commands verify that the user running the commands are the hdfs user by checking the `USER` environment variable. This also works for non-daemons. Setting `HADOOP_DISTCP_USER=jane` will verify that `USER` is set to `jane` before being allowed to execute the `hadoop distcp` command. Apache Hadoop provides a way to do a user check per-subcommand. While this method is easily circumvented and should not be considered a security-feature, it does provide a mechanism by which to prevent accidents. For example, setting `HDFS_NAMENODE_USER=hdfs` will make the `hdfs namenode` and `hdfs --daemon start namenode` commands verify that the user running the commands are the hdfs user by checking the `USER` environment variable. This also works for non-daemons. Setting `HADOOP_DISTCP_USER=jane` will verify that `USER` is set to `jane` before being allowed to execute the `hadoop distcp` command.
If a \_USER environment variable exists and commands are run with a privilege (e.g., as root; see hadoop_privilege_check in the API documentation), execution will switch to the specified user. For commands that support user account switching for security and therefore have a SECURE\_USER variable, the base \_USER variable needs to be the user that is expected to be used to switch to the SECURE\_USER account. For example:
```bash
HDFS_DATANODE_USER=root
HDFS_DATANODE_SECURE_USER=hdfs
```
Be aware that if the \-\-workers flag is used, the user switch happens *after* ssh is invoked. The multi-daemon start and stop commands in sbin will, however, switch (if appropriate) prior and will therefore use the keys of the specified \_USER.
## Developer and Advanced Administrator Environment ## Developer and Advanced Administrator Environment
### Shell Profiles ### Shell Profiles

View File

@ -16,9 +16,9 @@
setup() { setup() {
TMP="${BATS_TEST_DIRNAME}/../../../target/test-dir/bats.$$.${RANDOM}" RELTMP="${BATS_TEST_DIRNAME}/../../../target/test-dir/bats.$$.${RANDOM}"
mkdir -p ${TMP} mkdir -p ${RELTMP}
TMP=$(cd -P -- "${TMP}" >/dev/null && pwd -P) TMP=$(cd -P -- "${RELTMP}" >/dev/null && pwd -P)
export TMP export TMP
TESTBINDIR=$(cd -P -- "$(pwd)" >/dev/null && pwd -P) TESTBINDIR=$(cd -P -- "$(pwd)" >/dev/null && pwd -P)
HADOOP_LIBEXEC_DIR=${TESTBINDIR}/../../main/bin HADOOP_LIBEXEC_DIR=${TESTBINDIR}/../../main/bin

View File

@ -0,0 +1,65 @@
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements. See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
load hadoop-functions_test_helper
create_fake () {
mkdir ${TMP}/j
touch ${TMP}/j/k
ln -s j ${TMP}/l
}
@test "hadoop_abs (simple not exist)" {
run hadoop_abs fake
[ "${status}" -eq 1 ]
}
@test "hadoop_abs (simple dir)" {
create_fake
run hadoop_abs "${TMP}/j"
[ "${output}" = "${TMP}/j" ]
}
@test "hadoop_abs (simple file)" {
create_fake
run hadoop_abs "${TMP}/j/k"
[ "${output}" = "${TMP}/j/k" ]
}
@test "hadoop_abs (relative file1)" {
create_fake
run hadoop_abs "${TMP}/j/../j/k"
[ "${output}" = "${TMP}/j/k" ]
}
@test "hadoop_abs (relative file2)" {
create_fake
run hadoop_abs "${RELTMP}/j/../j/k"
[ "${output}" = "${TMP}/j/k" ]
}
@test "hadoop_abs (relative dir)" {
create_fake
fred=$(cd -P -- ".." >/dev/null && pwd -P)
run hadoop_abs ".."
[ "${output}" = "${fred}" ]
}
@test "hadoop_abs (symlink)" {
create_fake
run hadoop_abs "${TMP}/l"
[ "${output}" = "${TMP}/j" ]
}

