ExecuteLocalTaskList doesn't need scanState as it only uses
paramListInfo, distributedPlan and tupleStoreState. It is better to pass
only the variables that the function needs, so that we can call this
function from other places when we dont have scanState.
We had many fields in task related to query strings. It was kind of
complex, and only of them could be set at a time. Therefore it makes
more sense to abstract this and use a union so that it is clear that
only of them should be set.
We have three fields that could have query related strings:
- queryForLocation
- queryStringLazy
- perPlacementQueryStrings
Relatively, they can be set with:
- SetTaskQueryString
- SetTaskQueryIfShouldLazyDeparse
- SetTaskPerPlacementQueryStrings
The direct usage of the query related fields are also removed.
Rename queryForLocalExecution
Currently queryForLocalExecution is only used for deparsing purposes,
therefore it makes sense to rename it to what it is doing.
TaskQueryStringForPlacement simplifies how the executor gets the query
string for a given placement. Task will use the necessary fields to
return the correct query placement string. Executor doesn't need to know
the details for this.
rename TaskQueryString as TaskQueryStringAllPlacements
TaskQueryString returns the query string that will be the same for all
the placements. In INSERT..SELECT the query string can be different for
each placement. Adaptive executor uses TaskQueryStringForPlacement,
which returns the query string for a placement. It makes sense to rename
TaskQueryString as TaskQueryStringAllPlacements as it is returning the
query string for all placements.
rename SetTaskQuery as SetTaskQueryIfShouldLazyDeparse
SetTaskQuery does not always sets the task query. It can set the query
string as well. So it is more clear to name it
SetTaskQueryIfShouldLazyDeparse, since it will set the query not query
string only when we should deparse the query in a lazy way.
It is possible that a task will have different query string for each
placement. This is the case in INSERT..SELECT via repartitioning. When
we are setting task->perPlacementQueryString, we should set
queryStringLazy to NULL. Therefore a method for that purpose is created.
Some refactoring:
Consolidate expression which decides whether GROUP BY/HAVING are pushed down
Rename early pullUpIntermediateRows to hasNonDistributableAggregates
Create WorkerColumnName to handle formatting WORKER_COLUMN_FORMAT
Ignore NULL StringInfo pointers to SafeToPushdownWindowFunction
Fix bug where SubqueryPushdownMultiNodeTree mutates supplied Query,
SafeToPushdownWindowFunction requires the original query as it relies on rtable
We can use local copy in INSERT..SELECT, so the check that disables
local execution is removed.
Also a test for local copy where the data size >
LOCAL_COPY_FLUSH_THRESHOLD is added.
use local execution with insert..select
If current transaction is connected to local group we should not use
local copy, because we might not see some of the changes that are made
over the connection to the local group.
A copy will be executed locally if
- Local execution is enabled and current transaction accessed a local placement
- Local execution is enabled and we are inside a transaction block.
So even if local execution is enabled but we are not in a transaction block, the copy will not be run locally.
This will not run locally:
```
COPY distributed_table FROM STDIN;
....
```
This will run locally:
```
SET citus.enable_local_execution to 'on';
BEGIN;
COPY distributed_table FROM STDIN;
COMMIT;
....
```
.
There are 3 ways to do a copy in postgres programmatically:
- from a file
- from a program
- from a callback function
I have chosen to implement it with a callback function, which means that we write the rows of copy from a callback function to the output buffer, which is used to insert tuples into the actual table.
For each shard id, we have a buffer that keeps the current rows to be written, we perform the actual copy operation either when:
- copy buffer for the given shard id reaches to a threshold, which is currently 512KB
- we reach to the end of the copy
The buffer size is debatable(512KB). At a given time, we might allocate (local placement * buffer size) memory at most.
The local copy uses the same copy format as remote copy, which means that we serialize the data in the same format as remote copy and send it locally.
There was also the option to use ExecSimpleRelationInsert to insert
slots one by one, which would avoid the extra
serialization/deserialization but doing some benchmarks it seems that
using buffers are significantly better in terms of the performance.
