Set parallel merges to a value different from parallel loads.

[remoteglot-book] / build-book.sh

diff --git a/build-book.sh b/build-book.sh
index 01c51c7da920e5fb606b2e637f80ec06392b082e..60d5710ed481eda49ee4e372bbf778778688e1dd 100755 (executable)
--- a/build-book.sh
+++ b/build-book.sh
@@ -2,28 +2,30 @@
 
 set -e
 
-rm -f part-*.bin part-*.mtbl part-*.mtbl.part???? open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
+export SHARDS=40
+export PARALLEL_LOADS=15  # Reduce if you have problems with OOM
+export PARALLEL_MERGES=40
 
-for FILE in $@; do
+rm -f pgnnames.txt part-*.bin part-*.mtbl part-*.mtbl.part???? open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
+
+PGNNUM=0
+for FILE in "$@"; do
        date | tr -d "\n"
        echo "  $FILE"
-       ./parallel-parse-pgn.sh $FILE
+       ./parallel-parse-pgn.sh "$FILE" "$PGNNUM"
+       echo "$FILE" >> pgnnames.txt
+       PGNNUM=$(( PGNNUM + 1 ))
 done
 date 
 
-for FILE in part-*.bin; do
-       ( ./binloader $FILE ${FILE/bin/mtbl} 40 ) &
-done
-wait
+echo "Loading..."
+parallel -i -j ${PARALLEL_LOADS} bash -c 'FILE="{}"; set -x; ./binloader "$FILE" "${FILE/bin/mtbl}" $SHARDS' -- part-*.bin
 
 rm -f part-*.bin
 
-for X in $( seq 0 39 ); do
-       ( ./binmerger part-*.mtbl.part$( printf %04d $X ) open.mtbl.part$( printf %04d $X ).new ) &
-done
-wait
+parallel -i -j ${PARALLEL_MERGES} bash -c 'X={}; set -x; ./binmerger part-*.mtbl.part$( printf %04d $X ) open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) )
 
-for X in $( seq 0 39 ); do
-       mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X)
+for X in $( seq 0 $(( SHARDS - 1 )) ); do
+       mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X )
 done
 rm -f part-*.mtbl.part????