With this commit, we're introducing the Adaptive Executor.
The commit message consists of two distinct sections. The first part explains
how the executor works. The second part consists of the commit messages of
the individual smaller commits that resulted in this commit. The readers
can search for the each of the smaller commit messages on
https://github.com/citusdata/citus and can learn more about the history
of the change.
/*-------------------------------------------------------------------------
*
* adaptive_executor.c
*
* The adaptive executor executes a list of tasks (queries on shards) over
* a connection pool per worker node. The results of the queries, if any,
* are written to a tuple store.
*
* The concepts in the executor are modelled in a set of structs:
*
* - DistributedExecution:
* Execution of a Task list over a set of WorkerPools.
* - WorkerPool
* Pool of WorkerSessions for the same worker which opportunistically
* executes "unassigned" tasks from a queue.
* - WorkerSession:
* Connection to a worker that is used to execute "assigned" tasks
* from a queue and may execute unasssigned tasks from the WorkerPool.
* - ShardCommandExecution:
* Execution of a Task across a list of placements.
* - TaskPlacementExecution:
* Execution of a Task on a specific placement.
* Used in the WorkerPool and WorkerSession queues.
*
* Every connection pool (WorkerPool) and every connection (WorkerSession)
* have a queue of tasks that are ready to execute (readyTaskQueue) and a
* queue/set of pending tasks that may become ready later in the execution
* (pendingTaskQueue). The tasks are wrapped in a ShardCommandExecution,
* which keeps track of the state of execution and is referenced from a
* TaskPlacementExecution, which is the data structure that is actually
* added to the queues and describes the state of the execution of a task
* on a particular worker node.
*
* When the task list is part of a bigger distributed transaction, the
* shards that are accessed or modified by the task may have already been
* accessed earlier in the transaction. We need to make sure we use the
* same connection since it may hold relevant locks or have uncommitted
* writes. In that case we "assign" the task to a connection by adding
* it to the task queue of specific connection (in
* AssignTasksToConnections). Otherwise we consider the task unassigned
* and add it to the task queue of a worker pool, which means that it
* can be executed over any connection in the pool.
*
* A task may be executed on multiple placements in case of a reference
* table or a replicated distributed table. Depending on the type of
* task, it may not be ready to be executed on a worker node immediately.
* For instance, INSERTs on a reference table are executed serially across
* placements to avoid deadlocks when concurrent INSERTs take conflicting
* locks. At the beginning, only the "first" placement is ready to execute
* and therefore added to the readyTaskQueue in the pool or connection.
* The remaining placements are added to the pendingTaskQueue. Once
* execution on the first placement is done the second placement moves
* from pendingTaskQueue to readyTaskQueue. The same approach is used to
* fail over read-only tasks to another placement.
*
* Once all the tasks are added to a queue, the main loop in
* RunDistributedExecution repeatedly does the following:
*
* For each pool:
* - ManageWorkPool evaluates whether to open additional connections
* based on the number unassigned tasks that are ready to execute
* and the targetPoolSize of the execution.
*
* Poll all connections:
* - We use a WaitEventSet that contains all (non-failed) connections
* and is rebuilt whenever the set of active connections or any of
* their wait flags change.
*
* We almost always check for WL_SOCKET_READABLE because a session
* can emit notices at any time during execution, but it will only
* wake up WaitEventSetWait when there are actual bytes to read.
*
* We check for WL_SOCKET_WRITEABLE just after sending bytes in case
* there is not enough space in the TCP buffer. Since a socket is
* almost always writable we also use WL_SOCKET_WRITEABLE as a
* mechanism to wake up WaitEventSetWait for non-I/O events, e.g.
* when a task moves from pending to ready.
*
* For each connection that is ready:
* - ConnectionStateMachine handles connection establishment and failure
* as well as command execution via TransactionStateMachine.
*
* When a connection is ready to execute a new task, it first checks its
* own readyTaskQueue and otherwise takes a task from the worker pool's
* readyTaskQueue (on a first-come-first-serve basis).
