more tests + clickhouse-client options

programs/client/Client.cpp

@@ -2180,6 +2180,8 @@ public:
             ("log-level", po::value<std::string>(), "client log level")
             ("server_logs_file", po::value<std::string>(), "put server logs into specified file")
             ("query-fuzzer-runs", po::value<int>()->default_value(0), "query fuzzer runs")
+            ("opentelemetry-traceparent", po::value<std::string>(), "OpenTelemetry traceparent header as described by W3C Trace Context recommendation")
+            ("opentelemetry-tracestate", po::value<std::string>(), "OpenTelemetry tracestate header as described by W3C Trace Context recommendation")
         ;
 
         Settings cmd_settings;
@@ -2348,6 +2350,25 @@ public:
             ignore_error = true;
         }
 
+        if (options.count("opentelemetry-traceparent"))
+        {
+            std::string traceparent = options["opentelemetry-traceparent"].as<std::string>();
+            std::string error;
+            if (!context.getClientInfo().setOpenTelemetryTraceparent(
+                traceparent, error))
+            {
+                throw Exception(ErrorCodes::BAD_ARGUMENTS,
+                    "Cannot parse OpenTelemetry traceparent '{}': {}",
+                    traceparent, error);
+            }
+        }
+
+        if (options.count("opentelemetry-tracestate"))
+        {
+            context.getClientInfo().opentelemetry_tracestate =
+                options["opentelemetry-tracestate"].as<std::string>();
+        }
+
         argsToConfig(common_arguments, config(), 100);
 
         clearPasswordFromCommandLine(argc, argv);

tests/queries/0_stateless/01455_opentelemetry.sh

@@ -1,31 +0,0 @@
-#!/usr/bin/env bash
-set -ue
-
-CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
-. "$CURDIR"/../shell_config.sh
-
-# Generate some random trace id so that the prevous runs of the test do not interfere.
-trace_id=$(${CLICKHOUSE_CLIENT} -q "select lower(hex(reverse(reinterpretAsString(generateUUIDv4()))))")
-
-${CLICKHOUSE_CURL} --header "traceparent: 00-$trace_id-0000000000000010-01" --header "tracestate: some custom state" "http://localhost:8123/" --get --data-urlencode "query=select 1 from remote('127.0.0.2', system, one)"
-
-${CLICKHOUSE_CLIENT} -q "system flush logs"
-
-# Check that the HTTP traceparent was read, and then passed to `remote` instance.
-# We expect 4 queries, because there are two DESC TABLE queries for the shard.
-# This is bug-ish, see https://github.com/ClickHouse/ClickHouse/issues/14228
-${CLICKHOUSE_CLIENT} -q "select count(*) from system.opentelemetry_log where trace_id = reinterpretAsUUID(reverse(unhex('$trace_id')))"
-
-# Check that the tracestate header was read and passed. Must have
-# exactly the same value for all "query" spans in this trace.
-${CLICKHOUSE_CLIENT} -q "
-    select count(distinct attribute.values)
-    from system.opentelemetry_log
-        array join attribute.names, attribute.values
-    where
-        trace_id = reinterpretAsUUID(reverse(unhex('$trace_id')))
-        and operation_name = 'query'
-        and attribute.names = 'tracestate'
-"
-
-

tests/queries/0_stateless/01455_opentelemetry_distributed.reference

@@ -1,3 +1,6 @@
 1
 4
 1
+1
+2
+1

tests/queries/0_stateless/01455_opentelemetry_distributed.sh

@@ -0,0 +1,52 @@
+#!/usr/bin/env bash
+set -ue
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+. "$CURDIR"/../shell_config.sh
+
+function check_log
+{
+${CLICKHOUSE_CLIENT} -nq "
+system flush logs;
+
+-- Check the number of spans with given trace id, to verify it was propagated.
+select count(*)
+    from system.opentelemetry_log
+    where trace_id = reinterpretAsUUID(reverse(unhex('$trace_id')))
+        and operation_name = 'query'
+    ;
+
+-- Check that the tracestate header was propagated. It must have exactly the
+-- same non-empty value for all 'query' spans in this trace.
+select count(distinct value)
+    from system.opentelemetry_log
+        array join attribute.names as name, attribute.values as value
+    where
+        trace_id = reinterpretAsUUID(reverse(unhex('$trace_id')))
+        and operation_name = 'query'
+        and name = 'tracestate'
+        and length(value) > 0
+    ;
+"
+}
+
+# Generate some random trace id so that the prevous runs of the test do not interfere.
+trace_id=$(${CLICKHOUSE_CLIENT} -q "select lower(hex(reverse(reinterpretAsString(generateUUIDv4()))))")
+
+# Check that the HTTP traceparent is read, and then passed through `remote` table function.
+# We expect 4 queries, because there are two DESC TABLE queries for the shard.
+# This is bug-ish, see https://github.com/ClickHouse/ClickHouse/issues/14228
+${CLICKHOUSE_CURL} --header "traceparent: 00-$trace_id-0000000000000010-01" --header "tracestate: some custom state" "http://localhost:8123/" --get --data-urlencode "query=select 1 from remote('127.0.0.2', system, one)"
+
+check_log
+
+# With another trace id, check that clickhouse-client accepts traceparent, and
+# that it is passed through URL table function. We expect two query spans, one
+# for the initial query, and one for the HTTP query.
+trace_id=$(${CLICKHOUSE_CLIENT} -q "select lower(hex(reverse(reinterpretAsString(generateUUIDv4()))))")
+
+${CLICKHOUSE_CLIENT} --opentelemetry-traceparent "00-$trace_id-0000000000000020-02" --opentelemetry-tracestate "another custom state" --query "
+    select * from url('http://127.0.0.2:8123/?query=select%201', CSV, 'a int')
+"
+
+check_log