Large dataframe cannot be inserted to the clickhouse table

[SergeRomankevich]

Describe the bug
I am creating a clickhouse table.
CREATE TABLE Data_BI.Pt (DataVersionString,DeletionMarkUInt8,DescriptionString,post_KeyString,dog_KeyString,docString,doc_TypeString,store_KeyString,date_docDate,PredefinedUInt8,PredefinedDataName String ) ENGINE = MergeTree PARTITION BY toYYYYMM(date_doc) ORDER BY (post_Key, dog_Key, store_Key) SETTINGS index_granularity = 8192

Super fast inserting the data of the partion 2009.json (306Mb) file into the table.
Trying to insert partion 2013.json (874Mb) and fail.
Python gives no error. Clickhouse log has no errors. Clickhouse writes 0 lines inserted. What is the reason?

`import pandas as pd
import time
import datetime as dt
from clickhouse_driver.client import Client
s = time.time()

def readJSON(ye):
s0 = time.time()
f = pd.read_json('/tmp/Part/partion ' + str(ye) + '.json')
dd = pd.DataFrame.from_records(f['value'])
dd['ДатаДокумента'] = pd.to_datetime(dd['ДатаДокумента'], infer_datetime_format=True)
st=pd.to_datetime('1970-01-01')
dd['ДатаДокумента']=(dd['ДатаДокумента']-st).dt.days
s1 = time.time()
s11=s1-s0
print(f'Фрейм {str(ye)} года за {round(s11//60)} минут {round(s11%60)} сек.')
return dd

def insert_srv_etl(df0,ye):
#ch = Client('srv-etl', settings={'use_numpy': True})
ch = Client('srv-etl', settings={'use_numpy': True,'max_insert_block_size': 524288})
strok=ch.insert_dataframe('INSERT INTO Data_BI.Pt VALUES', df0)
ch.disconnect()
print(f'Вставлено {str(strok)} строк за {str(ye)} год')

df=readJSON('2013')
insert_srv_etl(df,'2013')`

Versions

• pandas 1.3.1
• clickhouse-driver 0.2.1
• ClickHouse server version 21.7.2.7 (official build).
• Python 3.7.3

[xzkostyan]

Fix was merged into master branch.