*
* In cases where the tasks finish quickly (e.g. <1ms), a single
* connection will often be sufficient to finish all tasks. It is
* therefore not necessary that all connections are established
* successfully or open a transaction (which may be blocked by an
* intermediate pgbouncer in transaction pooling mode). It is therefore
* essential that we take a task from the queue only after opening a
* transaction block.
*
* When a command on a worker finishes or the connection is lost, we call
* PlacementExecutionDone, which then updates the state of the task
* based on whether we need to run it on other placements. When a
* connection fails or all connections to a worker fail, we also call
* PlacementExecutionDone for all queued tasks to try the next placement
* and, if necessary, mark shard placements as inactive. If a task fails
* to execute on all placements, the execution fails and the distributed
* transaction rolls back.
*
* For multi-row INSERTs, tasks are executed sequentially by
* SequentialRunDistributedExecution instead of in parallel, which allows
* a high degree of concurrency without high risk of deadlocks.
* Conversely, multi-row UPDATE/DELETE/DDL commands take aggressive locks
* which forbids concurrency, but allows parallelism without high risk
* of deadlocks. Note that this is unrelated to SEQUENTIAL_CONNECTION,
* which indicates that we should use at most one connection per node, but
* can run tasks in parallel across nodes. This is used when there are
* writes to a reference table that has foreign keys from a distributed
* table.
*
* Execution finishes when all tasks are done, the query errors out, or
* the user cancels the query.
*
*-------------------------------------------------------------------------
*/
All the commits involved here:
* Initial unified executor prototype
* Latest changes
* Fix rebase conflicts to master branch
* Add missing variable for assertion
* Ensure that master_modify_multiple_shards() returns the affectedTupleCount
* Adjust intermediate result sizes
The real-time executor uses COPY command to get the results
from the worker nodes. Unified executor avoids that which
results in less data transfer. Simply adjust the tests to lower
sizes.
* Force one connection per placement (or co-located placements) when requested
The existing executors (real-time and router) always open 1 connection per
placement when parallel execution is requested.
That might be useful under certain circumstances:
(a) User wants to utilize as much as CPUs on the workers per
distributed query
(b) User has a transaction block which involves COPY command
Also, lots of regression tests rely on this execution semantics.
So, we'd enable few of the tests with this change as well.
* For parameters to be resolved before using them
For the details, see PostgreSQL's copyParamList()
* Unified executor sorts the returning output
* Ensure that unified executor doesn't ignore sequential execution of DDLJob's
Certain DDL commands, mainly creating foreign keys to reference tables,
should be executed sequentially. Otherwise, we'd end up with a self
distributed deadlock.
To overcome this situaiton, we set a flag `DDLJob->executeSequentially`
and execute it sequentially. Note that we have to do this because
the command might not be called within a transaction block, and
we cannot call `SetLocalMultiShardModifyModeToSequential()`.
This fixes at least two test: multi_insert_select_on_conflit.sql and
multi_foreign_key.sql
Also, I wouldn't mind scattering local `targetPoolSize` variables within
the code. The reason is that we'll soon have a GUC (or a global
variable based on a GUC) that'd set the pool size. In that case, we'd
simply replace `targetPoolSize` with the global variables.
* Fix 2PC conditions for DDL tasks
* Improve closing connections that are not fully established in unified execution
* Support foreign keys to reference tables in unified executor
The idea for supporting foreign keys to reference tables is simple:
Keep track of the relation accesses within a transaction block.
- If a parallel access happens on a distributed table which
has a foreign key to a reference table, one cannot modify
the reference table in the same transaction. Otherwise,
we're very likely to end-up with a self-distributed deadlock.
- If an access to a reference table happens, and then a parallel
access to a distributed table (which has a fkey to the reference
table) happens, we switch to sequential mode.
Unified executor misses the function calls that marks the relation
accesses during the execution. Thus, simply add the necessary calls
and let the logic kick in.
* Make sure to close the failed connections after the execution
* Improve comments
* Fix savepoints in unified executor.
* Rebuild the WaitEventSet only when necessary
* Unclaim connections on all errors.