You can see this comment for more details: https://github.com/citusdata/citus/pull/3557#discussion_r389499054
* reimplement ExecuteUtilityTaskListWithoutResults for local utility command execution
* introduce new functions for local execution of utility commands
* change ErrorIfTransactionAccessedPlacementsLocally logic for local utility command execution
* enable local execution for TRUNCATE command on distributed & reference tables
* update existing tests for local utility command execution
* enable local execution for DDL commands on distributed & reference tables
* enable local execution for DROP command on distributed & reference tables
* add normalization rules for cascaded commands
* add new tests for local utility command execution
New stack memory can contain anything including passwords/private keys.
In these functions we return structs that can have their padding
bytes uninitialized. By first zeroing out the struct fully, we try to
ensure that any data that is in these padding bytes is at least
overwritten once. It might not be zero anymore after setting the fields,
but at least it shouldn't be private data anymore.
When ExecutorSlowStartInterval is set to 0, it has a special meaning
that we do not want to use slow start. Therefore, in the code we have
checks such as ExecutorSlowStartInterval > 0 to understand if it is
enabled or not. However, this is kind of subtle, and it creates an extra
mapping in our mind. Therefore, I thought that using a variable for the
special value removes the mapping and makes it easier to understand.
Semmle reported quite some places where we use a value that could be NULL. Most of these are not actually a real issue, but better to be on the safe side with these things and make the static analysis happy.
DESCRIPTION: Replace the query planner for the coordinator part with the postgres planner
Closes#2761
Citus had a simple rule based planner for the query executed on the query coordinator. This planner grew over time with the addigion of SQL support till it was getting close to the functionality of the postgres planner. Except the code was brittle and its complexity rose which made it hard to add new SQL support.
Given its resemblance with the postgres planner it was a long outstanding wish to replace our hand crafted planner with the well supported postgres planner. This patch replaces our planner with a call to postgres' planner.
Due to the functionality of the postgres planner we needed to support both projections and filters/quals on the citus custom scan node. When a sort operation is planned above the custom scan it might require fields to be reordered in the custom scan before returning the tuple (projection). The postgres planner assumes every custom scan node implements projections. Because we controlled the plan that was created we prevented reordering in the custom scan and never had implemented it before.
A same optimisation applies to having clauses that could have been where clauses. Instead of applying the filter as a having on the aggregate it will push it down into the plan which could reach a custom scan node.
For both filters and projections we have implemented them when tuples are read from the tuple store. If no projections or filters are required it will directly return the tuple from the tuple store. Otherwise it will loop tuples from the tuple store through the filter and projection until a tuple is found and returned.
Besides filters being pushed down a side effect of having quals that could have been a where clause is that a call to read intermediate result could be called before the first tuple is fetched from the custom scan. This failed because the intermediate result would only be pulled to the coordinator on the first tuple fetch. To overcome this problem we do run the distributed subplans now before we run the postgres executor. This ensures the intermediate result is present on the coordinator in time. We do account for total time instrumentation by removing the instrumentation before handing control to the psotgres executor and update the timings our self.
For future SQL support it is enough to create a valid query structure for the part of the query to be executed on the query coordinating node. As a utility we do serialise and print the query at debug level4 for engineers to inspect what kind of query is being planned on the query coordinator.
The root of the problem is that, standard_planner() converts the following qual
```
{OPEXPR
:opno 98
:opfuncid 67
:opresulttype 16
:opretset false
:opcollid 0
:inputcollid 100
:args (
{VAR
:varno 1
:varattno 1
:vartype 25
:vartypmod -1
:varcollid 100
:varlevelsup 0
:varnoold 1
:varoattno 1
:location 45
}
{CONST
:consttype 25
:consttypmod -1
:constcollid 100
:constlen -1
:constbyval false
:constisnull true
:location 51
:constvalue <>
}
)
:location 49
}
```
To
```
(
{CONST
:consttype 16
:consttypmod -1
:constcollid 0
:constlen 1
:constbyval true
:constisnull true
:location -1
:constvalue <>
}
)
```
So, Citus doesn't deal with NULL values in real-time or non-fast path router queries.
And, in the FastPathRouter planner, we check constisnull in DistKeyInSimpleOpExpression().
However, in deferred pruning case, we do not check for isnull for const.
