Respect timezone env var

[iamtalhaasghar]

fix bug #250 which was introduced in #238

[bakwc]

Need to figure out why tests failing and fix

[iamtalhaasghar]

This test insert a datetime value of 2023-08-15 14:40:00

mysql_ch_replicator/tests/test_data_types.py

Line 235 in 05aa15f

f"(1, POINT(15.0, 14.0), '2,4,5', '2023-08-15 14:40:00', '110e6103-e39b-51d4-a716-826755413099', 'point');",

But the binlog replicator reads it as 2023-08-15 18:40:00 .i.e. adds an offset of +4

mysql_ch_replicator/mysql_ch_replicator/binlog_replicator.py

Line 524 in 05aa15f

f'records: {log_event.records}',

Output from binlog reader:
[binlogrepl 2026-03-23 00:40:54,359 DEBUG] store event ('mysql-bin.000003', 48309), event type: 1, database: replication-test_db table: test_table records: [[2, '1', b'\x00\x00\x00\x00\x01\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00.@\x00\x00\x00\x00\x00\x00,@', None, 26, datetime.datetime(2023, 8, 15, 18, 40), b'110e6103-e39b-51d4-a716-826755413099', 1]]

Binlogreader adds this offset ir-respective of the timezone value set in mysql_ch_replicator/__init__.py:

mysql_ch_replicator/mysql_ch_replicator/__init__.py

Line 4 in 05aa15f

os.environ['TZ'] = 'UTC'

So the actual root cause of the problem is binlogreader or the code which reads binlog files and prepares the events for mysql_ch_replicator to insert into clickhouse...

Edit:
i have made sure the timezone is matching for all containers. .i.e. mysql, clickhouse, test replicator etc.

❯ docker exec -it tests-mysql_db-1 /bin/bash
bash-5.1# date
Mon Mar 23 00:55:49 EDT 2026
bash-5.1#

❯ docker exec -it tests-clickhouse_db-1 /bin/bash
root@9958f9102306:/# date
Mon Mar 23 12:56:08 AM EDT 2026

❯ docker exec -it tests-replicator-1 /bin/bash
root@macbookpro:/app# date
Mon Mar 23 00:56:24 EDT 2026

EDIT 2:
when we do set timezone forcefully to UTC in mysql_ch_replicator/__init__.py the effect of additional +4 offset caused by binlogreader gets normalized and hence the test was NOT failing previously.

[bakwc]

Hm, need to think what could be done here, i will take a look, thanks.

[iamtalhaasghar]

But the binlog replicator reads it as 2023-08-15 18:40:00 .i.e. adds an offset of +4

mysql_ch_replicator/mysql_ch_replicator/binlog_replicator.py

Line 524 in 05aa15f

f'records: {log_event.records}',

Output from binlog reader: [binlogrepl 2026-03-23 00:40:54,359 DEBUG] store event ('mysql-bin.000003', 48309), event type: 1, database: replication-test_db table: test_table records: [[2, '1', b'\x00\x00\x00\x00\x01\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00.@\x00\x00\x00\x00\x00\x00,@', None, 26, datetime.datetime(2023, 8, 15, 18, 40), b'110e6103-e39b-51d4-a716-826755413099', 1]]

Binlogreader adds this offset ir-respective of the timezone value set in mysql_ch_replicator/__init__.py:

interestingly enough this ain't happening on my production server. The binlog reader isn't adding an extra offset while reading events from bin log files. The conversion happens only at insert time and that too only if the timezone is set to other than local one.

Edit: Looks like this is something which has to do with oracle's mysql. With mariadb this particular test passes just fine. Although some of the subsequent tests are failing which i am trying to fix one by one. That happened because the default mariadb server was set to UTC. Have changed the conf in tests dir and have set it to America/NewYork and test is failing again now.

edit 2:
the actual binlog file has correct time

#Q> INSERT INTO `test_table` (test1, test2, test4, test5, test6, test7) VALUES (1, POINT(15.0, 14.0), '2,4,5', '2023-08-15 14:40:00', '110e6103-e39b-51d4-a716-826755413099', 'point')
#260326  0:48:55 server id 1  end_log_pos 0 CRC32 0x47d9c0a2 	Table_map: `replication-test_db`.`test_table` mapped to number 25

which confirms that the problem is with mysql_ch_replicator's bin log reader as its adding an +4h offset