View File

@ -0,0 +1,21 @@
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements. See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
load hadoop-functions_test_helper
@test "hadoop_get_verify_uservar" {
run hadoop_get_verify_uservar cool program
[ "${output}" = "COOL_PROGRAM_USER" ]
}

View File

@ -0,0 +1,26 @@
# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements. See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
load hadoop-functions_test_helper
@test "hadoop_privilege_check " {
if [ ${EUID} = 0 ]; then
result=0
else
result=1
fi
run hadoop_privilege_check
[ "${status}" = "${result}" ]
}

View File

@ -239,6 +239,9 @@ else
exit 1 exit 1
fi fi
# now that we have support code, let's abs MYNAME so we can use it later
MYNAME=$(hadoop_abs "${MYNAME}")
if [[ $# = 0 ]]; then if [[ $# = 0 ]]; then
hadoop_exit_with_usage 1 hadoop_exit_with_usage 1
fi fi
@ -246,6 +249,14 @@ fi
HADOOP_SUBCMD=$1 HADOOP_SUBCMD=$1
shift shift
if hadoop_need_reexec hdfs "${HADOOP_SUBCMD}"; then
hadoop_uservar_su hdfs "${HADOOP_SUBCMD}" \
"${MYNAME}" \
"--reexec" \
"${HADOOP_USER_PARAMS[@]}"
exit $?
fi
hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}" hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}"
HADOOP_SUBCMD_ARGS=("$@") HADOOP_SUBCMD_ARGS=("$@")