Thus, the fix consists of two parts:
- Let PruneShards() not crash when NULL parameter is passed
- For deferred shard pruning in fast-path queries, explicitly check that we have CONST which is not NULL
Sometimes during errors workers will create files while we're deleting intermediate directories
example:
DEBUG: could not remove file "base/pgsql_job_cache/10_0_431": Directory not empty
DETAIL: WARNING from localhost:57637
Previously, the logic for evaluting the functions and the parameters
were the same. That ended-up evaluting the functions inaccurately
on the coordinator. Instead, split the function evaluation logic
from parameter evalution logic.
* Update shardPlacement->nodeId to uint
As the source of the shardPlacement->nodeId is always workerNode->nodeId,
and that is uint32.
We had this hack because of: 0ea4e52df5 (r266421409)
And, that is gone with: 90056f7d3c (diff-c532177d74c72d3f0e7cd10e448ab3c6L1123)
So, we're safe to do it now.
* Relax the restrictions on using the local execution
Previously, whenever any local execution happens, we disabled further
commands to do any remote queries. The basic motivation for doing that
is to prevent any accesses in the same transaction block to access the
same placements over multiple sessions: one is local session the other
is remote session to the same placement.
However, the current implementation does not distinguish local accesses
being to a placement or not. For example, we could have local accesses
that only touches intermediate results. In that case, we should not
implement the same restrictions as they become useless.
So, this is a pre-requisite for executing the intermediate result only
queries locally.
* Update the error messages
As the underlying implementation has changed, reflect it in the error
messages.
* Keep track of connections to local node
With this commit, we're adding infrastructure to track if any connection
to the same local host is done or not.
The main motivation for doing this is that we've previously were more
conservative about not choosing local execution. Simply, we disallowed
local execution if any connection to any remote node is done. However,
if we want to use local execution for intermediate result only queries,
this'd be annoying because we expect all queries to touch remote node
before the final query.
Note that this approach is still limiting in Citus MX case, but for now
we can ignore that.
* Formalize the concept of Local Node
Also some minor refactoring while creating the dummy placement
* Write intermediate results locally when the results are only needed locally
Before this commit, Citus used to always broadcast all the intermediate
results to remote nodes. However, it is possible to skip pushing
the results to remote nodes always.
There are two notable cases for doing that:
(a) When the query consists of only intermediate results
(b) When the query is a zero shard query
In both of the above cases, we don't need to access any data on the shards. So,
it is a valuable optimization to skip pushing the results to remote nodes.
The pattern mentioned in (a) is actually a common patterns that Citus users
use in practice. For example, if you have the following query:
WITH cte_1 AS (...), cte_2 AS (....), ... cte_n (...)
SELECT ... FROM cte_1 JOIN cte_2 .... JOIN cte_n ...;
The final query could be operating only on intermediate results. With this patch,
the intermediate results of the ctes are not unnecessarily pushed to remote
nodes.
* Add specific regression tests
As there are edge cases in Citus MX and with round-robin policy,
use the same queries on those cases as well.
* Fix failure tests
By forcing not to use local execution for intermediate results since
all the tests expects the results to be pushed remotely.
* Fix flaky test
* Apply code-review feedback
Mostly style changes
* Limit the max value of pg_dist_node_seq to reserve for internal use
adaptive_executor: sort includes, use foreach_ptr, remove lies from FinishDistributedExecution docs
connection_management: rename msecs, which isn't milliseconds
placement_connection: small typos
Deparsing and parsing a query can be heavy on CPU. When locally executing
the query we don't need to do this in theory most of the time.
This PR is the first step in allowing to skip deparsing and parsing
the query in these cases, by lazily creating the query string and
storing the query in the task. Future commits will make use of this and
not deparse and parse the query anymore, but use the one from the task
directly.
This is purely to enable better performance with prepared statements.
Before this commit, the fast path queries with prepared statements
where the distribution key includes a parameter always went through
distributed planning. After this change, we only go through distributed
planning on the first 5 executions.
We need to know which placement succeeded in executing the worker_partition_query_result() call. Otherwise we wouldn't know which node to fetch from. This change allows that by introducing Task::perPlacementQueryStrings.
Fixes#3331
In #2389, we've implemented support for partitioned tables with rep > 1.
The implementation is limiting the use of modification queries on the
partitions. In fact, we error out when any partition is modified via
EnsurePartitionTableNotReplicated().
