clickhouse client is going down beacuse of query on Materialized Views - clickhouse

Question

when creating mv with the target table
target table:
CREATE TABLE target_test
(
day date,
hour UInt32,
test_sum SimpleAggregateFunction(sum, Float64)
)
ENGINE = SummingMergeTree()
PARTITION BY toRelativeWeekNum(toDateTime(day))
ORDER BY (day, hour)
CREATE MATERIALIZED VIEW target_test_mv
TO target_test
as select
toDate(session_ts) as day,
toHour(toDateTime(session_ts)) as hour,
sum(if (test is null, 0,test)) as test_sum
from events
where session_ts>= now()-1000
group by day, hour
when the data is inserting into the target table and when I am running this query
clickhouse server is crashed because of the query why??
select sum(revenue_sum) from target_test_mv
this is from the log :
2020.12.07 12:03:48.059559 [ 19243 ] {} <Fatal> BaseDaemon: ########################################
2020.12.07 12:03:48.059595 [ 19244 ] {} <Fatal> BaseDaemon: ########################################
2020.12.07 12:03:48.059642 [ 19244 ] {} <Fatal> BaseDaemon: (version 20.4.2.9 (official build)) (from thread 19242) (query_id: 6651689c-484a-4b53-a174-06fd01848208) Received signal Segmentation fault (11).
2020.12.07 12:03:48.059644 [ 19243 ] {} <Fatal> BaseDaemon: (version 20.4.2.9 (official build)) (from thread 19241) (query_id: 6651689c-484a-4b53-a174-06fd01848208) Received signal Segmentation fault (11).
2020.12.07 12:03:48.059685 [ 19244 ] {} <Fatal> BaseDaemon: Address: NULL pointer. Access: read. Address not mapped to object.
2020.12.07 12:03:48.059691 [ 19243 ] {} <Fatal> BaseDaemon: Address: NULL pointer. Access: read. Address not mapped to object.
2020.12.07 12:03:48.059710 [ 19244 ] {} <Fatal> BaseDaemon: Stack trace: 0xcad040b 0xd26b20d 0xd281202 0xdcbb2f3 0xdcbf909 0xdb0b9a1 0xdb0f95d 0xdb10324 0x901e26b 0x901c753 0x7fec413f0dd5 0x7fec41a04ead
2020.12.07 12:03:48.059712 [ 19243 ] {} <Fatal> BaseDaemon: Stack trace: 0xcad040b 0xd26b20d 0xd281202 0xdcbb2f3 0xdcbf909 0xdb0b9a1 0xdb0f95d 0xdb10324 0x901e26b 0x901c753 0x7fec413f0dd5 0x7fec41a04ead
2020.12.07 12:03:48.059781 [ 19244 ] {} <Fatal> BaseDaemon: 3. DB::IAggregateFunctionHelper<DB::AggregateFunctionNullUnary<true> >::addBatchSinglePlace(unsigned long, char*, DB::IColumn const**, DB::Arena*) const # 0xcad040b in /opt/clickhouse/engine/clickhouse-common-static-20.4.2.9/usr/bin/clickhouse
2020.12.07 12:03:48.059781 [ 19243 ] {} <Fatal> BaseDaemon: 3. DB::IAggregateFunctionHelper<DB::AggregateFunctionNullUnary<true> >::addBatchSinglePlace(unsigned long, char*, DB::IColumn const**, DB::Arena*) const # 0xcad040b in /opt/clickhouse/engine/clickhouse-common-static-20.4.2.9/usr/bin/clickhouse
2020.12.07 12:03:48.059856 [ 19243 ] {} <Fatal> BaseDaemon: 4. DB::Aggregator::executeWithoutKeyImpl(char*&, unsigned long, DB::Aggregator::AggregateFunctionInstruction*, DB::Arena*) # 0xd26b20d in /opt/clickhouse/engine/clickhouse-common-static-20.4.2.9/usr/bin/clickhouse
2020.12.07 12:03:48.059858 [ 19244 ] {} <Fatal> BaseDaemon: 4. DB::Aggregator::executeWithoutKeyImpl(char*&, unsigned long, DB::Aggregator::AggregateFunctionInstruction*, DB::Arena*) # 0xd26b20d in /opt/clickhouse/engine/clickhouse-common-static-20.4.2.9/usr/bin/clickhouse
2020.12.07 12:03:48.059902 [ 19243 ] {} <Fatal> BaseDaemon: 5. DB::Aggregator::executeOnBlock(std::__1::vector<COW<DB::IColumn>::immutable_ptr<DB::IColumn>, std::__1::allocator<COW<DB::IColumn>::immutable_ptr<DB::IColumn> > >, unsigned long, DB::AggregatedDataVariants&, std::__1::vector<DB::IColumn const*, std::__1::allocator<DB::IColumn const*> >&, std::__1::vector<std::__1::vector<DB::IColumn const*, std::__1::allocator<DB::IColumn const*> >, std::__1::allocator<std::__1::vector<DB::IColumn const*, std::__1::allocator<DB::IColumn const*> > > >&, bool&) # 0xd281202 in /opt/clickhouse/engine/clickhouse-common-static-20.4.2.9/usr/bin/clickhouse
2020.12.07 12:03:48.059917 [ 19244 ] {} <Fatal> BaseDaemon: 5. DB::Aggregator::executeOnBlock(std::__1::vector<COW<DB::IColumn>::immutable_ptr<DB::IColumn>, std::__1::allocator<COW<DB::IColumn>::immutable_ptr<DB::IColumn> > >, unsigned long, DB::AggregatedDataVariants&, std::__1::vector<DB::IColumn const*, std::__1::allocator<DB::IColumn const*> >&, std::__1::vector<std::__1::vector<DB::IColumn const*, std::__1::allocator<DB::IColumn const*> >, std::__1::allocator<std::__1::vector<DB::IColumn const*, std::__1::allocator<DB::IColumn const*> > > >&, bool&) # 0xd281202 in /opt/clickhouse/engine/clickhouse-common-static-20.4.2.9/usr/bin/clickhouse

Answer 1

I've tried a scenario similar to yours in a more recent version of ClickHouse and it works for me:
CREATE TABLE target_test
( day date,
hour UInt32,
test_sum Float64 )
ENGINE = SummingMergeTree() PARTITION BY toRelativeWeekNum(toDateTime(day)) ORDER BY (day, hour)
CREATE MATERIALIZED VIEW target_test_mv TO target_test as
select
toDate(event_time) as day,
toHour(toDateTime(event_time)) as hour,
sum(if (read_rows is null, 0,read_rows)) as test_sum
from system.query_log
group by day, hour
limit 10000
:) select sum(test_sum) from target_test;
SELECT sum(test_sum)
FROM target_test
┌─sum(test_sum)─┐
│ 20661 │
└───────────────┘
Note I'm using the query_log for the test and I've removed the SimpleAggregateFunction since the SummingMergeTree takes care of that.
You could try several things:
Try to reproduce the scenario above with the system.query_log instead of your events table, to discard there's any problem with your table.
Try on a more recent version of ClickHouse (20.4.2.9 is from May)
If your scenario still fails and given there's a segmentation fault in the logs, you can open a case in the ClickHouse GitHub repo.