View File

@ -20,6 +20,21 @@
# Optinally upgrade or rollback dfs state. # Optinally upgrade or rollback dfs state.
# Run this on master node. # Run this on master node.
## startup matrix:
#
# if $EUID != 0, then exec
# if $EUID =0 then
# if hdfs_subcmd_user is defined, su to that user, exec
# if hdfs_subcmd_user is not defined, error
#
# For secure daemons, this means both the secure and insecure env vars need to be
# defined. e.g., HDFS_DATANODE_USER=root HADOOP_SECURE_DN_USER=hdfs
#
## @description usage info
## @audience private
## @stability evolving
## @replaceable no
function hadoop_usage function hadoop_usage
{ {
echo "Usage: start-dfs.sh [-upgrade|-rollback] [-clusterId]" echo "Usage: start-dfs.sh [-upgrade|-rollback] [-clusterId]"
@ -45,7 +60,6 @@ else
exit 1 exit 1
fi fi
# get arguments # get arguments
if [[ $# -ge 1 ]]; then if [[ $# -ge 1 ]]; then
startOpt="$1" startOpt="$1"
@ -77,32 +91,25 @@ if [[ -z "${NAMENODES}" ]]; then
fi fi
echo "Starting namenodes on [${NAMENODES}]" echo "Starting namenodes on [${NAMENODES}]"
hadoop_uservar_su hdfs namenode "${HADOOP_HDFS_HOME}/bin/hdfs" \
"${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${NAMENODES}" \ --hostnames "${NAMENODES}" \
--daemon start \ --daemon start \
namenode ${nameStartOpt} namenode ${nameStartOpt}
HADOOP_JUMBO_RETCOUNTER=$?
#--------------------------------------------------------- #---------------------------------------------------------
# datanodes (using default workers file) # datanodes (using default workers file)
if [[ -n "${HADOOP_SECURE_DN_USER}" ]] &&
[[ -z "${HADOOP_SECURE_COMMAND}" ]]; then
hadoop_error "ERROR: Attempting to start secure cluster, skipping datanodes. "
hadoop_error "ERROR: Run start-secure-dns.sh as root or configure "
hadoop_error "ERROR: \${HADOOP_SECURE_COMMAND} to complete startup."
else
echo "Starting datanodes" echo "Starting datanodes"
hadoop_uservar_su hdfs datanode "${HADOOP_HDFS_HOME}/bin/hdfs" \
"${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--daemon start \ --daemon start \
datanode ${dataStartOpt} datanode ${dataStartOpt}
fi (( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
#--------------------------------------------------------- #---------------------------------------------------------
# secondary namenodes (if any) # secondary namenodes (if any)
@ -113,8 +120,8 @@ if [[ -n "${SECONDARY_NAMENODES}" ]]; then
if [[ "${NAMENODES}" =~ , ]]; then if [[ "${NAMENODES}" =~ , ]]; then
hadoop_error "ERROR: Highly available NameNode is configured." hadoop_error "WARNING: Highly available NameNode is configured."
hadoop_error "ERROR: Skipping SecondaryNameNode." hadoop_error "WARNING: Skipping SecondaryNameNode."
else else
@ -124,12 +131,13 @@ if [[ -n "${SECONDARY_NAMENODES}" ]]; then
echo "Starting secondary namenodes [${SECONDARY_NAMENODES}]" echo "Starting secondary namenodes [${SECONDARY_NAMENODES}]"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs secondarynamenode "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${SECONDARY_NAMENODES}" \ --hostnames "${SECONDARY_NAMENODES}" \
--daemon start \ --daemon start \
secondarynamenode secondarynamenode
(( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
fi fi
fi fi
@ -143,12 +151,13 @@ case "${SHARED_EDITS_DIR}" in
JOURNAL_NODES=$(echo "${SHARED_EDITS_DIR}" | sed 's,qjournal://\([^/]*\)/.*,\1,g; s/;/ /g; s/:[0-9]*//g') JOURNAL_NODES=$(echo "${SHARED_EDITS_DIR}" | sed 's,qjournal://\([^/]*\)/.*,\1,g; s/;/ /g; s/:[0-9]*//g')
echo "Starting journal nodes [${JOURNAL_NODES}]" echo "Starting journal nodes [${JOURNAL_NODES}]"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs journalnode "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${JOURNAL_NODES}" \ --hostnames "${JOURNAL_NODES}" \
--daemon start \ --daemon start \
journalnode journalnode
(( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
;; ;;
esac esac
@ -158,12 +167,15 @@ AUTOHA_ENABLED=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey dfs.ha.automati
if [[ "${AUTOHA_ENABLED}" = "true" ]]; then if [[ "${AUTOHA_ENABLED}" = "true" ]]; then
echo "Starting ZK Failover Controllers on NN hosts [${NAMENODES}]" echo "Starting ZK Failover Controllers on NN hosts [${NAMENODES}]"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs zkfc "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${NAMENODES}" \ --hostnames "${NAMENODES}" \
--daemon start \ --daemon start \
zkfc zkfc
(( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
fi fi
exit ${HADOOP_JUMBO_RETCOUNTER}
# eof # eof

View File

@ -17,8 +17,12 @@
# Run as root to start secure datanodes in a security-enabled cluster. # Run as root to start secure datanodes in a security-enabled cluster.
## @description usage info
function hadoop_usage { ## @audience private
## @stability evolving
## @replaceable no
function hadoop_usage()
{
echo "Usage: start-secure-dns.sh" echo "Usage: start-secure-dns.sh"
} }
@ -42,12 +46,9 @@ else
exit 1 exit 1
fi fi
if [[ "${EUID}" -eq 0 ]] && [[ -n "${HADOOP_SECURE_DN_USER}" ]]; then echo "Starting datanodes"
exec "${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs datanode "${HADOOP_HDFS_HOME}/bin/hdfs" \
--config "${HADOOP_CONF_DIR}" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \
--daemon start \ --daemon start \
datanode datanode
else
echo hadoop_usage_and_exit 1
fi