However, we seem to forgot an important case, where the parent table's
partition is marked as INVALID. In that case, at least one of the partition
becomes INVALID. However, we do not mark partitions as INVALID ever.
If the user queries the partition table directly, Citus could happily send
the query to INVALID placements -- which are not marked as INVALID.
This PR fixes it by marking the placements of the partitions as INVALID
as well.
The shard placement repair logic already re-creates all the partitions,
so should be fine in that front.
Different versions of reindent tool reformatted citus_custom_scan.c
and citus_copyfuncs.c differently. So some developers spent some
extra attention not to commit these two files after reindent.
This PR tries to address this.
* WIP
* wip
* add basic logic to run a single job with repartioning joins with adaptive executor
* fix some warnings and return in ExecuteDependedTasks if there is none
* Add the logic to run depended jobs in adaptive executor
The execution of depended tasks logic is changed. With the current
logic:
- All tasks are created from the top level task list.
- At one iteration:
- CurTasks whose dependencies are executed are found.
- CurTasks are executed in parallel with adapter executor main
logic.
- The iteration is repeated until all tasks are completed.
* Separate adaptive executor repartioning logic
* Remove duplicate parts
* cleanup directories and schemas
* add basic repartion tests for adaptive executor
* Use the first placement to fetch data
In task tracker, when there are replicas, we try to fetch from a replica
for which a map task is succeeded. TaskExecution is used for this,
however TaskExecution is not used in adaptive executor. So we cannot use
the same thing as task tracker.
Since adaptive executor fails when a map task fails (There is no retry
logic yet). We know that if we try to execute a fetch task, all of its
map tasks already succeeded, so we can just use the first one to fetch
from.
* fix clean directories logic
* do not change the search path while creating a udf
* Enable repartition joins with adaptive executor with only enable_reparitition_joins guc
* Add comments to adaptive_executor_repartition
* dont run adaptive executor repartition test in paralle with other tests
* execute cleanup only in the top level execution
* do cleanup only in the top level ezecution
* not begin a transaction if repartition query is used
* use new connections for repartititon specific queries
New connections are opened to send repartition specific queries. The
opened connections will be closed at the FinishDistributedExecution.
While sending repartition queries no transaction is begun so that
we can see all changes.
* error if a modification was done prior to repartition execution
* not start a transaction if a repartition query and sql task, and clean temporary files and schemas at each subplan level
* fix cleanup logic
* update tests
* add missing function comments
* add test for transaction with DDL before repartition query
* do not close repartition connections in adaptive executor
* rollback instead of commit in repartition join test
* use close connection instead of shutdown connection
* remove unnecesary connection list, ensure schema owner before removing directory
* rename ExecuteTaskListRepartition
* put fetch query string in planner not executor as we currently support only replication factor = 1 with adaptive executor and repartition query and we know the query string in the planner phase in that case
* split adaptive executor repartition to DAG execution logic and repartition logic
* apply review items
* apply review items
* use an enum for remote transaction state and fix cleanup for repartition
* add outside transaction flag to find connections that are unclaimed instead of always opening a new transaction
* fix style
* wip
* rename removejobdir to partition cleanup
* do not close connections at the end of repartition queries
* do repartition cleanup in pg catch
* apply review items
* decide whether to use transaction or not at execution creation
* rename isOutsideTransaction and add missing comment
* not error in pg catch while doing cleanup
* use replication factor of the creation time, not current time to decide if task tracker should be chosen
* apply review items
* apply review items
* apply review item
DESCRIPTION: Fix counter that keeps track of internal depth in executor
While reviewing #3302 I ran into the `ExecutorLevel` variable which used a variable to keep the original value to restore on successful exit. I haven't explored the full space and if it is possible to get into an inconsistent state. However using `PG_TRY`/`PG_CATCH` seems generally more correct.
Given very bad things will happen if this level is not reset, I kept the failsafe of setting the variiable back to 0 on the `XactCallback` but I did add an assert to treat it as a developer bug.
Test ALTER ROLE doesn't deadlock when coordinator added, or propagate from mx workers
Consolidate wait_until_metadata_sync & verify_metadata to multi_test_helpers
In plain words, each distributed plan pulls the necessary intermediate
results to the worker nodes that the plan hits. This is primarily useful
in three ways.