* Improve failure handling for unified executor
- Implement the notion of errorOnAnyFailure. This is similar to
Critical Connections that the connection managament APIs provide
- If the nodes inside a modifying transaction expand, activate 2PC
- Fix few bugs related to wait event sets
- Mark placement INACTIVE during the execution as much as possible
as opposed to we do in the COMMIT handler
- Fix few bugs related to scheduling next placement executions
- Improve decision on when to use 2PC
Improve the logic to start a transaction block for distributed transactions
- Make sure that only reference table modifications are always
executed with distributed transactions
- Make sure that stored procedures and functions are executed
with distributed transactions
* Move waitEventSet to DistributedExecution
This could also be local to RunDistributedExecution(), but in that case
we had to mark it as "volatile" to avoid PG_TRY()/PG_CATCH() issues, and
cast it to non-volatile when doing WaitEventSetFree(). We thought that
would make code a bit harder to read than making this non-local, so we
move it here. See comments for PG_TRY() in postgres/src/include/elog.h
and "man 3 siglongjmp" for more context.
* Fix multi_insert_select test outputs
Two things:
1) One complex transaction block is now supported. Simply update
the test output
2) Due to dynamic nature of the unified executor, the orders of
the errors coming from the shards might change (e.g., all of
the queries on the shards would fail, but which one appears
on the error message?). To fix that, we simply added it to
our shardId normalization tool which happens just before diff.
* Fix subeury_and_cte test
The error message is updated from:
failed to execute task
To:
more than one row returned by a subquery or an expression
which is a lot clearer to the user.
* Fix intermediate_results test outputs
Simply update the error message from:
could not receive query results
to
result "squares" does not exist
which makes a lot more sense.
* Fix multi_function_in_join test
The error messages update from:
Failed to execute task XXX
To:
function f(..) does not exist
* Fix multi_query_directory_cleanup test
The unified executor does not create any intermediate files.
* Fix with_transactions test
A test case that just started to work fine
* Fix multi_router_planner test outputs
The error message is update from:
Could not receive query results
To:
Relation does not exists
which is a lot more clearer for the users
* Fix multi_router_planner_fast_path test
The error message is update from:
Could not receive query results
To:
Relation does not exists
which is a lot more clearer for the users
* Fix isolation_copy_placement_vs_modification by disabling select_opens_transaction_block
* Fix ordering in isolation_multi_shard_modify_vs_all
* Add executor locks to unified executor
* Make sure to allocate enought WaitEvents
The previous code was missing the waitEvents for the latch and
postmaster death.
* Fix rebase conflicts for master rebase
* Make sure that TRUNCATE relies on unified executor
* Implement true sequential execution for multi-row INSERTS
Execute the individual tasks executed one by one. Note that this is different than
MultiShardConnectionType == SEQUENTIAL_CONNECTION case (e.g., sequential execution
mode). In that case, running the tasks across the nodes in parallel is acceptable
and implemented in that way.
However, the executions that are qualified here would perform poorly if the
tasks across the workers are executed in parallel. We currently qualify only
one class of distributed queries here, multi-row INSERTs. If we do not enforce
true sequential execution, concurrent multi-row upserts could easily form
a distributed deadlock when the upserts touch the same rows.
* Remove SESSION_LIFESPAN flag in unified_executor
* Apply failure test updates
We've changed the failure behaviour a bit, and also the error messages
that show up to the user. This PR covers majority of the updates.
* Unified executor honors citus.node_connection_timeout
With this commit, unified executor errors out if even
a single connection cannot be established within
citus.node_connection_timeout.
And, as a side effect this fixes failure_connection_establishment
test.
* Properly increment/decrement pool size variables
Before this commit, the idle and active connection
counts were not properly calculated.
* insert_select_executor goes through unified executor.
* Add missing file for task tracker
* Modify ExecuteTaskListExtended()'s signature
* Sort output of INSERT ... SELECT ... RETURNING
* Take partition locks correctly in unified executor
* Alternative implementation for force_max_query_parallelization
* Fix compile warnings in unified executor
* Fix style issues
* Decrement idleConnectionCount when idle connection is lost
* Always rebuild the wait event sets
In the previous implementation, on waitFlag changes, we were only
modifying the wait events. However, we've realized that it might
be an over optimization since (a) we couldn't see any performance
benefits (b) we see some errors on failures and because of (a)
we prefer to disable it now.