View File

@ -19,9 +19,13 @@
# Stop hadoop dfs daemons. # Stop hadoop dfs daemons.
# Run this on master node. # Run this on master node.
## @description usage info
## @audience private
## @stability evolving
## @replaceable no
function hadoop_usage function hadoop_usage
{ {
echo "Usage: stop-dfs.sh [-upgrade|-rollback] [-clusterId]" echo "Usage: stop-dfs.sh"
} }
this="${BASH_SOURCE-$0}" this="${BASH_SOURCE-$0}"
@ -55,7 +59,7 @@ fi
echo "Stopping namenodes on [${NAMENODES}]" echo "Stopping namenodes on [${NAMENODES}]"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs namenode "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${NAMENODES}" \ --hostnames "${NAMENODES}" \
@ -65,21 +69,13 @@ echo "Stopping namenodes on [${NAMENODES}]"
#--------------------------------------------------------- #---------------------------------------------------------
# datanodes (using default workers file) # datanodes (using default workers file)
if [[ -n "${HADOOP_SECURE_DN_USER}" ]] &&
[[ -z "${HADOOP_SECURE_COMMAND}" ]]; then
echo "ERROR: Attempting to stop secure cluster, skipping datanodes. "
echo "Run stop-secure-dns.sh as root or configure "
echo "\${HADOOP_SECURE_COMMAND} to complete stop."
else
echo "Stopping datanodes" echo "Stopping datanodes"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs datanode "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--daemon stop \ --daemon stop \
datanode datanode
fi
#--------------------------------------------------------- #---------------------------------------------------------
# secondary namenodes (if any) # secondary namenodes (if any)
@ -93,7 +89,7 @@ fi
if [[ -n "${SECONDARY_NAMENODES}" ]]; then if [[ -n "${SECONDARY_NAMENODES}" ]]; then
echo "Stopping secondary namenodes [${SECONDARY_NAMENODES}]" echo "Stopping secondary namenodes [${SECONDARY_NAMENODES}]"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs secondarynamenode "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${SECONDARY_NAMENODES}" \ --hostnames "${SECONDARY_NAMENODES}" \
@ -111,7 +107,7 @@ case "${SHARED_EDITS_DIR}" in
JOURNAL_NODES=$(echo "${SHARED_EDITS_DIR}" | sed 's,qjournal://\([^/]*\)/.*,\1,g; s/;/ /g; s/:[0-9]*//g') JOURNAL_NODES=$(echo "${SHARED_EDITS_DIR}" | sed 's,qjournal://\([^/]*\)/.*,\1,g; s/;/ /g; s/:[0-9]*//g')
echo "Stopping journal nodes [${JOURNAL_NODES}]" echo "Stopping journal nodes [${JOURNAL_NODES}]"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs journalnode "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${JOURNAL_NODES}" \ --hostnames "${JOURNAL_NODES}" \
@ -126,7 +122,7 @@ AUTOHA_ENABLED=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey dfs.ha.automati
if [[ "${AUTOHA_ENABLED}" = "true" ]]; then if [[ "${AUTOHA_ENABLED}" = "true" ]]; then
echo "Stopping ZK Failover Controllers on NN hosts [${NAMENODES}]" echo "Stopping ZK Failover Controllers on NN hosts [${NAMENODES}]"
"${HADOOP_HDFS_HOME}/bin/hdfs" \ hadoop_uservar_su hdfs zkfc "${HADOOP_HDFS_HOME}/bin/hdfs" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--hostnames "${NAMENODES}" \ --hostnames "${NAMENODES}" \

View File

@ -17,8 +17,12 @@
# Run as root to stop secure datanodes in a security-enabled cluster. # Run as root to stop secure datanodes in a security-enabled cluster.
## @description usage info
function hadoop_usage { ## @audience private
## @stability evolving
## @replaceable no
function hadoop_usage()
{
echo "Usage: stop-secure-dns.sh" echo "Usage: stop-secure-dns.sh"
} }
@ -42,12 +46,8 @@ else
exit 1 exit 1
fi fi
if [[ "${EUID}" -eq 0 ]] && [[ -n "${HADOOP_SECURE_DN_USER}" ]]; then hadoop_uservar_su hdfs datanode "${HADOOP_HDFS_HOME}/bin/hdfs" \
exec "${HADOOP_HDFS_HOME}/bin/hdfs" \
--config "${HADOOP_CONF_DIR}" \
--workers \ --workers \
--config "${HADOOP_CONF_DIR}" \
--daemon stop \ --daemon stop \
datanode datanode
else
echo hadoop_usage_and_exit 1
fi