(i) If the distributed plan that uses intermediate
result(s) is a router query, then the intermediate results are only
broadcasted to a single node.
(ii) If a distributed plan consists of only intermediate results, which
is not uncommon, the intermediate results are broadcasted to a single
node only.
(iii) If a distributed query hits a sub-set of the shards in multiple
workers, the intermediate results will be broadcasted to the relevant
node(s).
The final item (iii) becomes crucial for append/range distributed
tables where typically the distributed queries hit a small subset of
shards/workers.
To do this, for each query that Citus creates a distributed plan, we keep
track of the subPlans used in the queryTree, and save it in the distributed
plan. Just before Citus executes each subPlan, Citus first keeps track of
every worker node that the distributed plan hits, and marks every subPlan
should be broadcasted to these nodes. Later, for each subPlan which is a
distributed plan, Citus does this operation recursively since these
distributed plans may access to different subPlans, and those have to be
recorded as well.
When the user picks "round-robin" policy, the aim is that the load
is distributed across nodes. However, for reference tables on the
coordinator, since local execution kicks in immediately, round-robin
is ignored.
With this change, we're excluding the placement on the coordinator.
Although the approach seems a little bit invasive because of
modifications in the placement list, that sounds acceptable.
We could have done this in some other ways such as:
1) Add a field to "Task->roundRobinPlacement" (or such), which is
updated as the first element after RoundRobinPolicy is applied.
During the execution, if that placement is local to the coordinator,
skip it and try the other remote placements.
2) On TaskAccessesLocalNode()@local_execution.c, check
task_assignment_policy, if round-robin selected and there is local
placement on the coordinator, skip it. However, task assignment is done
on planning, but this decision is happening on the execution, which
could create weird edge cases.
It looks like the logic to prevent RETURNING in reference tables to
have duplicate entries that comes from local and remote executions
leads to missing some tuples for distributed tables.
With this PR, we're ensuring to kick in the logic for reference tables
only.
* Remove unused executor codes
All of the codes of real-time executor. Some functions
in router executor still remains there because there
are common functions. We'll move them to accurate places
in the follow-up commits.
* Move GUCs to transaction mngnt and remove unused struct
* Update test output
* Get rid of references of real-time executor from code
* Warn if real-time executor is picked
* Remove lots of unused connection codes
* Removed unused code for connection restrictions
Real-time and router executors cannot handle re-using of the existing
connections within a transaction block.
Adaptive executor and COPY can re-use the connections. So, there is no
reason to keep the code around for applying the restrictions in the
placement connection logic.
When citus.enable_repartition_joins guc is set to on, and we have
adaptive executor, there was a typo in the debug message, which was
saying realtime executor no adaptive executor.
See #3125 for details on each item.
* Remove real-time/router executor tests-1
These are the ones which doesn't have '_%d' in the test
output files.
* Remove real-time/router executor tests-2
These are the ones which has in the test
output files.
* Move the tests outputs to correct place
* Make sure that single shard commits use 2PC on adaptive executor
It looks like we've messed the tests in #2891. Fixing back.
* Use adaptive executor for all router queries
This becomes important because when task-tracker is picked, we
used to pick router executor, which doesn't make sense.
* Remove explicit references to real-time/router executors in the tests
* JobExecutorType never picks real-time/router executors
* Make sure to go incremental in test output numbers
* Even users cannot pick real-time anymore
* Do not use real-time/router custom scans
* Get rid of unnecessary normalizations
* Reflect unneeded normalizations
* Get rid of unnecessary test output file
It turns out that TupleDescGetAttInMetadata() allocates quite a lot
of memory. And, if the target list is long and there are too many rows
returning, the leak becomes appereant.
