feat: automatically skip large file systems

config/uac.conf

@@ -24,6 +24,14 @@ exclude_name_pattern: []
 # running kernel.
 exclude_file_system: [9p, afs, autofs, cifs, davfs, fuse, kernfs, nfs, nfs4, rpc_pipefs, smbfs, sysfs]
 
+# Exclude any file systems with more than this much space *used* (root file
+# system will never be excluded).
+# File systems will be excluded from find, file, hash and stat collectors.
+# Valid units are b|c (bytes), k (kilobytes), and M,G,T for mega-, giga-,
+# and tera-bytes respectively. Value is assumed to be bytes if no unit given.
+# Set to zero to disable.
+exclude_mount_size: 0
+
 # Hash algorithms
 # Used by the hash collector.
 # Accepted values: md5, sha1 and sha256

lib/convert_size_to_kb.sh

@@ -0,0 +1,58 @@
+#!/bin/sh
+# SPDX-License-Identifier: Apache-2.0
+# shellcheck disable=SC2006,SC2005,SC2003
+
+# Take a string like "10T" and convert to a number representing kilobytes.
+# Valid units are b|c (bytes), k (kilobytes), and M,G,T for mega-, giga-,
+# and tera-bytes respectively. Null unit specifier means bytes.
+# Arguments:
+#   string like "10T"
+# Returns:
+#   integer number of kilobytes
+#   returns zero on any errors
+_convert_size_to_kb()
+{
+  __cs_arg="${1:-0}"
+
+  # return error if format not recognized
+  if ! echo "${__cs_arg}" | grep -q -E '^[0-9]*[bckMGT]?$'; then
+    echo 0
+    return 1
+  fi
+
+  __cs_number="`echo "${__cs_arg}" | sed 's/[^0-9]//g'`"
+
+  # short circuit for lunatics who enter things like "0T"
+  if [ "${__cs_number}" -eq 0 ]; then
+      echo 0
+      return 0
+  fi
+
+  __cs_unit="`echo "${__cs_arg}" | tr -d 0-9`"
+
+  # if the unit is "c" or empty, convert the unit to "b"
+  [ -n "${__cs_unit}" ] || __cs_unit='b'
+  [ "${__cs_unit}" = "c" ] && __cs_unit='b'
+
+  # If they enter a number of bytes less than 1024, expr will round down
+  # to zero kilobytes, which is surely not what the user intended.
+  # Set exclude size to 1kb.
+  [ "${__cs_unit}" = "b" ] && [ "${__cs_number}" -lt 1024 ] && __cs_number=1024
+
+  case "${__cs_unit}" in
+    b) echo "`expr "${__cs_number}" / 1024`"
+	 ;;
+    k) echo "${__cs_number}"
+       ;;
+    M) echo "`expr "${__cs_number}" \* 1024`"
+       ;;
+    G) echo "`expr "${__cs_number}" \* 1024 \* 1024`"
+       ;;
+    T) echo "`expr "${__cs_number}" \* 1024 \* 1024 \* 1024`"
+       ;;
+    *) echo 0
+       return 1
+       ;;
+  esac
+  return 0
+}

lib/get_large_file_systems.sh

@@ -0,0 +1,24 @@
+#!/bin/sh
+# SPDX-License-Identifier: Apache-2.0
+# 
+
+# Returns a pipe delimited list of file systems that are using more than
+# ${__UAC_CONF_EXCLUDE_MOUNT_SIZE} kilobytes. Root file system will always
+# be ignored.
+#
+# Arguments:
+#   none
+# Returns:
+#   pipe delimited string, may be empty
+_get_large_file_systems()
+{
+  [ "${__UAC_CONF_EXCLUDE_MOUNT_SIZE}" -gt 0 ] || return 0
+
+  # shellcheck disable=SC2162
+  df | while read __gl_dev __gl_size __gl_used __gl_avail __gl_pct __gl_mtpt
+  do
+    echo "${__gl_used}" | grep -q -E '^[0-9]*$' || continue
+    [ "${__gl_used}" -gt "${__UAC_CONF_EXCLUDE_MOUNT_SIZE}" ] && [ "${__gl_mtpt}" != '/' ] && echo "${__gl_mtpt}"
+  done | tr \\n \| | sed 's/|$//'
+  return 0
+}

lib/load_config_file.sh

@@ -19,6 +19,7 @@ _load_config_file()
   __UAC_CONF_EXCLUDE_PATH_PATTERN="${__UAC_CONF_EXCLUDE_PATH_PATTERN:-}"
   __UAC_CONF_EXCLUDE_NAME_PATTERN="${__UAC_CONF_EXCLUDE_NAME_PATTERN:-}"
   __UAC_CONF_EXCLUDE_FILE_SYSTEM="${__UAC_CONF_EXCLUDE_FILE_SYSTEM:-9p|afs|autofs|cifs|davfs|fuse|kernfs|nfs|nfs4|rpc_pipefs|smbfs|sysfs}"
+  __UAC_CONF_EXCLUDE_MOUNT_SIZE="${__UAC_CONF_EXCLUDE_MOUNT_SIZE:-0}"
   __UAC_CONF_HASH_ALGORITHM="${__UAC_CONF_HASH_ALGORITHM:-md5|sha1}"
   __UAC_CONF_MAX_DEPTH="${__UAC_CONF_MAX_DEPTH:-0}"
   __UAC_CONF_ENABLE_FIND_MTIME="${__UAC_CONF_ENABLE_FIND_MTIME:-true}"
@@ -63,6 +64,13 @@ _load_config_file()
     __UAC_CONF_EXCLUDE_FILE_SYSTEM=`echo "${__lc_value}" | _array_to_psv 2>/dev/null`
   fi
 