* Make sure to allocate enough sized waitEventSet
With multi-row INSERTs, we might have more sessions than
task*workerCount after few calls of RunDistributedExecution()
because the previous sessions would also be alive.
Instead, re-allocate events when the connectino set changes.
* Implement SELECT FOR UPDATE on reference tables
On master branch, we do two extra things on SELECT FOR UPDATE
queries on reference tables:
- Acquire executor locks
- Execute the query on all replicas
With this commit, we're implementing the same logic on the
new executor.
* SELECT FOR UPDATE opens transaction block even if SelectOpensTransactionBlock disabled
Otherwise, users would be very confused and their logic is very likely
to break.
* Fix build error
* Fix the newConnectionCount calculation in ManageWorkerPool
* Fix rebase conflicts
* Fix minor test output differences
* Fix citus indent
* Remove duplicate sorts that is added with rebase
* Create distributed table via executor
* Fix wait flags in CheckConnectionReady
* failure_savepoints output for unified executor.
* failure_vacuum output (pg 10) for unified executor.
* Fix WaitEventSetWait timeout in unified executor
* Stabilize failure_truncate test output
* Add an ORDER BY to multi_upsert
* Fix regression test outputs after rebase to master
* Add executor.c comment
* Rename executor.c to adaptive_executor.c
* Do not schedule tasks if the failed placement is not ready to execute
Before the commit, we were blindly scheduling the next placement executions
even if the failed placement is not on the ready queue. Now, we're ensuring
that if failed placement execution is on a failed pool or session where the
execution is on the pendingQueue, we do not schedule the next task. Because
the other placement execution should be already running.
* Implement a proper custom scan node for adaptive executor
- Switch between the executors, add GUC to set the pool size
- Add non-adaptive regression test suites
- Enable CIRCLE CI for non-adaptive tests
- Adjust test output files
* Add slow start interval to the executor
* Expose max_cached_connection_per_worker to user
* Do not start slow when there are cached connections
* Consider ExecutorSlowStartInterval in NextEventTimeout
* Fix memory issues with ReceiveResults().
* Disable executor via TaskExecutorType
* Make sure to execute the tests with the other executor
* Use task_executor_type to enable-disable adaptive executor
* Remove useless code
* Adjust the regression tests
* Add slow start regression test
* Rebase to master
* Fix test failures in adaptive executor.
* Rebase to master - 2
* Improve comments & debug messages
* Set force_max_query_parallelization in isolation_citus_dist_activity
* Force max parallelization for creating shards when asked to use exclusive connection.
* Adjust the default pool size
* Expand description of max_adaptive_executor_pool_size GUC
* Update warnings in FinishRemoteTransactionCommit()
* Improve session clean up at the end of execution
Explicitly list all the states that the execution might end,
otherwise warn.
* Remove MULTI_CONNECTION_WAIT_RETRY which is not used at all
* Add more ORDER BYs to multi_mx_partitioning
We used to error out if the join clause includes filters like
t1.a < t2.a even if other filter like t1.key = t2.key exists.
Recently we lifted that restriction in subquery planning by
not lifting that restriction and focusing on equivalance classes
provided by postgres.
This checkin forwards previously erroring out real-time queries
due to join clauses to subquery planner and let it handle the
join even if the query does not have a subquery.
We are now pushing down queries that do not have any
subqueries in it. Error message looked misleading, changed to a more descriptive one.
* Enabling physical planner for subquery pushdown changes
This commit applies the logic that exists in INSERT .. SELECT
planning to the subquery pushdown changes.
The main algorithm is followed as :
- pick an anchor relation (i.e., target relation)
- per each target shard interval
- add the target shard interval's shard range
as a restriction to the relations (if all relations
joined on the partition keys)
- Check whether the query is router plannable per
target shard interval.
- If router plannable, create a task
* Add union support within the JOINS
This commit adds support for UNION/UNION ALL subqueries that are
in the following form:
.... (Q1 UNION Q2 UNION ...) as union_query JOIN (QN) ...