You can reproduce the issue wout the fix with the following commands:
```SQL
CREATE TABLE users_table (user_id int, time timestamp, value_1 int, value_2 int, value_3 float, value_4 bigint);
SELECT create_distributed_table('users_table', 'user_id');
insert into users_table SELECT i, now(), i, i, i, i FROM generate_series(0,99999)i;
-- load faster
-- 200,000
INSERT INTO users_table SELECT * FROM users_table;
-- 400,000
INSERT INTO users_table SELECT * FROM users_table;
-- 800,000
INSERT INTO users_table SELECT * FROM users_table;
-- 1,600,000
INSERT INTO users_table SELECT * FROM users_table;
-- 3,200,000
INSERT INTO users_table SELECT * FROM users_table;
-- 6,400,000
INSERT INTO users_table SELECT * FROM users_table;
-- 12,800,000
INSERT INTO users_table SELECT * FROM users_table;
-- making the target list entry wider speeds up the leak to show up
select *,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,* FROM users_table ;
```
Areas for further optimization:
- Don't save subquery results to a local file on the coordinator when the subquery is not in the having clause
- Push the the HAVING with subquery to the workers if there's a group by on the distribution column
- Don't push down the results to the workers when we don't push down the HAVING clause, only the coordinator needs it
Fixes#520Fixes#756Closes#2047
When a function is marked as colocated with a distributed table,
we try delegating queries of kind "SELECT func(...)" to workers.
We currently only support this simple form, and don't delegate
forms like "SELECT f1(...), f2(...)", "SELECT f1(...) FROM ...",
or function calls inside transactions.
As a side effect, we also fix the transactional semantics of DO blocks.
Previously we didn't consider a DO block a multi-statement transaction.
Now we do.
Co-authored-by: Marco Slot <marco@citusdata.com>
Co-authored-by: serprex <serprex@users.noreply.github.com>
Co-authored-by: pykello <hadi.moshayedi@microsoft.com>
DESCRIPTION: Distribute Types to worker nodes
When to propagate
==============
There are two logical moments that types could be distributed to the worker nodes
- When they get used ( just in time distribution )
- When they get created ( proactive distribution )
The just in time distribution follows the model used by how schema's get created right before we are going to create a table in that schema, for types this would be when the table uses a type as its column.
The proactive distribution is suitable for situations where it is benificial to have the type on the worker nodes directly. They can later on be used in queries where an intermediate result gets created with a cast to this type.
Just in time creation is always the last resort, you cannot create a distributed table before the type gets created. A good example use case is; you have an existing postgres server that needs to scale out. By adding the citus extension, add some nodes to the cluster, and distribute the table. The type got created before citus existed. There was no moment where citus could have propagated the creation of a type.
Proactive is almost always a good option. Types are not resource intensive objects, there is no performance overhead of having 100's of types. If you want to use them in a query to represent an intermediate result (which happens in our test suite) they just work.
There is however a moment when proactive type distribution is not beneficial; in transactions where the type is used in a distributed table.
Lets assume the following transaction:
```sql
BEGIN;
CREATE TYPE tt1 AS (a int, b int);
CREATE TABLE t1 AS (a int PRIMARY KEY, b tt1);
SELECT create_distributed_table('t1', 'a');
\copy t1 FROM bigdata.csv
```
Types are node scoped objects; meaning the type exists once per worker. Shards however have best performance when they are created over their own connection. For the type to be visible on all connections it needs to be created and committed before we try to create the shards. Here the just in time situation is most beneficial and follows how we create schema's on the workers. Outside of a transaction block we will just use 1 connection to propagate the creation.
How propagation works
=================
Just in time
-----------
Just in time propagation hooks into the infrastructure introduced in #2882. It adds types as a supported object in `SupportedDependencyByCitus`. This will make sure that any object being distributed by citus that depends on types will now cascade into types. When types are depending them self on other objects they will get created first.
Creation later works by getting the ddl commands to create the object by its `ObjectAddress` in `GetDependencyCreateDDLCommands` which will dispatch types to `CreateTypeDDLCommandsIdempotent`.
For the correct walking of the graph we follow array types, when later asked for the ddl commands for array types we return `NIL` (empty list) which makes that the object will not be recorded as distributed, (its an internal type, dependant on the user type).
Proactive distribution
---------------------
When the user creates a type (composite or enum) we will have a hook running in `multi_ProcessUtility` after the command has been applied locally. Running after running locally makes that we already have an `ObjectAddress` for the type. This is required to mark the type as being distributed.
Keeping the type up to date
====================
For types that are recorded in `pg_dist_object` (eg. `IsObjectDistributed` returns true for the `ObjectAddress`) we will intercept the utility commands that alter the type.
- `AlterTableStmt` with `relkind` set to `OBJECT_TYPE` encapsulate changes to the fields of a composite type.
- `DropStmt` with removeType set to `OBJECT_TYPE` encapsulate `DROP TYPE`.
