Comment from the code:
/*
* Iterate until all the tasks are finished. Once all the tasks
* are finished, ensure that that all the connection initializations
* are also finished. Otherwise, those connections are terminated
* abruptly before they are established (or failed). Instead, we let
* the ConnectionStateMachine() to properly handle them.
*
* Note that we could have the connections that are not established
* as a side effect of slow-start algorithm. At the time the algorithm
* decides to establish new connections, the execution might have tasks
* to finish. But, the execution might finish before the new connections
* are established.
*/
Note that the abruptly terminated connections lead to the following errors:
2020-11-16 21:09:09.800 CET [16633] LOG: could not accept SSL connection: Connection reset by peer
2020-11-16 21:09:09.872 CET [16657] LOG: could not accept SSL connection: Undefined error: 0
2020-11-16 21:09:09.894 CET [16667] LOG: could not accept SSL connection: Connection reset by peer
To easily reproduce the issue:
- Create a single node Citus
- Add the coordinator to the metadata
- Create a distributed table with shards on the coordinator
- f.sql: select count(*) from test;
- pgbench -f /tmp/f.sql postgres -T 12 -c 40 -P 1 or pgbench -f /tmp/f.sql postgres -T 12 -c 40 -P 1 -C
With this commit, the executor becomes smarter about refrain to open
new connections. The very basic example is that, if the connection
establishments take 1000ms and task executions as 5 msecs, the executor
becomes smart enough to not establish new connections.
With https://github.com/citusdata/citus/pull/4806 we enabled
2PC for any non-read-only local task. However, if the execution
is a single task, enabling 2PC (CoordinatedTransactionShouldUse2PC)
hits an assertion as we are not in a coordinated transaction.
There is no downside of using a coordinated transaction for single
task local queries.
Because setting the flag doesn't necessarily mean that we'll
use 2PC. If connections are read-only, we will not use 2PC.
In other words, we'll use 2PC only for connections that modified
any placements.
Before this commit, Citus used 2PC no matter what kind of
local query execution happens.
For example, if the coordinator has shards (and the workers as well),
even a simple SELECT query could start 2PC:
```SQL
WITH cte_1 AS (SELECT * FROM test LIMIT 10) SELECT count(*) FROM cte_1;
```
In this query, the local execution of the shards (and also intermediate
result reads) triggers the 2PC.
To prevent that, Citus now distinguishes local reads and local writes.
And, Citus switches to 2PC only if a modification happens. This may
still lead to unnecessary 2PCs when there is a local modification
and remote SELECTs only. Though, we handle that separately
via #4587.
When we use PROCESS_UTILITY_TOPLEVEL it causes some problems when
combined with other extensions such as pg_audit. With this commit we use
PROCESS_UTILITY_QUERY in the codebase to fix those problems.
When COPY is used for copying into co-located files, it was
not allowed to use local execution. The primary reason was
Citus treating co-located intermediate results as co-located
shards, and COPY into the distributed table was done via
"format result". And, local execution of such COPY commands
was not implemented.
With this change, we implement support for local execution with
"format result". To do that, we use the buffer for every file
on shardState->copyOutState, similar to how local copy on
shards are implemented. In fact, the logic is similar to
local copy on shards, but instead of writing to the shards,
Citus writes the results to a file.
The logic relies on LOCAL_COPY_FLUSH_THRESHOLD, and flushes
only when the size exceeds the threshold. But, unlike local
copy on shards, in this case we write the headers and footers
just once.
* Make undistribute_table() and citus_create_local_table() work with columnar
* Rename and use LocallyExecuteUtilityTask for UDF check
* Remove 'local' references in ExecuteUtilityCommand
As described in the comment, we have observed crashes in production
due to a segfault caused by the dereference of a NULL pointer in our
connection statemachine.
As a mitigation, preventing system crashes, we provide an error with
a small explanation of the issue. Unfortunately the case is not
reliably reproduced yet, hence the inability to add tests.
DESCRIPTION: Prevent segfaults when SAVEPOINT handling cannot recover from connection failures
Baseinfo also has pushed down filters etc, so it makes more sense to use
BaseRestrictInfo to determine what columns have constant equality
filters.
Also RteIdentity is used for removing conversion candidates instead of
rteIndex.
It seems that most of the updates were broken, we weren't aware of it
because there wasn't any data in the tables. They are broken mostly
because local tables do not have a shard id and some code paths should
be updated with that information, currently when there is an invalid
shard id, it is assumed to be pruned.
