thevar1able/ClickHouse
1
0
Fork 0
mirror of https://github.com/ClickHouse/ClickHouse.git synced 2024-11-23 08:02:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

add none_only_active DDL output mode

Browse Source
This commit is contained in:
Alexander Tokmakov 2024-02-23 01:34:17 +01:00
parent f83ddb1a70
commit 011c694117
9 changed files with 29 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
docs/en/operations/settings/settings.md
View File

@ -3956,6 +3956,7 @@ Possible values:
- `none` — Is similar to throw, but distributed DDL query returns no result set.
- `null_status_on_timeout` — Returns `NULL` as execution status in some rows of result set instead of throwing `TIMEOUT_EXCEEDED` if query is not finished on the corresponding hosts.
- `never_throw` — Do not throw `TIMEOUT_EXCEEDED` and do not rethrow exceptions if query has failed on some hosts.
- `none_only_active` - similar to `none`, but doesn't wait for inactive replicas of the `Replicated` database. Note: with this mode it's impossible to figure out that the query was not executed on some replica and will be executed in background.
- `null_status_on_timeout_only_active` — similar to `null_status_on_timeout`, but doesn't wait for inactive replicas of the `Replicated` database
- `throw_only_active` — similar to `throw`, but doesn't wait for inactive replicas of the `Replicated` database

2
src/Core/Settings.h
View File

@ -695,7 +695,7 @@ class IColumn;
M(Bool, database_replicated_allow_replicated_engine_arguments, true, "Allow to create only Replicated tables in database with engine Replicated with explicit arguments", 0) \
M(Bool, cloud_mode, false, "Only available in ClickHouse Cloud", 0) \
M(UInt64, cloud_mode_engine, 1, "Only available in ClickHouse Cloud", 0) \
M(DistributedDDLOutputMode, distributed_ddl_output_mode, DistributedDDLOutputMode::THROW, "Format of distributed DDL query result, one of: 'none', 'throw', 'null_status_on_timeout', 'never_throw'", 0) \
M(DistributedDDLOutputMode, distributed_ddl_output_mode, DistributedDDLOutputMode::THROW, "Format of distributed DDL query result, one of: 'none', 'throw', 'null_status_on_timeout', 'never_throw', 'none_only_active', 'throw_only_active', 'null_status_on_timeout_only_active'", 0) \
M(UInt64, distributed_ddl_entry_format_version, 5, "Compatibility version of distributed DDL (ON CLUSTER) queries", 0) \
\
M(UInt64, external_storage_max_read_rows, 0, "Limit maximum number of rows when table with external engine should flush history data. Now supported only for MySQL table engine, database engine, dictionary and MaterializedMySQL. If equal to 0, this setting is disabled", 0) \

1
src/Core/SettingsEnums.cpp
View File

@ -117,6 +117,7 @@ IMPLEMENT_SETTING_ENUM(DistributedDDLOutputMode, ErrorCodes::BAD_ARGUMENTS,
{"null_status_on_timeout", DistributedDDLOutputMode::NULL_STATUS_ON_TIMEOUT},
{"throw_only_active", DistributedDDLOutputMode::THROW_ONLY_ACTIVE},
{"null_status_on_timeout_only_active", DistributedDDLOutputMode::NULL_STATUS_ON_TIMEOUT_ONLY_ACTIVE},
{"none_only_active", DistributedDDLOutputMode::NONE_ONLY_ACTIVE},
{"never_throw", DistributedDDLOutputMode::NEVER_THROW}})
IMPLEMENT_SETTING_ENUM(StreamingHandleErrorMode, ErrorCodes::BAD_ARGUMENTS,

1
src/Core/SettingsEnums.h
View File

@ -175,6 +175,7 @@ enum class DistributedDDLOutputMode
NEVER_THROW,
THROW_ONLY_ACTIVE,
NULL_STATUS_ON_TIMEOUT_ONLY_ACTIVE,
NONE_ONLY_ACTIVE,
};
DECLARE_SETTING_ENUM(DistributedDDLOutputMode)

