test_upgrade: factor out hardcoded assumptions

The following concepts are different between Concourse and a local
deployment:

- master hostname
- data directory locations and prefix
- gpinitsystem configuration location

Give them their own variables in preparation for supporting local runs.
Co-authored-by: NDavid Krieger <dkrieger@pivotal.io>

concourse/scripts/test_upgrade.bash

@@ -10,6 +10,26 @@ DEBUG_UPGRADE=${DEBUG_UPGRADE:-}
 
 ./ccp_src/scripts/setup_ssh_to_cluster.sh
 
+# The host running the master GPDB segment.
+MASTER_HOST=mdw
+
+# The GPHOME directories (containing greenplum_path.sh, the bin/ directory, and
+# so on) for the old and new clusters, respectively.
+OLD_GPHOME=/usr/local/greenplum-db-devel
+NEW_GPHOME=/usr/local/gpdb_master
+
+# The old and new clusters' master data directories.
+OLD_MASTER_DATA_DIRECTORY=/data/gpdata/master/gpseg-1
+NEW_MASTER_DATA_DIRECTORY=/data/gpdata/master-new/gpseg-1
+
+# The data directory prefix, assumed to be shared by all segments across the
+# cluster.
+DATADIR_PREFIX=/data/gpdata
+
+# The locations of the gpinitsystem configuration and hostfile.
+GPINITSYSTEM_CONFIG=gpinitsystem_config
+GPINITSYSTEM_HOSTFILE=segment_host_list
+
 DIRNAME=$(dirname "$0")
 
 cat << EOF
@@ -34,9 +54,9 @@ load_old_db_data() {
 
     echo 'Loading test database...'
 
-    scp "$dumpfile" mdw:/tmp/dump.sql.xz
-    ssh -n mdw '
-        source /usr/local/greenplum-db-devel/greenplum_path.sh
+    scp "$dumpfile" ${MASTER_HOST}:/tmp/dump.sql.xz
+    ssh -n ${MASTER_HOST} '
+        source '"${OLD_GPHOME}"'/greenplum_path.sh
         unxz < /tmp/dump.sql.xz | '"${psql_env}"' psql '"${psql_opts}"' -f - postgres
     '
 }
@@ -45,8 +65,8 @@ dump_cluster() {
     # Dump the entire cluster contents to file, using the new pg_dumpall.
     local dumpfile=$1
 
-    ssh -n mdw "
-        source /usr/local/gpdb_master/greenplum_path.sh
+    ssh -n ${MASTER_HOST} "
+        source ${NEW_GPHOME}/greenplum_path.sh
         pg_dumpall -f '$dumpfile'
     "
 }
@@ -60,10 +80,10 @@ extract_gpdb_tarball() {
     # the naming convention.
     scp ${tarball_dir}/*.tar.gz $node_hostname:/tmp/gpdb_binary.tar.gz
     ssh -ttn $node_hostname "sudo bash -c \"\
-      mkdir -p /usr/local/gpdb_master; \
-      tar -xf /tmp/gpdb_binary.tar.gz -C /usr/local/gpdb_master; \
-      chown -R gpadmin:gpadmin /usr/local/gpdb_master; \
-      sed -ie 's|^GPHOME=.*|GPHOME=/usr/local/gpdb_master|' /usr/local/gpdb_master/greenplum_path.sh ; \
+      mkdir -p ${NEW_GPHOME}; \
+      tar -xf /tmp/gpdb_binary.tar.gz -C ${NEW_GPHOME}; \
+      chown -R gpadmin:gpadmin ${NEW_GPHOME}; \
+      sed -ie 's|^GPHOME=.*|GPHOME=${NEW_GPHOME}|' ${NEW_GPHOME}/greenplum_path.sh ; \
     \""
 }
 
@@ -73,7 +93,7 @@ create_new_datadir() {
     # Create a -new directory for every data directory that already exists.
     # This is what we'll be init'ing the new database into.
     ssh -ttn "$node_hostname" 'sudo bash -c '\''
-        for dir in $(find /data/gpdata/* -maxdepth 0 -type d); do
+        for dir in $(find '"${DATADIR_PREFIX}"'/* -maxdepth 0 -type d); do
             newdir="${dir}-new"
 
             mkdir -p "$newdir"
@@ -88,17 +108,17 @@ gpinitsystem_for_upgrade() {
     # FIXME: the checksum/string settings below need to be pulled from the
     # previous database, not hardcoded. And long-term, we need to decide how
     # Greenplum clusters should be upgraded when GUC settings' defaults change.
-    ssh -ttn mdw '
-        source /usr/local/greenplum-db-devel/greenplum_path.sh
-        gpstop -a -d /data/gpdata/master/gpseg-1
+    ssh -ttn ${MASTER_HOST} '
+        source '"${OLD_GPHOME}"'/greenplum_path.sh
+        gpstop -a -d '"${OLD_MASTER_DATA_DIRECTORY}"'
 
