This will fix race with DatabaseCatalog::loadMarkedAsDroppedTables(),
since MaterializeMySQL, and MaterializedMySQLSyncThread in background,
will be started earlier then
DatabaseCatalog::loadMarkedAsDroppedTables() and will move those tables
to metadata_dropped, and after loadMarkedAsDroppedTables() will start
and try to load partially dropped tables and will hit UUID overlap:
12:02:51.536783 [ 3026034 ] {} <Information> Application: starting up
12:02:53.019282 [ 3026034 ] {} <Information> DatabaseMaterializeMySQL<Atomic> (mysql): Total 9 tables and 0 dictionaries.
12:02:53.041699 [ 3026200 ] {} <Debug> mysql.data (7143b65f-6982-4600-b143-b65f6982e600): Loading data parts
12:02:53.041740 [ 3026200 ] {} <Debug> mysql.data (7143b65f-6982-4600-b143-b65f6982e600): There are no data parts
12:02:53.620382 [ 3026034 ] {} <Information> DatabaseMaterializeMySQL<Atomic> (mysql): Starting up tables.
12:03:00.669730 [ 3026183 ] {} <Debug> executeQuery: (internal) /*Materialize MySQL step 1: execute MySQL DDL for dump data*/ DROP TABLE mysql.data
12:03:00.741894 [ 3026269 ] {} <Information> DatabaseCatalog: Trying load partially dropped table mysql.data (7143b65f-6982-4600-b143-b65f6982e600) from /var/lib/clickhouse/metadata_dropped/mysql.data.7143b65f-6982-4600-b143-b65f6982e600.sql
12:03:00.742582 [ 3026269 ] {} <Debug> mysql.data (7143b65f-6982-4600-b143-b65f6982e600): Loading data parts
12:03:00.742650 [ 3026269 ] {} <Debug> mysql.data (7143b65f-6982-4600-b143-b65f6982e600): There are no data parts
12:03:00.773137 [ 3026034 ] {} <Error> Application: Caught exception while loading metadata: Code: 57, e.displayText() = DB::Exception: Mapping for table with UUID=7143b65f-6982-4600-b143-b65f6982e600 already exists. It happened due to UUID collision, most likely because some not random UUIDs were manually specified in CREATE queries., Stack trace (when copying this message, always include the lines below):
12:03:01.224557 [ 3026034 ] {} <Error> Application: DB::Exception: Mapping for table with UUID=7143b65f-6982-4600-b143-b65f6982e600 already exists. It happened due to UUID collision, most likely because some not random UUIDs were manually specified in CREATE queries.
Cc: @zhang2014
Which logs all the info about LogIn, LogOut and LogIn Failure events.
Additional info that is logged:
- User name
- event type (LogIn, LogOut, LoginFailure)
- Event date\time\time with microseconds
- authentication type (same as for IDENTIFIED BY of CREATE USER statement)
- array of active settings profiles upon login
- array of active roles upon login
- array of changed settings with corresponding values
- client address and port
- interface (TCP\HTTP\MySQL\PostgreSQL, etc.)
- client info (name, version info)
- optional LoginFailure reason text message.
Added some tests to verify that events are properly saved with all necessary info via following interfaces:
- TCP
- HTTP
- MySQL
Known limitations
- Not tested against named HTTP sessions, PostgreSQL and gRPC, hence those are not guaranteed to work 100% properly.
* Add a codec Encrypted() for encrypting columns on disk
While this is implemented as a compression codec, it does not actually compress data. It instead encrypts data on disk. The key is obtained by executing a user-specified command at the server startup, or if it's not specified the codec refuses to process any data. For now the only supported cipher is 'AES-128-GCM-SIV'.
- Uses a small assembly file to include binary resources, rather than
objcopy
- Updates `base/common/getResource.cpp` for this new method of inclusion
- Removes linux-only guards in CMake files, as this solution is
cross-platform.
The resulting binary resources are available in the ClickHouse server
binary on Linux, macOS, and illumos platforms. FreeBSD has not been
tested, but will likely work as well.
TODO (suggested by Nikolai)
1. Build query plan fro current query (inside storage::read) up to WithMergableState
2. Check, that plan is simple enough: Aggregating - Expression - Filter - ReadFromStorage (or simplier)
3. Check, that filter is the same as filter in projection, and also expression calculates the same aggregation keys as in projection
4. Return WithMergableState if projection applies
3 will be easier to do with ActionsDAG, cause it sees all functions, and dependencies are direct (but it is possible with ExpressionActions also)
Also need to figure out how prewhere works for projections, and
row_filter_policies.
wip
* master: (694 commits)
Fix integration test test_storage_kafka failed error
Fix test 00163_column_oriented_formats failed error
Read ORC file by stripe to reduce memory cost
Function toDateTime decimal overflow ubsan fix
Revert "[RFC] Fix memory tracking with min_bytes_to_use_mmap_io"
Zlib use attribute constructor for functable initialization
Translate to Russian (clickhouse-client documentation)
Simple key dictionary primary key wrong order fix
Disable hedged requests
Added integration test
Revert "Function `arrayFold` for folding over array with accumulator"
Fix documentation for the GRANT command.
Added system query reload model
Make function `unhex` case insensitive for compatibility
Improve documentation for CREATE ROW POLICY command #2.
Add exception message
Fix tidy
Fix waiting for all connections closed on shutdown.
Disable postgresql_port in perf tests
Mark 01605_adaptive_granularity_block_borders as long
...