25
src/Interpreters/executeDDLQueryOnCluster.cpp
View File

@ -252,7 +252,8 @@ BlockIO getDistributedDDLStatus(const String & node_path, const DDLLogEntry & en
auto source = std::make_shared<DDLQueryStatusSource>(node_path, entry, context, hosts_to_wait);
io.pipeline = QueryPipeline(std::move(source));
if (context->getSettingsRef().distributed_ddl_output_mode == DistributedDDLOutputMode::NONE)
if (context->getSettingsRef().distributed_ddl_output_mode == DistributedDDLOutputMode::NONE ||
context->getSettingsRef().distributed_ddl_output_mode == DistributedDDLOutputMode::NONE_ONLY_ACTIVE)
io.pipeline.complete(std::make_shared<EmptySink>(io.pipeline.getHeader()));
return io;
@ -264,7 +265,9 @@ Block DDLQueryStatusSource::getSampleBlock(ContextPtr context_, bool hosts_to_wa
auto maybe_make_nullable = [&](const DataTypePtr & type) -> DataTypePtr
{
if (output_mode == DistributedDDLOutputMode::THROW || output_mode == DistributedDDLOutputMode::NONE)
if (output_mode == DistributedDDLOutputMode::THROW ||
output_mode == DistributedDDLOutputMode::NONE ||
output_mode == DistributedDDLOutputMode::NONE_ONLY_ACTIVE)
return type;
return std::make_shared<DataTypeNullable>(type);
};
@ -313,14 +316,15 @@ DDLQueryStatusSource::DDLQueryStatusSource(
{
auto output_mode = context->getSettingsRef().distributed_ddl_output_mode;
throw_on_timeout = output_mode == DistributedDDLOutputMode::THROW || output_mode == DistributedDDLOutputMode::THROW_ONLY_ACTIVE
|| output_mode == DistributedDDLOutputMode::NONE;
|| output_mode == DistributedDDLOutputMode::NONE || output_mode == DistributedDDLOutputMode::NONE_ONLY_ACTIVE;
if (hosts_to_wait)
{
waiting_hosts = NameSet(hosts_to_wait->begin(), hosts_to_wait->end());
is_replicated_database = true;
only_running_hosts = output_mode == DistributedDDLOutputMode::THROW_ONLY_ACTIVE ||
output_mode == DistributedDDLOutputMode::NULL_STATUS_ON_TIMEOUT_ONLY_ACTIVE;
output_mode == DistributedDDLOutputMode::NULL_STATUS_ON_TIMEOUT_ONLY_ACTIVE ||
output_mode == DistributedDDLOutputMode::NONE_ONLY_ACTIVE;
}
else
{
@ -442,14 +446,16 @@ Chunk DDLQueryStatusSource::generate()
size_t num_unfinished_hosts = waiting_hosts.size() - num_hosts_finished;
size_t num_active_hosts = current_active_hosts.size();
constexpr auto msg_format = "Watching task {} is executing longer than distributed_ddl_task_timeout (={}) seconds. "
"There are {} unfinished hosts ({} of them are currently executing the task), "
"they are going to execute the query in background";
constexpr auto msg_format = "Distributed DDL task {} is not finished on {} of {} hosts "
"({} of them are currently executing the task, {} are inactive). "
"They are going to execute the query in background. Was waiting for {} seconds{}";
if (throw_on_timeout)
{
if (!first_exception)
first_exception = std::make_unique<Exception>(Exception(ErrorCodes::TIMEOUT_EXCEEDED,
msg_format, node_path, timeout_seconds, num_unfinished_hosts, num_active_hosts));
msg_format, node_path, num_unfinished_hosts, waiting_hosts.size(), num_active_hosts, offline_hosts.size(),
watch.elapsedSeconds(), stop_waiting_offline_hosts ? "" : ", which is longer than distributed_ddl_task_timeout"));
/// For Replicated database print a list of unfinished hosts as well. Will return empty block on next iteration.
if (is_replicated_database)
@ -457,7 +463,8 @@ Chunk DDLQueryStatusSource::generate()
return {};
}
LOG_INFO(log, msg_format, node_path, timeout_seconds, num_unfinished_hosts, num_active_hosts);
LOG_INFO(log, msg_format, node_path, num_unfinished_hosts, waiting_hosts.size(), num_active_hosts, offline_hosts.size(),
watch.elapsedSeconds(), stop_waiting_offline_hosts ? "" : "which is longer than distributed_ddl_task_timeout");
return generateChunkWithUnfinishedHosts();
}