-        source /usr/local/gpdb_master/greenplum_path.sh
-        sed -e '\''s|\(/data/gpdata/\w\+\)|\1-new|g'\'' gpinitsystem_config > gpinitsystem_config_new
+        source '"${NEW_GPHOME}"'/greenplum_path.sh
+        sed -e '\''s|\('"${DATADIR_PREFIX}"'/\w\+\)|\1-new|g'\'' '"${GPINITSYSTEM_CONFIG}"' > gpinitsystem_config_new
         # echo "HEAP_CHECKSUM=off" >> gpinitsystem_config_new
         # echo "standard_conforming_strings = off" >> upgrade_addopts
         # echo "escape_string_warning = off" >> upgrade_addopts
-        gpinitsystem -a -c ~gpadmin/gpinitsystem_config_new -h ~gpadmin/segment_host_list # -p ~gpadmin/upgrade_addopts
-        gpstop -a -d /data/gpdata/master-new/gpseg-1
+        gpinitsystem -a -c gpinitsystem_config_new -h '"${GPINITSYSTEM_HOSTFILE}"' # -p ~gpadmin/upgrade_addopts
+        gpstop -a -d '"${NEW_MASTER_DATA_DIRECTORY}"'
     '
 }
 
@@ -119,18 +139,18 @@ run_upgrade() {
     fi
 
     ssh -ttn "$hostname" '
-        source /usr/local/gpdb_master/greenplum_path.sh
+        source '"${NEW_GPHOME}"'/greenplum_path.sh
         time pg_upgrade '"${upgrade_opts}"' '"$*"' \
-            -b /usr/local/greenplum-db-devel/bin/ -B /usr/local/gpdb_master/bin/ \
+            -b '"${OLD_GPHOME}"'/bin/ -B '"${NEW_GPHOME}"'/bin/ \
             -d '"$datadir"' \
-            -D '"$(sed -e 's|\(/data/gpdata/\w\+\)|\1-new|g' <<< "$datadir")"
+            -D '"$(sed -e 's|\('"${DATADIR_PREFIX}"'/\w\+\)|\1-new|g' <<< "$datadir")"
 }
 
 dump_old_master_query() {
     # Prints the rows generated by the given SQL query to stdout. The query is
     # run on the old master, pre-upgrade.
-    ssh -n mdw '
-        source /usr/local/greenplum-db-devel/greenplum_path.sh
+    ssh -n ${MASTER_HOST} '
+        source '"${OLD_GPHOME}"'/greenplum_path.sh
         psql postgres --quiet --no-align --tuples-only -F"'$'\t''" -c "'$1'"
     '
 }
@@ -150,10 +170,10 @@ get_segment_datadirs() {
 }
 
 start_upgraded_cluster() {
-    ssh -n mdw '
-        source /usr/local/gpdb_master/greenplum_path.sh
-        MASTER_DATA_DIRECTORY=/data/gpdata/master-new/gpseg-1 gpstart -a -v
-    '
+    ssh -n ${MASTER_HOST} "
+        source ${NEW_GPHOME}/greenplum_path.sh
+        MASTER_DATA_DIRECTORY='${NEW_MASTER_DATA_DIRECTORY}' gpstart -a -v
+    "
 }
 
 apply_sql_fixups() {
@@ -172,8 +192,8 @@ apply_sql_fixups() {
 
     # FIXME: we need a generic way for gp_upgrade to figure out which SQL fixup
     # files need to be applied to the cluster before it is used.
-    ssh -n mdw '
-        source /usr/local/gpdb_master/greenplum_path.sh
+    ssh -n ${MASTER_HOST} '
+        source '"${NEW_GPHOME}"'/greenplum_path.sh
         if [ -f reindex_all.sql ]; then
             '"${psql_env}"' psql '"${psql_opts}"' -f reindex_all.sql template1
         fi
@@ -186,9 +206,9 @@ compare_dumps() {
     local old_dump=$1
     local new_dump=$2
 
-    scp "$DIRNAME/dumpsort.gawk" mdw:~
+    scp "$DIRNAME/dumpsort.gawk" ${MASTER_HOST}:~
 
-    ssh -n mdw "
+    ssh -n ${MASTER_HOST} "
         diff -U3 --speed-large-files --ignore-space-change \
             <(gawk -f ~/dumpsort.gawk < '$old_dump') \
             <(gawk -f ~/dumpsort.gawk < '$new_dump')
@@ -235,14 +255,17 @@ get_segment_datadirs > /tmp/segment_datadirs.txt
 gpinitsystem_for_upgrade
 
 # TODO: we need to switch the mode argument according to GPDB version
-echo "Upgrading master at mdw..."
-run_upgrade mdw "/data/gpdata/master/gpseg-1" --mode=dispatcher
+echo "Upgrading master at ${MASTER_HOST}..."
+run_upgrade ${MASTER_HOST} "${OLD_MASTER_DATA_DIRECTORY}" --mode=dispatcher
 
 while read -u30 hostname datadir; do
     echo "Upgrading segment at '$hostname' ($datadir)..."
 
-    newdatadir=$(sed -e 's|\(/data/gpdata/\w\+\)|\1-new|g' <<< "$datadir")
-    ssh -n mdw rsync -r --delete /data/gpdata/master-new/gpseg-1/ "$hostname:$newdatadir" \
+    newdatadir=$(sed -e 's|\('"${DATADIR_PREFIX}"'/\w\+\)|\1-new|g' <<< "$datadir")
+
+    # NOTE: the trailing slash on the rsync source directory is important! It
+    # means to transfer the directory's contents and not the directory itself.
+    ssh -n ${MASTER_HOST} rsync -r --delete "${NEW_MASTER_DATA_DIRECTORY}/" "$hostname:$newdatadir" \
         --exclude /postgresql.conf \
         --exclude /pg_hba.conf \
         --exclude /postmaster.opts \