View File

@ -121,6 +121,8 @@ else
exit 1 exit 1
fi fi
# now that we have support code, let's abs MYNAME so we can use it later
MYNAME=$(hadoop_abs "${MYNAME}")
if [ $# = 0 ]; then if [ $# = 0 ]; then
hadoop_exit_with_usage 1 hadoop_exit_with_usage 1
@ -129,6 +131,14 @@ fi
HADOOP_SUBCMD=$1 HADOOP_SUBCMD=$1
shift shift
if hadoop_need_reexec mapred "${HADOOP_SUBCMD}"; then
hadoop_uservar_su mapred "${HADOOP_SUBCMD}" \
"${MYNAME}" \
"--reexec" \
"${HADOOP_USER_PARAMS[@]}"
exit $?
fi
hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}" hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}"
HADOOP_SUBCMD_ARGS=("$@") HADOOP_SUBCMD_ARGS=("$@")

View File

@ -15,14 +15,17 @@
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
# limitations under the License. # limitations under the License.
## @description usage info
MYNAME="${BASH_SOURCE-$0}" ## @audience private
## @stability evolving
## @replaceable no
function hadoop_usage function hadoop_usage
{ {
hadoop_generate_usage "${MYNAME}" false hadoop_generate_usage "${MYNAME}" false
} }
MYNAME="${BASH_SOURCE-$0}"
bin=$(cd -P -- "$(dirname -- "${MYNAME}")" >/dev/null && pwd -P) bin=$(cd -P -- "$(dirname -- "${MYNAME}")" >/dev/null && pwd -P)
# let's locate libexec... # let's locate libexec...
@ -42,14 +45,17 @@ else
exit 1 exit 1
fi fi
HADOOP_JUMBO_RETCOUNTER=0
# start resourceManager # start resourceManager
HARM=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.resourcemanager.ha.enabled 2>&-) HARM=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.resourcemanager.ha.enabled 2>&-)
if [[ ${HARM} = "false" ]]; then if [[ ${HARM} = "false" ]]; then
echo "Starting resourcemanager" echo "Starting resourcemanager"
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn resourcemanager "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--daemon start \ --daemon start \
resourcemanager resourcemanager
(( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
else else
logicals=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.resourcemanager.ha.rm-ids 2>&-) logicals=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.resourcemanager.ha.rm-ids 2>&-)
logicals=${logicals//,/ } logicals=${logicals//,/ }
@ -59,30 +65,35 @@ else
RMHOSTS="${RMHOSTS} ${rmhost}" RMHOSTS="${RMHOSTS} ${rmhost}"
done done
echo "Starting resourcemanagers on [${RMHOSTS}]" echo "Starting resourcemanagers on [${RMHOSTS}]"
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--daemon start \ --daemon start \
--workers \ --workers \
--hostnames "${RMHOSTS}" \ --hostnames "${RMHOSTS}" \
resourcemanager resourcemanager
(( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
fi fi
# start nodemanager # start nodemanager
echo "Starting nodemanagers" echo "Starting nodemanagers"
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn nodemanager "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--workers \ --workers \
--daemon start \ --daemon start \
nodemanager nodemanager
(( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
# start proxyserver # start proxyserver
PROXYSERVER=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.web-proxy.address 2>&- | cut -f1 -d:) PROXYSERVER=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.web-proxy.address 2>&- | cut -f1 -d:)
if [[ -n ${PROXYSERVER} ]]; then if [[ -n ${PROXYSERVER} ]]; then
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn proxyserver "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--workers \ --workers \
--hostnames "${PROXYSERVER}" \ --hostnames "${PROXYSERVER}" \
--daemon start \ --daemon start \
proxyserver proxyserver
(( HADOOP_JUMBO_RETCOUNTER=HADOOP_JUMBO_RETCOUNTER + $? ))
fi fi
exit ${HADOOP_JUMBO_RETCOUNTER}