+  # load exclude_mount_size option
+  if echo "${__lc_config_file_content}" | grep -q "exclude_mount_size:"; then
+    __lc_value=`echo "${__lc_config_file_content}" | sed -n -e 's|exclude_mount_size\: *\(.*\)|\1|p'`
+    __UAC_CONF_EXCLUDE_MOUNT_SIZE=`_convert_size_to_kb "${__lc_value}"`
+  fi
+
+
   # load hash_algorithm option
   if echo "${__lc_config_file_content}" | grep -q "hash_algorithm:"; then
     __lc_value=`echo "${__lc_config_file_content}" | sed -n -e 's|hash_algorithm\: *\(.*\)|\1|p'`
@@ -119,4 +127,4 @@ _load_config_file()
     __UAC_CONF_ENABLE_FIND_CTIME="${__lc_value}"    
   fi
 
-}
+}

lib/load_libraries.sh

@@ -9,6 +9,7 @@
 . "${__UAC_DIR}/lib/build_find_command.sh"
 . "${__UAC_DIR}/lib/command_collector.sh"
 . "${__UAC_DIR}/lib/command_exists.sh"
+. "${__UAC_DIR}/lib/convert_size_to_kb.sh"
 . "${__UAC_DIR}/lib/copy_data.sh"
 . "${__UAC_DIR}/lib/create_acquisition_log.sh"
 . "${__UAC_DIR}/lib/error_msg.sh"
@@ -22,6 +23,7 @@
 . "${__UAC_DIR}/lib/get_days_since_date_until_now.sh"
 . "${__UAC_DIR}/lib/get_epoch_date.sh"
 . "${__UAC_DIR}/lib/get_hostname.sh"
+. "${__UAC_DIR}/lib/get_large_file_systems.sh"
 . "${__UAC_DIR}/lib/get_mount_point_by_file_system.sh"
 . "${__UAC_DIR}/lib/get_operating_system.sh"
 . "${__UAC_DIR}/lib/get_profile_by_name.sh"

uac

@@ -355,7 +355,6 @@ _log_msg INF "PATH: ${PATH}"
 _log_msg INF "Loading configuration file(s)"
 _log_msg INF "Exclude path pattern: ${__UAC_CONF_EXCLUDE_PATH_PATTERN}"
 _log_msg INF "Exclude name pattern: ${__UAC_CONF_EXCLUDE_NAME_PATTERN}"
-_log_msg INF "Exclude file system: ${__UAC_CONF_EXCLUDE_FILE_SYSTEM}"
 _log_msg INF "Hash algorithm: ${__UAC_CONF_HASH_ALGORITHM}"
 _log_msg INF "Max depth: ${__UAC_CONF_MAX_DEPTH}"
 _log_msg INF "Enable find mtime: ${__UAC_CONF_ENABLE_FIND_MTIME}"
@@ -382,6 +381,12 @@ if [ -n "${__UAC_CONF_EXCLUDE_FILE_SYSTEM}" ]; then
   __UAC_EXCLUDE_MOUNT_POINTS=`_get_mount_point_by_file_system "${__UAC_CONF_EXCLUDE_FILE_SYSTEM}" "${__UAC_OPERATING_SYSTEM}"`
 fi
 
+# add in file systems excluded by size
+if [ "${__UAC_CONF_EXCLUDE_MOUNT_SIZE}" -gt 0 ]; then
+  __ua_large_mounts=`_get_large_file_systems`
+  __UAC_EXCLUDE_MOUNT_POINTS=`echo "${__UAC_EXCLUDE_MOUNT_POINTS}|${__ua_large_mounts}" | sed 's/^|//; s/|$//'`
+fi
+
 # check whether temp data dir's file system supports symlink creation.
 if ln -s "${__UAC_DIR}/uac" "${__UAC_TEMP_DATA_DIR}/uac-symlink-support-test.tmp" >/dev/null; then
   __UAC_TEMP_DATA_DIR_NO_SYMLINK_SUPPORT=false
@@ -436,6 +441,7 @@ if ${__UAC_TEMP_DATA_DIR_NO_SYMLINK_SUPPORT}; then
 else
   _log_msg INF "Temp directory symlink creation support: true"
 fi
+_log_msg INF "Exclude file system size: ${__UAC_CONF_EXCLUDE_MOUNT_SIZE} kb"
 _log_msg INF "Exclude mount points: ${__UAC_EXCLUDE_MOUNT_POINTS}"
 
 _verbose_msg "Building artifact list..."