Consider local tables in router planner
In case there is a local table, the shard id will not be valid and there
are some checks that rely on shard id, we should skip these in case of
local tables, which is handled with a dummy placement.
Add citus local table dist table join tests
add local-dist table mixed joins tests
When Citus needs to parallelize queries on the local node (e.g., the node
executing the distributed query and the shards are the same), we need to
be mindful about the connection management. The reason is that the client
backends that are running distributed queries are competing with the client
backends that Citus initiates to parallelize the queries in order to get
a slot on the max_connections.
In that regard, we implemented a "failover" mechanism where if the distributed
queries cannot get a connection, the execution failovers the tasks to the local
execution.
The failover logic is follows:
- As the connection manager if it is OK to get a connection
- If yes, we are good.
- If no, we fail the workerPool and the failure triggers
the failover of the tasks to local execution queue
The decision of getting a connection is follows:
/*
* For local nodes, solely relying on citus.max_shared_pool_size or
* max_connections might not be sufficient. The former gives us
* a preview of the future (e.g., we let the new connections to establish,
* but they are not established yet). The latter gives us the close to
* precise view of the past (e.g., the active number of client backends).
*
* Overall, we want to limit both of the metrics. The former limit typically
* kics in under regular loads, where the load of the database increases in
* a reasonable pace. The latter limit typically kicks in when the database
* is issued lots of concurrent sessions at the same time, such as benchmarks.
*/
Multi-row execution already uses sequential execution. When shards
are local, using local execution is profitable as it avoids
an extra connection establishment to the local node.
* Move local execution after the remote execution
Before this commit, when both local and remote tasks
exist, the executor was starting the execution with
local execution. There is no strict requirements on
this.
Especially considering the adaptive connection management
improvements that we plan to roll soon, moving the local
execution after to the remote execution makes more sense.
The adaptive connection management for single node Citus
would look roughly as follows:
- Try to connect back to the coordinator for running
parallel queries.
- If succeeds, go on and execute tasks in parallel
- If fails, fallback to the local execution
So, we'll use local execution as a fallback mechanism. And,
moving it after to the remote execution allows us to implement
such further scenarios.
The adaptive executor emulates the TCP's slow start algorithm.
Whenever the executor needs new connections, it doubles the number
of connections established in the previous iteration.
This approach is powerful. When the remote queries are very short
(like index lookup with < 1ms), even a single connection is sufficent
most of the time. When the remote queries are long, the executor
can quickly establish necessary number of connections.
One missing piece on our implementation seems that the executor
keeps doubling the number of connections even if the previous
connection attempts have been finalized. Instead, we should
wait until all the attempts are finalized. This is how TCP's
slow-start works. Plus, it decreases the unnecessary pressure
on the remote nodes.
Before this commit, we let AdaptiveExecutorPreExecutorRun()
to be effective multiple times on every FETCH on cursors.
That does not affect the correctness of the query results,
but adds significant overhead.
With this commit, we make sure that local execution adds the
intermediate result size as the distributed execution adds. Plus,
it enforces the citus.max_intermediate_result_size value.
We should not access CurrentLocalExecutionStatus directly because that
would mean that we could also set it directly, which we shouldn't
because we have checks to see if the new state is possible, otherwise we
error.
After the connection timeout, we fail the session/pool. However, the
underlying connection can still be trying to connect. That is dangerous
because the new placement executions have already been in place. The
executor cannot handle the situation where multiple of
EXECUTION_ORDER_ANY task executions succeeds.
Adding a regression test doesn't seem easily doable. To reproduce the issue
- Add 2 worker nodes
- create a reference table
- set citus.node_connection_timeout to 1ms (requires code change)
- Continiously execute `SELECT count(*) FROM ref_table`
- Sometime later, you hit an out-of-array access in
`ScheduleNextPlacementExecution()` hence crashing.
- The reason for that is sometimes the first connection
successfully established while the executor is already
trying to execute the query on the second node.
Introduce table entry utility functions
Citus table cache entry utilities are introduced so that we can easily
extend existing functionality with minimum changes, specifically changes
to these functions. For example IsNonDistributedTableCacheEntry can be
extended for citus local tables without the need to scan the whole
codebase and update each relevant part.
* Introduce utility functions to find the type of tables
A table type can be a reference table, a hash/range/append distributed
table. Utility methods are created so that we don't have to worry about
how a table is considered as a reference table etc. This also makes it
easy to extend the table types.