In other words, we currently do NOT support the queries that are
in the following form where union query is not JOINed with
other relations/subqueries :
.... (Q1 UNION Q2 UNION ...) as union_query ....
* Subquery pushdown planner uses original query
With this commit, we change the input to the logical planner for
subquery pushdown. Before this commit, the planner was relying
on the query tree that is transformed by the postgresql planner.
After this commit, the planner uses the original query. The main
motivation behind this change is the simplify deparsing of
subqueries.
* Enable top level subquery join queries
This work enables
- Top level subquery joins
- Joins between subqueries and relations
- Joins involving more than 2 range table entries
A new regression test file is added to reflect enabled test cases
* Add top level union support
This commit adds support for UNION/UNION ALL subqueries that are
in the following form:
.... (Q1 UNION Q2 UNION ...) as union_query ....
In other words, Citus supports allow top level
unions being wrapped into aggregations queries
and/or simple projection queries that only selects
some fields from the lower level queries.
* Disallow subqueries without a relation in the range table list for subquery pushdown
This commit disallows subqueries without relation in the range table
list. This commit is only applied for subquery pushdown. In other words,
we do not add this limitation for single table re-partition subqueries.
The reasoning behind this limitation is that if we allow pushing down
such queries, the result would include (shardCount * expectedResults)
where in a non distributed world the result would be (expectedResult)
only.
* Disallow subqueries without a relation in the range table list for INSERT .. SELECT
This commit disallows subqueries without relation in the range table
list. This commit is only applied for INSERT.. SELECT queries.
The reasoning behind this limitation is that if we allow pushing down
such queries, the result would include (shardCount * expectedResults)
where in a non distributed world the result would be (expectedResult)
only.
* Change behaviour of subquery pushdown flag (#1315)
This commit changes the behaviour of the citus.subquery_pushdown flag.
Before this commit, the flag is used to enable subquery pushdown logic. But,
with this commit, that behaviour is enabled by default. In other words, the
flag is now useless. We prefer to keep the flag since we don't want to break
the backward compatibility. Also, we may consider using that flag for other
purposes in the next commits.
* Require subquery_pushdown when limit is used in subquery
Using limit in subqueries may cause returning incorrect
results. Therefore we allow limits in subqueries only
if user explicitly set subquery_pushdown flag.
* Evaluate expressions on the LIMIT clause (#1333)
Subquery pushdown uses orignal query, the LIMIT and OFFSET clauses
are not evaluated. However, logical optimizer expects these expressions
are already evaluated by the standard planner. This commit manually
evaluates the functions on the logical planner for subquery pushdown.
* Better format subquery regression tests (#1340)
* Style fix for subquery pushdown regression tests
With this commit we intented a more consistent style for the
regression tests we've added in the
- multi_subquery_union.sql
- multi_subquery_complex_queries.sql
- multi_subquery_behavioral_analytics.sql
* Enable the tests that are temporarily commented
This commit enables some of the regression tests that were commented
out until all the development is done.
* Fix merge conflicts (#1347)
- Update regression tests to meet the changes in the regression
test output.
- Replace Ifs with Asserts given that the check is already done
- Update shard pruning outputs
* Add view regression tests for increased subquery coverage (#1348)
- joins between views and tables
- joins between views
- union/union all queries involving views
- views with limit
- explain queries with view
* Improve btree operators for the subquery tests
This commit adds the missing comprasion for subquery composite key
btree comparator.
So far placements were assigned an Oid, but that was just used to track
insertion order. It also did so incompletely, as it was not preserved
across changes of the shard state. The behaviour around oid wraparound
was also not entirely as intended.
The newly introduced, explicitly assigned, IDs are preserved across
shard-state changes.
The prime goal of this change is not to improve ordering of task
assignment policies, but to make it easier to reference shards. The
newly introduced UpdateShardPlacementState() makes use of that, and so
will the in-progress connection and transaction management changes.
Before this change, count on a distributed returned NULL if all shards
were pruned away, because on the master we replace with count(..) call
with a sum(..) call to sum the counts from the shards. However, sum
returns NULL when there are no rows, whereas count is expected to return
0.
pg_toast_* oids are constantly changing, and this causes regression tests to
fail time to time. With this commit, we remove all of the pg_toast_* references
from regression test outputs.