- `AlterEnumStmt` encapsulates changes to enum values.
Enum types can not be changed transactionally. When the execution on a worker fails a warning will be shown to the user the propagation was incomplete due to worker communication failure. An idempotent command is shown for the user to re-execute when the worker communication is fixed.
Keeping types up to date is done via the executor. Before the statement is executed locally we create a plan on how to apply it on the workers. This plan is executed after we have applied the statement locally.
All changes to types need to be done in the same transaction for types that have already been distributed and will fail with an error if parallel queries have already been executed in the same transaction. Much like foreign keys to reference tables.
/*
* local_executor.c
*
* The scope of the local execution is locally executing the queries on the
* shards. In other words, local execution does not deal with any local tables
* that are not shards on the node that the query is being executed. In that sense,
* the local executor is only triggered if the node has both the metadata and the
* shards (e.g., only Citus MX worker nodes).
*
* The goal of the local execution is to skip the unnecessary network round-trip
* happening on the node itself. Instead, identify the locally executable tasks and
* simply call PostgreSQL's planner and executor.
*
* The local executor is an extension of the adaptive executor. So, the executor uses
* adaptive executor's custom scan nodes.
*
* One thing to note that Citus MX is only supported with replication factor = 1, so
* keep that in mind while continuing the comments below.
*
* On the high level, there are 3 slightly different ways of utilizing local execution:
*
* (1) Execution of local single shard queries of a distributed table
*
* This is the simplest case. The executor kicks at the start of the adaptive
* executor, and since the query is only a single task the execution finishes
* without going to the network at all.
*
* Even if there is a transaction block (or recursively planned CTEs), as long
* as the queries hit the shards on the same, the local execution will kick in.
*
* (2) Execution of local single queries and remote multi-shard queries
*
* The rule is simple. If a transaction block starts with a local query execution,
* all the other queries in the same transaction block that touch any local shard
* have to use the local execution. Although this sounds restrictive, we prefer to
* implement in this way, otherwise we'd end-up with as complex scenarious as we
* have in the connection managements due to foreign keys.
*
* See the following example:
* BEGIN;
* -- assume that the query is executed locally
* SELECT count(*) FROM test WHERE key = 1;
*
* -- at this point, all the shards that reside on the
* -- node is executed locally one-by-one. After those finishes
* -- the remaining tasks are handled by adaptive executor
* SELECT count(*) FROM test;
*
*
* (3) Modifications of reference tables
*
* Modifications to reference tables have to be executed on all nodes. So, after the
* local execution, the adaptive executor keeps continuing the execution on the other
* nodes.
*
* Note that for read-only queries, after the local execution, there is no need to
* kick in adaptive executor.
*
* There are also few limitations/trade-offs that is worth mentioning. First, the
* local execution on multiple shards might be slow because the execution has to
* happen one task at a time (e.g., no parallelism). Second, if a transaction
* block/CTE starts with a multi-shard command, we do not use local query execution
* since local execution is sequential. Basically, we do not want to lose parallelism
* across local tasks by switching to local execution. Third, the local execution
* currently only supports queries. In other words, any utility commands like TRUNCATE,
* fails if the command is executed after a local execution inside a transaction block.
* Forth, the local execution cannot be mixed with the executors other than adaptive,
* namely task-tracker, real-time and router executors. Finally, related with the
* previous item, COPY command cannot be mixed with local execution in a transaction.
* The implication of that any part of INSERT..SELECT via coordinator cannot happen
* via the local execution.
*/
Before this patch, when a connection is lost, we'd have the following
situation:
- Pop a task execution from readyQueue
- Lost connection
- Fail the session/pool. -> This step was not acting properly
because we've popped the task, but not set to session->currentTask
yet
After the patch:
- Pop a task execution from readyQueue
- Immediately set it to session->currentTask
- Lost connection
- Fail the session/pool. -> At this step, failing the
session would trigger query failures (or failovers)
properly.