* Add IsCitusTableType utilities
* Rename IsCacheEntryCitusTableType -> IsCitusTableTypeCacheEntry
* Change citus table types in some checks
CMDTAG_SELECT exists in PG12 hence defining a MACRO such as
CMDTAG_SELECT -> "SELECT" is not possible. I chose CMDTAG_SELECT_COMPAT
because with the COMPAT suffix it is explicit that it maps to different
things in different versions and also has a less chance of mapping
something irrevelant. For example if we used SELECT as a macro, then it
would map every SELECT to whatever it is mapping to, which might have
unexpected/undesired behaviour.
The error message when index has opclassopts is improved and the commit
from postgres side is also included for future reference.
Also some minor style related changes are applied.
This commit mostly adds pg_get_triggerdef_command to our ruleutils_13.
This doesn't add anything extra for ruleutils 13 so it is basically a copy
of the change on ruleutils_12
addRangeTableEntryXXX methods return a ParseNamespaceItem with pg >= 13.
RangeTableEntryFromNSItem macro is added so that we return the range
table entry from the ParseNamespaceItem in pg>=13 and for pg < 13 rte
would already be returned with addRangeTableEntryXXX methods.
Commit on Postgres side:
5815696bc66b3092f6361f53e0394909647042c8
PortalDefineQuery doesn't accept char* for command tag anymore with PG
>= 13. We are currently only using it with Select, therefore a Portal
define query compat for select is created.
Commit on PG side:
2f9661311b83dc481fc19f6e3bda015392010a40
As the new planner and pg_plan_query_compat methods expect the query
string as well, macros are defined to be compatible in different
versions of postgres.
Relevant commit on Postgres:
6aba63ef3e606db71beb596210dd95fa73c44ce2
Command on Postgres:
git log --all --grep="pg_plan_query"
With PG13 heap_* (heap_open, heap_close etc) are replaced with table_*
(table_open, table_close etc).
It is better to use the new table access methods in the codebase and
define the macros for the previous versions as we can easily remove the
macro without having to change the codebase when we drop the support for
the old version.
Commits that introduced this change on Postgres:
f25968c49697db673f6cd2a07b3f7626779f1827
e0c4ec07284db817e1f8d9adfb3fffc952252db0
4b21acf522d751ba5b6679df391d5121b6c4a35f
Command to see relevant commits on Postgres side:
git log --all --grep="heap_open"
With this patch, we introduce `locally_reserved_shared_connections.c/h` files
which are responsible for reserving some space in shared memory counters
upfront.
We sometimes need to reserve connections, but not necessarily
establish them. For example:
- COPY command should reserve connections as it cannot know which
connections it needs in which order. COPY establishes connections
as any input data hits the workers. For example, for router COPY
command, it only establishes 1 connection.
As discussed here (https://github.com/citusdata/citus/pull/3849#pullrequestreview-431792473),
COPY needs to reserve connections up-front, otherwise we can end
up with resource starvation/un-detected deadlocks.
The executor relies on WorkerPool, and many other places rely on WorkerNode.
With this commit, we make sure that they are sorted via the same function/logic.
* use adaptive executor even if task-tracker is set
* Update check-multi-mx tests for adaptive executor
Basically repartition joins are enabled where necessary. For parallel
tests max adaptive executor pool size is decresed to 2, otherwise we
would get too many clients error.
* Update limit_intermediate_size test
It seems that when we use adaptive executor instead of task tracker, we
exceed the intermediate result size less in the test. Therefore updated
the tests accordingly.
* Update multi_router_planner
It seems that there is one problem with multi_router_planner when we use
adaptive executor, we should fix the following error:
+ERROR: relation "authors_range_840010" does not exist
+CONTEXT: while executing command on localhost:57637
* update repartition join tests for check-multi
* update isolation tests for repartitioning
* Error out if shard_replication_factor > 1 with repartitioning
As we are removing the task tracker, we cannot switch to it if
shard_replication_factor > 1. In that case, we simply error out.
* Remove MULTI_EXECUTOR_TASK_TRACKER
* Remove multi_task_tracker_executor
Some utility methods are moved to task_execution_utils.c.
* Remove task tracker protocol methods
* Remove task_tracker.c methods
* remove unused methods from multi_server_executor
* fix style
* remove task tracker specific tests from worker_schedule
* comment out task tracker udf calls in tests
We were using task tracker udfs to test permissions in
multi_multiuser.sql. We should find some other way to test them, then we
should remove the commented out task tracker calls.
