DevOps-Bash-tools/bigdata/impala_shell.sh

#!/usr/bin/env bash
#  vim:ts=4:sts=4:sw=4:et
#
#  Author: Hari Sekhon
#  Date: 2019-12-10 11:33:52 +0000 (Tue, 10 Dec 2019)
#
#  https://github.com/HariSekhon/DevOps-Bash-tools
#
#  License: see accompanying Hari Sekhon LICENSE file
#
#  If you're using my code you're welcome to connect with me on LinkedIn and optionally send me feedback to help steer this or other code I publish
#
#  https://www.linkedin.com/in/HariSekhon
#

set -euo pipefail
[ -n "${DEBUG:-}" ] && set -x
srcdir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"

# shellcheck disable=SC1090,SC1091
. "$srcdir/lib/utils.sh"

# shellcheck disable=SC2034,SC2154
usage_description="
Script to more easily connect to Impala without having to find an impalad and repeatedly specify options like -k for kerberos


Tested on Impala 2.7.0, 2.12.0 on CDH 5.10, 5.16 with Kerberos and SSL


If using dedicated coordinators then consider setting IMPALA_HOST to one of those explicitly, see

   https://docs.cloudera.com/documentation/enterprise/5-16-x/topics/impala_dedicated_coordinator.html


See also:

  find_active_impalad.py - https://github.com/HariSekhon/DevOps-Python-tools

  HAProxy Configs for Impala and many other technologies - https://github.com/HariSekhon/HAProxy-configs


If you get an error such as:

Error connecting: TTransportException, TSocket read 0 bytes

then check if you need to add --ssl to the command line (or export IMPALA_SSL=1 to do this automatically, eg. put in .bashrc or similar)


Useful options for scripting:

  -q --query
  -B --delimited
  --output_delimiter=\\t   # default
  --quiet


See adjacent impala_*.sh scripts for slightly better versions of these quick command line examples, including better escaping


Examples:


# List all databases:

  ./impala_shell.sh -Bq 'show databases' | awk '{print \$1}'


# List all tables in all databases:

  ./impala_shell.sh -Bq 'show databases' | while read db rest; do ./impala_shell.sh -Bq \"use \$db; show tables\" | sed \"s/^/\$db./\"; done


# Row counts for all tables in all databases:

  ./impala_shell.sh --quiet -Bq 'show databases' | while read db rest; do ./impala_shell.sh --quiet -Bq \"use \$db; show tables\" | while read table; do printf \"%s\\t\" \"\$db.\$table\"; ./impala_shell.sh --quiet -Bq \"use \$db; SELECT COUNT(*) FROM \$table\"; done; done > row_counts.tsv
"

# used by usage() in lib/utils.sh
# shellcheck disable=SC2034
usage_args="[<impala_shell_options>]"

help_usage "$@"


opts="${IMPALA_OPTS:-}"

core_site_xml="${HADOOP_CORE_SITE_XML:-/etc/hadoop/conf/core-site.xml}"

#if ! [ -f "$core_site_xml" ]; then
#    echo "File not found: $core_site_xml. Did you run this on a Hadoop node?" >&2
#    exit 1
#fi

if [ -n "${IMPALA_KERBEROS:-}" ] ||
   grep -A 1 hadoop.security.authentication "$core_site_xml" 2>/dev/null | grep -q kerberos; then
    opts="$opts -k"
fi

if [ -n "${IMPALA_SSL:-}" ]; then
    opts="$opts --ssl"
fi

topology_map="${HADOOP_TOPOLOGY_MAP:-/etc/hadoop/conf/topology.map}"

if [ -n "${IMPALA_HOST:-}" ]; then
    impalad="$IMPALA_HOST"
elif [ -f "$topology_map" ]; then
    #echo "picking random impala from hadoop topology map" >&2
    # nodes in the topology map that aren't masters, namenodes, controlnodes etc probably have impalad running on them, so pick one at random to connect to
    # or alternatively use HAProxy config for load balanced impala clusters - see https://github.com/HariSekhon/HAProxy-configs
    impalad="$(
        awk -F'"' '/<node name="[A-Za-z]/{print $2}' "$topology_map" |
        grep -Ev '^[^.]*(name|master|control)' |
        shuf -n 1
    )"
else
    impalad="$(hostname -f)"
    #echo "IMPALA_HOST not set and topology map '$topology_map' not found, defaulting to local host $impalad"
fi

# split opts
# shellcheck disable=SC2086
exec impala-shell $opts -i "$impalad" "$@"