From 6d786f9734e8226fa23135ee26236af695203717 Mon Sep 17 00:00:00 2001
From: Azat Khuzhin <a3at.mail@gmail.com>
Date: Fri, 16 Jul 2021 22:59:02 +0300
Subject: [PATCH] Rewrite 01882_total_rows_approx using http interface w/o
 system.processes

---
 .../01882_total_rows_approx.reference         |  9 +---
 .../0_stateless/01882_total_rows_approx.sh    | 52 +++++--------------
 2 files changed, 13 insertions(+), 48 deletions(-)

diff --git a/tests/queries/0_stateless/01882_total_rows_approx.reference b/tests/queries/0_stateless/01882_total_rows_approx.reference
index 7f2070fc9cb..fd1fb9b7231 100644
--- a/tests/queries/0_stateless/01882_total_rows_approx.reference
+++ b/tests/queries/0_stateless/01882_total_rows_approx.reference
@@ -1,8 +1 @@
-Waiting for query to be started...
-Query started.
-Checking total_rows_approx.
-10
-10
-10
-10
-10
+"total_rows_to_read":"10"
diff --git a/tests/queries/0_stateless/01882_total_rows_approx.sh b/tests/queries/0_stateless/01882_total_rows_approx.sh
index f51e95b15c0..26333f61692 100755
--- a/tests/queries/0_stateless/01882_total_rows_approx.sh
+++ b/tests/queries/0_stateless/01882_total_rows_approx.sh
@@ -1,23 +1,12 @@
 #!/usr/bin/env bash
 
-# Check that total_rows_approx (via system.processes) includes all rows from
+# Check that total_rows_approx (via http headers) includes all rows from
 # all parts at the query start.
 #
 # At some point total_rows_approx was accounted only when the query starts
 # reading the part, and so total_rows_approx wasn't reliable, even for simple
 # SELECT FROM MergeTree()
 # It was fixed by take total_rows_approx into account as soon as possible.
-#
-# To check total_rows_approx this query starts the query in background,
-# that sleep's 1 second for each part, and by using max_threads=1 the query
-# reads parts sequentially and sleeps 1 second between parts.
-# Also the test spawns background process to check total_rows_approx for this
-# query.
-# It checks multiple times since at first few iterations the query may not
-# start yet (since there are 3 excessive sleep calls - 1 for primary key
-# analysis and 2 for partition pruning), and get only last 5 total_rows_approx
-# rows (one row is not enough since when the query finishes total_rows_approx
-# will be set to 10 anyway, regardless proper accounting).
 
 CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 # shellcheck source=../shell_config.sh
@@ -25,31 +14,14 @@ CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
 
 $CLICKHOUSE_CLIENT -q "drop table if exists data_01882"
 $CLICKHOUSE_CLIENT -q "create table data_01882 (key Int) Engine=MergeTree() partition by key order by key as select * from numbers(10)"
-QUERY_ID="$CLICKHOUSE_TEST_NAME-$(tr -cd '[:lower:]' < /dev/urandom | head -c10)"
-
-function check_background_query()
-{
-    echo "Waiting for query to be started..."
-    while [[ $($CLICKHOUSE_CLIENT --param_query_id="$QUERY_ID" -q 'select count() from system.processes where query_id = {query_id:String}') != 1 ]]; do
-        sleep 0.01
-    done
-    echo "Query started."
-
-    echo "Checking total_rows_approx."
-    # check total_rows_approx multiple times
-    # (to make test more reliable to what it covers)
-    local i=0
-    for ((i = 0; i < 20; ++i)); do
-        $CLICKHOUSE_CLIENT --param_query_id="$QUERY_ID" -q 'select total_rows_approx from system.processes where query_id = {query_id:String}'
-        (( ++i ))
-        sleep 1
-    done | tail -n5
-}
-check_background_query &
-
-# this query will sleep 10 seconds in total, 1 seconds for each part (10 parts).
-$CLICKHOUSE_CLIENT -q "select *, sleepEachRow(1) from data_01882" --max_threads=1 --format Null --query_id="$QUERY_ID" --max_block_size=1
-
-wait
-
-$CLICKHOUSE_CLIENT -q "drop table data_01882"
+# send_progress_in_http_headers will periodically send the progress
+# but this is not stable, i.e. it can be dumped on query end,
+# thus check few times to be sure that this is not coincidence.
+for _ in {1..30}; do
+    $CLICKHOUSE_CURL -vsS "${CLICKHOUSE_URL}&max_threads=1&default_format=Null&send_progress_in_http_headers=1&http_headers_progress_interval_ms=1" --data-binary @- <<< "select * from data_01882" |& {
+        grep -o -F '"total_rows_to_read":"10"'
+    } | {
+        # grep out final result
+        grep -v -F '"read_rows":"10"'
+    }
+done | uniq