View File

@ -15,14 +15,17 @@
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
# limitations under the License. # limitations under the License.
## @description usage info
MYNAME="${BASH_SOURCE-$0}" ## @audience private
## @stability evolving
## @replaceable no
function hadoop_usage function hadoop_usage
{ {
hadoop_generate_usage "${MYNAME}" false hadoop_generate_usage "${MYNAME}" false
} }
MYNAME="${BASH_SOURCE-$0}"
bin=$(cd -P -- "$(dirname -- "${MYNAME}")" >/dev/null && pwd -P) bin=$(cd -P -- "$(dirname -- "${MYNAME}")" >/dev/null && pwd -P)
# let's locate libexec... # let's locate libexec...
@ -44,7 +47,7 @@ fi
# stop nodemanager # stop nodemanager
echo "Stopping nodemanagers" echo "Stopping nodemanagers"
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn nodemanager "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--workers \ --workers \
--daemon stop \ --daemon stop \
@ -54,7 +57,7 @@ echo "Stopping nodemanagers"
HARM=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.resourcemanager.ha.enabled 2>&-) HARM=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.resourcemanager.ha.enabled 2>&-)
if [[ ${HARM} = "false" ]]; then if [[ ${HARM} = "false" ]]; then
echo "Stopping resourcemanager" echo "Stopping resourcemanager"
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn resourcemanager "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--daemon stop \ --daemon stop \
resourcemanager resourcemanager
@ -67,7 +70,7 @@ else
RMHOSTS="${RMHOSTS} ${rmhost}" RMHOSTS="${RMHOSTS} ${rmhost}"
done done
echo "Stopping resourcemanagers on [${RMHOSTS}]" echo "Stopping resourcemanagers on [${RMHOSTS}]"
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn resourcemanager "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--daemon stop \ --daemon stop \
--workers \ --workers \
@ -79,7 +82,7 @@ fi
PROXYSERVER=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.web-proxy.address 2>&- | cut -f1 -d:) PROXYSERVER=$("${HADOOP_HDFS_HOME}/bin/hdfs" getconf -confKey yarn.web-proxy.address 2>&- | cut -f1 -d:)
if [[ -n ${PROXYSERVER} ]]; then if [[ -n ${PROXYSERVER} ]]; then
echo "Stopping proxy server [${PROXYSERVER}]" echo "Stopping proxy server [${PROXYSERVER}]"
"${HADOOP_YARN_HOME}/bin/yarn" \ hadoop_uservar_su yarn proxyserver "${HADOOP_YARN_HOME}/bin/yarn" \
--config "${HADOOP_CONF_DIR}" \ --config "${HADOOP_CONF_DIR}" \
--workers \ --workers \
--hostnames "${PROXYSERVER}" \ --hostnames "${PROXYSERVER}" \

View File

@ -219,6 +219,9 @@ else
exit 1 exit 1
fi fi
# now that we have support code, let's abs MYNAME so we can use it later
MYNAME=$(hadoop_abs "${MYNAME}")
# if no args specified, show usage # if no args specified, show usage
if [[ $# = 0 ]]; then if [[ $# = 0 ]]; then
hadoop_exit_with_usage 1 hadoop_exit_with_usage 1
@ -228,6 +231,14 @@ fi
HADOOP_SUBCMD=$1 HADOOP_SUBCMD=$1
shift shift
if hadoop_need_reexec yarn "${HADOOP_SUBCMD}"; then
hadoop_uservar_su yarn "${HADOOP_SUBCMD}" \
"${MYNAME}" \
"--reexec" \
"${HADOOP_USER_PARAMS[@]}"
exit $?
fi
hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}" hadoop_verify_user "${HADOOP_SHELL_EXECNAME}" "${HADOOP_SUBCMD}"
HADOOP_SUBCMD_ARGS=("$@") HADOOP_SUBCMD_ARGS=("$@")