fix(ci): run entire snapshot in main, pr and main-cron workflows (#8523)

Author: kwannoel

ci/scripts/deterministic-e2e-test.sh

@@ -45,4 +45,4 @@ echo "--- deterministic simulation e2e, ci-3cn-2fe, parallel, batch"
 seq $TEST_NUM | parallel MADSIM_TEST_SEED={} './risingwave_simulation -j 16 ./e2e_test/batch/\*\*/\*.slt 2> $LOGDIR/parallel-batch-{}.log && rm $LOGDIR/parallel-batch-{}.log'
 
 echo "--- deterministic simulation e2e, ci-3cn-2fe, fuzzing (pre-generated-queries)"
-seq $TEST_NUM | parallel MADSIM_TEST_SEED={} './risingwave_simulation  --run-sqlsmith-queries ./src/tests/sqlsmith/tests/sqlsmith-query-snapshots/{} 2> $LOGDIR/fuzzing-{}.log && rm $LOGDIR/fuzzing-{}.log'
+seq 64 | parallel MADSIM_TEST_SEED={} './risingwave_simulation  --run-sqlsmith-queries ./src/tests/sqlsmith/tests/sqlsmith-query-snapshots/{} 2> $LOGDIR/fuzzing-{}.log && rm $LOGDIR/fuzzing-{}.log'

ci/workflows/pull-request.yml

@@ -228,6 +228,7 @@ steps:
       - ./ci/plugins/upload-failure-logs
     timeout_in_minutes: 15
     retry: *auto-retry
+    soft_fail: true
 
   - label: "check"
     command: "ci/scripts/check.sh"

src/tests/sqlsmith/scripts/gen_queries.sh

@@ -142,11 +142,25 @@ check_failed_to_generate_queries() {
   fi
 }
 
+# sync step
+# Some queries maybe be added
+sync_queries() {
+  set +x
+  pushd $OUTDIR
+  git checkout main
+  git pull
+  set +e
+  git branch -D stage
+  set -e
+  git checkout -b stage
+  popd
+  set -x
+}
+
 # Upload step
 upload_queries() {
   set +x
   pushd "$OUTDIR"
-  git checkout -b stage
   git add .
   git commit -m 'update queries'
   git push -f origin stage
@@ -209,6 +223,11 @@ validate() {
   echo_err "[INFO] Passed checks"
 }
 
+sync() {
+  sync_queries
+  echo_err "[INFO] Synced"
+}
+
 upload() {
   upload_queries
   echo_err "[INFO] Uploaded"
@@ -223,6 +242,7 @@ main() {
   setup
 
   build
+  sync
   generate
   validate
   upload

src/tests/sqlsmith/src/sql_gen/scalar.rs

@@ -148,15 +148,19 @@ impl<'a, R: Rng> SqlGenerator<'a, R> {
     fn gen_temporal_scalar(&mut self, typ: &DataType) -> String {
         use DataType as T;
 
-        let rand_secs = self.rng.gen_range(2..1000000) as u64;
         let minute = 60;
         let hour = 60 * minute;
         let day = 24 * hour;
         let week = 7 * day;
-        let choices = [0, 1, minute, hour, day, week, rand_secs];
-        let secs = choices.choose(&mut self.rng).unwrap();
+        let choices = [0, 1, minute, hour, day, week];
 
-        let tm = DateTime::<Utc>::from(SystemTime::now() - Duration::from_secs(*secs));
+        let secs = match self.rng.gen_range(1..=100) {
+            1..=30 => *choices.choose(&mut self.rng).unwrap(),
+            31..=100 => self.rng.gen_range(2..100) as u64,
+            _ => unreachable!(),
+        };
+
+        let tm = DateTime::<Utc>::from(SystemTime::now() - Duration::from_secs(secs));
         match typ {
             T::Date => tm.format("%F").to_string(),
             T::Timestamp | T::Timestamptz => tm.format("%Y-%m-%d %H:%M:%S").to_string(),
@@ -169,7 +173,7 @@ impl<'a, R: Rng> SqlGenerator<'a, R> {
             T::Time => tm.format("%T").to_string(),
             T::Interval => {
                 if self.rng.gen_bool(0.5) {
-                    (-(*secs as i64)).to_string()
+                    (-(secs as i64)).to_string()
                 } else {
                     secs.to_string()
                 }