4
tests/queries/0_stateless/01175_distributed_ddl_output_mode_long.reference
View File

@ -3,7 +3,7 @@ Received exception from server:
Code: 57. Error: Received from localhost:9000. Error: There was an error on [localhost:9000]: Code: 57. Error: Table default.none already exists. (TABLE_ALREADY_EXISTS)
(query: create table none on cluster test_shard_localhost (n int) engine=Memory;)
Received exception from server:
Code: 159. Error: Received from localhost:9000. Error: Watching task <task> is executing longer than distributed_ddl_task_timeout (=1) seconds. There are 1 unfinished hosts (0 of them are currently executing the task), they are going to execute the query in background. (TIMEOUT_EXCEEDED)
Code: 159. Error: Received from localhost:9000. Error: Distributed DDL task <task> is not finished on 1 of 2 hosts (0 of them are currently executing the task, 0 are inactive). They are going to execute the query in background. Was waiting for <sec> seconds, which is longer than distributed_ddl_task_timeout. (TIMEOUT_EXCEEDED)
(query: drop table if exists none on cluster test_unavailable_shard;)
throw
localhost 9000 0 0 0
@ -12,7 +12,7 @@ Code: 57. Error: Received from localhost:9000. Error: There was an error on [loc
(query: create table throw on cluster test_shard_localhost (n int) engine=Memory format Null;)
localhost 9000 0 1 0
Received exception from server:
Code: 159. Error: Received from localhost:9000. Error: Watching task <task> is executing longer than distributed_ddl_task_timeout (=1) seconds. There are 1 unfinished hosts (0 of them are currently executing the task), they are going to execute the query in background. (TIMEOUT_EXCEEDED)
Code: 159. Error: Received from localhost:9000. Error: Distributed DDL task <task> is not finished on 1 of 2 hosts (0 of them are currently executing the task, 0 are inactive). They are going to execute the query in background. Was waiting for <sec> seconds, which is longer than distributed_ddl_task_timeout. (TIMEOUT_EXCEEDED)
(query: drop table if exists throw on cluster test_unavailable_shard;)
null_status_on_timeout
localhost 9000 0 0 0