Before this change, we do not check whether given table which already contains any data
in master_create_distributed_table command. If that table contains any data, making it
it distributed, makes that data hidden to user. With this change, we now gave error to
user if the table contains data.
Fixes#547
This change removes all references to \stage in the regression tests
and puts \COPY instead. Doing so changed shard counts, min/max
values on some test tables (lineitem, orders, etc.).
This adds support for SERIAL/BIGSERIAL column types. Because we now can
evaluate functions on the master (during execution), adding this is a
matter of ensuring the table creation step works properly.
To accomplish this, I've added some logic to detect sequences owned by
a table (i.e. those related to its columns). Simply creating a sequence
and using it in a default value is insufficient; users who do so must
ensure the sequence is owned by the column using it.
Fortunately, this is exactly what SERIAL and BIGSERIAL do, which is the
use case we're targeting with this feature. While testing this, I found
that worker_apply_shard_ddl_command actually adds shard identifiers to
sequence names, though I found no places that use or test this path. I
removed that code so that sequence names are not mutated and will match
those used by a SERIAL default value expression.
Our use of the new-to-9.5 CREATE SEQUENCE IF NOT EXISTS syntax means we
are dropping support for 9.4 (which is being done regardless, but makes
this change simpler). I've removed 9.4 from the Travis build matrix.
Some edge cases are possible in ALTER SEQUENCE, COPY FROM (on workers),
and CREATE SEQUENCE OWNED BY. I've added errors for each so that users
understand when and why certain operations are prohibited.
Fixes#676
We added old versions (i.e. without schema name) of worker_apply_shard_ddl_command,
worker_fetch_foreign_file and worker_fetch_regular_table back. During function call
of one of these functions, we set schema name as public schema and call the newer
version of the functions.
Fixes#565Fixes#626
To add schema support to citus, we need to schema-prefix all table names, object names etc.
in the queries sent to worker nodes. However; query deparsing is not available for most of
DDL commands, therefore it is not easy to generate worker query in the master node.
As a solution we are sending schema names along with shard id and query to run to worker
nodes with worker_apply_shard_ddl_command.
To not break \STAGE command we pass public schema as paramater while calling
worker_apply_shard_ddl_command from there. This will not cause problem if user uses \STAGE
in different schema because passes schema name is used only if there is no schema name is
given in the query.
Fixes#394
This change adds LIMIT/OFFSET support for non router-plannable
distributed queries.
In cases that we can push the LIMIT down, we add the OFFSET value to
that LIMIT in the worker queries. When a query with LIMIT x OFFSET y is issued,
the query is propagated to the workers as LIMIT (x+y) OFFSET 0, and on the
master table, the original LIMIT and OFFSET values are used. With this change,
we can use OFFSET wherever we can use LIMIT.
Fixes#271
This change sets ShardIds and JobIds for each test case. Before this change,
when a new test that somehow increments Job or Shard IDs is added, then
the tests after the new test should be updated.
ShardID and JobID sequences are set at the beginning of each file with the
following commands:
```
ALTER SEQUENCE pg_catalog.pg_dist_shardid_seq RESTART 290000;
ALTER SEQUENCE pg_catalog.pg_dist_jobid_seq RESTART 290000;
```
ShardIds and JobIds are multiples of 10000. Exceptions are:
- multi_large_shardid: shardid and jobid sequences are set to much larger values
- multi_fdw_large_shardid: same as above
- multi_join_pruning: Causes a race condition with multi_hash_pruning since
they are run in parallel.
This change removes the whitelisting check on the WHERE clauses. Note that, before
this change, citus was already allowing all types of nodes with the following
format (i.e., wrap with a boolean test):
* SELECT col FROM table WHERE (ANY EXPRESSION) is TRUE;
Thus, this change is mostly useful for allowing the expressions in the WHERE clause
directly and avoiding "unsupport clause type" errors.
Jelte Fennema
Önder Kalacı
velioglu
Murat Tuncer
Jason Petersen
Marco Slot
Murat Tuncer
Andres Freund
Metin Doslu
Burak Yucesoy
Eren Basak