chore: fix cr issues & update results

Author: shuiyisong

greptimedb/count.sh

@@ -2,6 +2,6 @@
 
 curl -s --fail http://localhost:4000/v1/sql \
     -H 'Content-Type: application/x-www-form-urlencoded' \
-    -d "sql=select count(1) as cnt from jsontable"  \
+    -d "sql=select count(*) as cnt from bluesky"  \
     -d "format=json" \
     | grep -o "cnt\":[0-9]*" | sed 's/cnt\"://g'

greptimedb/data_size.sh

@@ -2,6 +2,6 @@
 
 curl -s --fail http://localhost:4000/v1/sql \
     -H 'Content-Type: application/x-www-form-urlencoded' \
-    -d "sql=SELECT sum(r.sst_size) as data_size FROM information_schema.REGION_STATISTICS r LEFT JOIN information_schema.TABLES t on r.table_id = t.table_id WHERE t.table_name = 'jsontable'"  \
+    -d "sql=SELECT sum(r.sst_size) as data_size FROM information_schema.REGION_STATISTICS r LEFT JOIN information_schema.TABLES t on r.table_id = t.table_id WHERE t.table_name = 'bluesky'"  \
     -d "format=json"  \
     | grep -o "data_size\":[0-9]*" | sed 's/data_size\"://g'

greptimedb/index_size.sh

@@ -2,6 +2,6 @@
 
 curl -s --fail http://localhost:4000/v1/sql \
     -H 'Content-Type: application/x-www-form-urlencoded' \
-    -d "sql=SELECT sum(r.index_size) as index_size FROM information_schema.REGION_STATISTICS r LEFT JOIN information_schema.TABLES t on r.table_id = t.table_id WHERE t.table_name = 'jsontable'"  \
+    -d "sql=SELECT sum(r.index_size) as index_size FROM information_schema.REGION_STATISTICS r LEFT JOIN information_schema.TABLES t on r.table_id = t.table_id WHERE t.table_name = 'bluesky'"  \
     -d "format=json"  \
     | grep -o "index_size\":[0-9]*" | sed 's/index_size\"://g'

greptimedb/load_data.sh

@@ -21,10 +21,11 @@ counter=0
 for file in $(ls *.json.gz | head -n $MAX_FILES); do
     echo "Processing file: $file"
 
-    curl "http://localhost:4000/v1/events/logs?table=jsontable&pipeline_name=jsonbench&ignore_errors=true" \
+    curl "http://localhost:4000/v1/events/logs?table=bluesky&pipeline_name=jsonbench&ignore_errors=true" \
          -H "Content-Type: application/x-ndjson" \
          -H "Content-Encoding: gzip" \
          --data-binary @$file
+    echo ""
 
     first_attempt=$?
     if [[ $first_attempt -eq 0 ]]; then
@@ -41,6 +42,7 @@ done
 
 curl -XPOST -H 'Content-Type: application/x-www-form-urlencoded' \
           http://localhost:4000/v1/sql \
-          -d "sql=admin flush_table('jsontable')"
+          -d "sql=admin flush_table('bluesky')" \
+          -d "format=json"
 
-echo "Loaded $MAX_FILES data files from $DATA_DIRECTORY to greptimedb."
+echo -e "\nLoaded $MAX_FILES data files from $DATA_DIRECTORY to GreptimeDB."

greptimedb/main.sh

@@ -40,7 +40,6 @@ benchmark() {
 
     ./start.sh
     ./load_data.sh "$DATA_DIRECTORY" "$size" "$SUCCESS_LOG" "$ERROR_LOG"
-    sleep 1 
     ./total_size.sh | tee "${OUTPUT_PREFIX}_bluesky_${size}m.total_size"
     ./data_size.sh | tee "${OUTPUT_PREFIX}_bluesky_${size}m.data_size"
     ./index_size.sh | tee "${OUTPUT_PREFIX}_bluesky_${size}m.index_size"