4
tests/queries/0_stateless/01175_distributed_ddl_output_mode_long.sh
View File

@ -54,14 +54,14 @@ $CLIENT --distributed_ddl_output_mode=none -q "create table none on cluster test
$CLIENT --distributed_ddl_output_mode=none -q "create table none on cluster test_shard_localhost (n int) engine=Memory;" 2>&1 | sed "s/DB::Exception/Error/g" | sed "s/ (version.*)//"
# Timeout
run_until_out_contains 'There are 1 unfinished hosts' $CLICKHOUSE_CLIENT_WITH_SETTINGS --distributed_ddl_output_mode=none -q "drop table if exists none on cluster test_unavailable_shard;" 2>&1 | sed "s/DB::Exception/Error/g" | sed "s/ (version.*)//" | sed "s/Watching task .* is executing longer/Watching task <task> is executing longer/"
run_until_out_contains 'not finished on 1 ' $CLICKHOUSE_CLIENT_WITH_SETTINGS --distributed_ddl_output_mode=none -q "drop table if exists none on cluster test_unavailable_shard;" 2>&1 | sed "s/DB::Exception/Error/g" | sed "s/ (version.*)//" | sed "s/Distributed DDL task .* is not finished/Distributed DDL task <task> is not finished/" | sed "s/for .* seconds/for <sec> seconds/"
$CLIENT --distributed_ddl_output_mode=throw -q "select value from system.settings where name='distributed_ddl_output_mode';"
$CLIENT --distributed_ddl_output_mode=throw -q "create table throw on cluster test_shard_localhost (n int) engine=Memory;"
$CLIENT --distributed_ddl_output_mode=throw -q "create table throw on cluster test_shard_localhost (n int) engine=Memory format Null;" 2>&1 | sed "s/DB::Exception/Error/g" | sed "s/ (version.*)//"
run_until_out_contains 'There are 1 unfinished hosts' $CLICKHOUSE_CLIENT_WITH_SETTINGS --distributed_ddl_output_mode=throw -q "drop table if exists throw on cluster test_unavailable_shard;" 2>&1 | sed "s/DB::Exception/Error/g" | sed "s/ (version.*)//" | sed "s/Watching task .* is executing longer/Watching task <task> is executing longer/"
run_until_out_contains 'not finished on 1 ' $CLICKHOUSE_CLIENT_WITH_SETTINGS --distributed_ddl_output_mode=throw -q "drop table if exists throw on cluster test_unavailable_shard;" 2>&1 | sed "s/DB::Exception/Error/g" | sed "s/ (version.*)//" | sed "s/Distributed DDL task .* is not finished/Distributed DDL task <task> is not finished/" | sed "s/for .* seconds/for <sec> seconds/"
$CLIENT --distributed_ddl_output_mode=null_status_on_timeout -q "select value from system.settings where name='distributed_ddl_output_mode';"

2
tests/queries/0_stateless/02447_drop_database_replica.reference
View File

@ -13,6 +13,7 @@ t
rdb_default 1 1 s1 r1 1
2
2
2
s1 r1 OK 2 0
s1 r2 QUEUED 2 0
s2 r1 QUEUED 2 0
@ -24,4 +25,5 @@ rdb_default 1 2 s1 r2 0
t
t2
t3
t4
rdb_default_4 1 1 s1 r1 1

5
tests/queries/0_stateless/02447_drop_database_replica.sh
View File

@ -33,8 +33,9 @@ $CLICKHOUSE_CLIENT -q "select cluster, shard_num, replica_num, database_shard_na
$CLICKHOUSE_CLIENT -q "system drop database replica 's1|r1' from database $db2" 2>&1| grep -Fac "is active, cannot drop it"
# Also check that it doesn't exceed distributed_ddl_task_timeout waiting for inactive replicas
timeout 60s $CLICKHOUSE_CLIENT --distributed_ddl_task_timeout=1000 --distributed_ddl_output_mode=throw_only_active -q "create table $db.t2 (n int) engine=Log" 2>&1| grep -Fac "TIMEOUT_EXCEEDED"
timeout 60s $CLICKHOUSE_CLIENT --distributed_ddl_task_timeout=1000 --distributed_ddl_output_mode=null_status_on_timeout_only_active -q "create table $db.t3 (n int) engine=Log" | sort
timeout 60s $CLICKHOUSE_CLIENT --distributed_ddl_task_timeout=1000 --distributed_ddl_output_mode=none_only_active -q "create table $db.t2 (n int) engine=Log" 2>&1| grep -Fac "TIMEOUT_EXCEEDED"
timeout 60s $CLICKHOUSE_CLIENT --distributed_ddl_task_timeout=1000 --distributed_ddl_output_mode=throw_only_active -q "create table $db.t3 (n int) engine=Log" 2>&1| grep -Fac "TIMEOUT_EXCEEDED"
timeout 60s $CLICKHOUSE_CLIENT --distributed_ddl_task_timeout=1000 --distributed_ddl_output_mode=null_status_on_timeout_only_active -q "create table $db.t4 (n int) engine=Log" | sort
$CLICKHOUSE_CLIENT -q "detach database $db3"
$CLICKHOUSE_CLIENT -q "system drop database replica 'r1' from shard 's2' from database $db"