* Add tuplestore helpers
* More detailed error messages in tuplestore
* Add CreateTupleDescCopy to SetupTuplestore
* Use new SetupTuplestore helper function
* Remove unnecessary copy
* Remove comment about undefined behaviour
This causes no behaviorial changes, only organizes better to implement modifying CTEs
Also rename ExtactInsertRangeTableEntry to ExtractResultRelationRTE,
as the source of this function didn't match the documentation
Remove Task's upsertQuery in favor of ROW_MODIFY_NONCOMMUTATIVE
Split up AcquireExecutorShardLock into more internal functions
Tests: Normalize multi_reference_table multi_create_table_constraints
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
Instead of scattering the code around, we move all the
logic into a single function.
This will help supporting foreign keys to reference tables
in the unified executor with a single line of change, just
calling this function.
The feature is only intended for getting consistent outputs for the regression tests.
RETURNING does not have any ordering gurantees and with unified executor, the ordering
of query executions on the shards are also becoming unpredictable. Thus, we're enforcing
ordering when a GUC is set.
We implicitly add an `ORDER BY` something equivalent of
`
RETURNING expr1, expr2, .. ,exprN
ORDER BY expr1, expr2, .. ,exprN
`
As described in the code comments as well, this is probably not the most
performant approach we could implement. However, since we're only
targeting regression tests, I don't see any issues with that. If we
decide to expand this to a feature to users, we should revisit the
implementation and improve the performance.
Before this commit, Citus supported INSERT...SELECT queries with
ON CONFLICT or RETURNING clauses only for pushdownable ones, since
queries supported via coordinator were utilizing COPY infrastructure
of PG to send selected tuples to the target worker nodes.
After this PR, INSERT...SELECT queries with ON CONFLICT or RETURNING
clauses will be performed in two phases via coordinator. In the first
phase selected tuples will be saved to the intermediate table which
is colocated with target table of the INSERT...SELECT query. Note that,
a utility function to save results to the colocated intermediate result
also implemented as a part of this commit. In the second phase, INSERT..
SELECT query is directly run on the worker node using the intermediate
table as the source table.
The file handling the utility functions (DDL) for citus organically grew over time and became unreasonably large. This refactor takes that file and refactored the functionality into separate files per command. Initially modeled after the directory and file layout that can be found in postgres.
Although the size of the change is quite big there are barely any code changes. Only one two functions have been added for readability purposes:
- PostProcessIndexStmt which is extracted from PostProcessUtility
- PostProcessAlterTableStmt which is extracted from multi_ProcessUtility
A README.md has been added to `src/backend/distributed/commands` describing the contents of the module and every file in the module.
We need more documentation around the overloading of the COPY command, for now the boilerplate has been added for people with better knowledge to fill out.
Assign the distributed transaction id before trying to acquire the
executor advisory locks. This is useful to show this backend in citus
lock graphs (e.g., dump_global_wait_edges() and citus_lock_waits).
We used to set the execution mode in the truncate trigger. However,
when multiple tables are truncated with a single command, we could
set the execution mode very late. Instead, now set the execution mode
on the utility hook.
With this commit, we all partitioned distributed tables with
replication factor > 1. However, we also have many restrictions.
In summary, we disallow all kinds of modifications (including DDLs)
on the partition tables. Instead, the user is allowed to run the
modifications over the parent table.
The necessity for such a restriction have two aspects:
- We need to acquire shard resource locks appropriately
- We need to handle marking partitions INVALID in case
of any failures. Note that, in theory, the parent table
should also become INVALID, which is too aggressive.
We acquire distributed lock on all mx nodes for truncated
tables before actually doing truncate operation.
This is needed for distributed serialization of the truncate
command without causing a deadlock.
Reason for the failure is that PG11 introduced a new relation kind
RELKIND_PARTITIONED_INDEX to be used for partitioned indices.
We expanded our check to cover that case.
Make sure that the coordinator sends the commands when the search
path synchronised with the coordinator's search_path. This is only
important when Citus sends the commands that are directly relayed
to the worker nodes. For example, the deparsed DLL commands or
queries always adds schema qualifications to the queries. So, they
do not require this change.
SaitTalhaNisanci
Philip Dubé
Onur Tirtir
Jelte Fennema
Marco Slot
Philip Dubé
Nils Dijk
Onder Kalaci
Hadi Moshayedi
Önder Kalacı
Hanefi Onaldi
Marco Slot
Hadi Moshayedi
Jason Petersen
Jason Petersen
velioglu
Murat Tuncer
mehmet furkan şahin