greptimedb/queries.sql

@@ -1,5 +1,5 @@
-SELECT commit_collection AS event, count(1) AS cnt FROM jsontable GROUP BY event ORDER BY cnt DESC;
-SELECT commit_collection AS event, count(1) AS cnt, count(DISTINCT did) AS users FROM jsontable WHERE kind = 'commit' AND commit_operation = 'create' GROUP BY event ORDER BY cnt DESC;
-SELECT commit_collection AS event, date_part('hour', time_us) AS hour_of_day, count(1) AS cnt FROM jsontable WHERE kind = 'commit' AND commit_operation = 'create' AND commit_collection IN('app.bsky.feed.post', 'app.bsky.feed.repost', 'app.bsky.feed.like') GROUP BY event, hour_of_day ORDER BY hour_of_day, event;
-SELECT did AS user_id, min(time_us) AS first_post_ts FROM jsontable WHERE kind = 'commit' AND commit_operation = 'create' AND commit_collection = 'app.bsky.feed.post' GROUP BY user_id ORDER BY first_post_ts ASC LIMIT 3;
-SELECT did AS user_id, date_part('millisecond',(max(time_us) - min(time_us))) AS activity_span FROM jsontable WHERE kind = 'commit' AND commit_operation = 'create' AND commit_collection = 'app.bsky.feed.post' GROUP BY user_id ORDER BY activity_span DESC LIMIT 3;
+SELECT commit_collection AS event, count(1) AS cnt FROM bluesky GROUP BY event ORDER BY cnt DESC;
+SELECT commit_collection AS event, count(1) AS cnt, count(DISTINCT did) AS users FROM bluesky WHERE kind = 'commit' AND commit_operation = 'create' GROUP BY event ORDER BY cnt DESC;
+SELECT commit_collection AS event, date_part('hour', time_us) AS hour_of_day, count(1) AS cnt FROM bluesky WHERE kind = 'commit' AND commit_operation = 'create' AND commit_collection IN('app.bsky.feed.post', 'app.bsky.feed.repost', 'app.bsky.feed.like') GROUP BY event, hour_of_day ORDER BY hour_of_day, event;
+SELECT did AS user_id, min(time_us) AS first_post_ts FROM bluesky WHERE kind = 'commit' AND commit_operation = 'create' AND commit_collection = 'app.bsky.feed.post' GROUP BY user_id ORDER BY first_post_ts ASC LIMIT 3;
+SELECT did AS user_id, date_part('millisecond',(max(time_us) - min(time_us))) AS activity_span FROM bluesky WHERE kind = 'commit' AND commit_operation = 'create' AND commit_collection = 'app.bsky.feed.post' GROUP BY user_id ORDER BY activity_span DESC LIMIT 3;

greptimedb/queries_formatted.sql

@@ -3,7 +3,7 @@
 ------------------------------------------------------------------------------------------------------------------------
 SELECT commit_collection AS event,
        count(1) AS cnt
-FROM jsontable
+FROM bluesky
 GROUP BY event
 ORDER BY cnt DESC;
 
@@ -13,7 +13,7 @@ ORDER BY cnt DESC;
 SELECT commit_collection AS event,
        count(1) AS cnt,
        count(DISTINCT did) AS users
-FROM jsontable
+FROM bluesky
 WHERE kind = 'commit'
   AND commit_operation = 'create'
 GROUP BY event
@@ -25,7 +25,7 @@ ORDER BY cnt DESC;
 SELECT commit_collection AS event,
        date_part('hour', time_us) AS hour_of_day,
        count(1) AS cnt
-FROM jsontable
+FROM bluesky
 WHERE kind = 'commit'
   AND commit_operation = 'create'
   AND commit_collection IN('app.bsky.feed.post', 'app.bsky.feed.repost', 'app.bsky.feed.like')
@@ -39,7 +39,7 @@ ORDER BY hour_of_day,
 ------------------------------------------------------------------------------------------------------------------------
 SELECT did AS user_id,
        min(time_us) AS first_post_ts
-FROM jsontable
+FROM bluesky
 WHERE kind = 'commit'
   AND commit_operation = 'create'
   AND commit_collection = 'app.bsky.feed.post'
@@ -51,7 +51,7 @@ ORDER BY first_post_ts ASC LIMIT 3;
 ------------------------------------------------------------------------------------------------------------------------
 SELECT did AS user_id,
        date_part('millisecond',(max(time_us) - min(time_us))) AS activity_span
-FROM jsontable
+FROM bluesky
 WHERE kind = 'commit'
   AND commit_operation = 'create'
   AND commit_collection = 'app.bsky.feed.post'