* remove task tracker test from follower schedule
* remove task tracker tests from multi mx schedule
* Remove task-tracker specific functions from worker functions
* remove multi task tracker extra schedule
* Remove unused methods from multi physical planner
* remove task_executor_type related things in tests
* remove LoadTuplesIntoTupleStore
* Do initial cleanup for repartition leftovers
During startup, task tracker would call TrackerCleanupJobDirectories and
TrackerCleanupJobSchemas to clean up leftover directories and job
schemas. With adaptive executor, while doing repartitions it is possible
to leak these things as well. We don't retry cleanups, so it is possible
to have leftover in case of errors.
TrackerCleanupJobDirectories is renamed as
RepartitionCleanupJobDirectories since it is repartition specific now,
however TrackerCleanupJobSchemas cannot be used currently because it is
task tracker specific. The thing is that this function is a no-op
currently.
We should add cleaning up intermediate schemas to DoInitialCleanup
method when that problem is solved(We might want to solve it in this PR
as well)
* Revert "remove task tracker tests from multi mx schedule"
This reverts commit 03ecc0a681.
* update multi mx repartition parallel tests
* not error with task_tracker_conninfo_cache_invalidate
* not run 4 repartition queries in parallel
It seems that when we run 4 repartition queries in parallel we get too
many clients error on CI even though we don't get it locally. Our guess
is that, it is because we open/close many connections without doing some
work and postgres has some delay to close the connections. Hence even
though connections are removed from the pg_stat_activity, they might
still not be closed. If the above assumption is correct, it is unlikely
for it to happen in practice because:
- There is some network latency in clusters, so this leaves some times
for connections to be able to close
- Repartition joins return some data and that also leaves some time for
connections to be fully closed.
As we don't get this error in our local, we currently assume that it is
not a bug. Ideally this wouldn't happen when we get rid of the
task-tracker repartition methods because they don't do any pruning and
might be opening more connections than necessary.
If this still gives us "too many clients" error, we can try to increase
the max_connections in our test suite(which is 100 by default).
Also there are different places where this error is given in postgres,
but adding some backtrace it seems that we get this from
ProcessStartupPacket. The backtraces can be found in this link:
https://circleci.com/gh/citusdata/citus/138702
* Set distributePlan->relationIdList when it is needed
It seems that we were setting the distributedPlan->relationIdList after
JobExecutorType is called, which would choose task-tracker if
replication factor > 1 and there is a repartition query. However, it
uses relationIdList to decide if the query has a repartition query, and
since it was not set yet, it would always think it is not a repartition
query and would choose adaptive executor when it should choose
task-tracker.
* use adaptive executor even with shard_replication_factor > 1
It seems that we were already using adaptive executor when
replication_factor > 1. So this commit removes the check.
* remove multi_resowner.c and deprecate some settings
* remove TaskExecution related leftovers
* change deprecated API error message
* not recursively plan single relatition repartition subquery
* recursively plan single relation repartition subquery
* test depreceated task tracker functions
* fix overlapping shard intervals in range-distributed test
* fix error message for citus_metadata_container
* drop task-tracker deprecated functions
* put the implemantation back to worker_cleanup_job_schema_cachesince citus cloud uses it
* drop some functions, add downgrade script
Some deprecated functions are dropped.
Downgrade script is added.
Some gucs are deprecated.
A new guc for repartition joins bucket size is added.
* order by a test to fix flappiness
The reason we should use ActiveReadableNodeList instead of ActiveReadableNonCoordinatorNodeList is that if coordinator is added to cluster as a worker, it should be counted as well. Otherwise if there is only coordinator in the cluster, the count will be 0, hence we get a warning.
In MultiTaskTrackerExecute, we should connect to coordinator if it is
added to the cluster because it will also be assigned tasks.
We were using ALL_WORKERS TargetWorkerSet while sending temporary schema
creation and cleanup. We(well mostly I) thought that ALL_WORKERS would also include coordinator when it is added as a worker. It turns out that it was FILTERING OUT the coordinator even if it is added as a worker to the cluster.
So to have some context here, in repartitions, for each jobId we create
(at least we were supposed to) a schema in each worker node in the cluster. Then we partition each shard table into some intermediate files, which is called the PARTITION step. So after this partition step each node has some intermediate files having tuples in those nodes. Then we fetch the partition files to necessary worker nodes, which is called the FETCH step. Then from the files we create intermediate tables in the temporarily created schemas, which is called a MERGE step. Then after evaluating the result, we remove the temporary schemas(one for each job ID in each node) and files.
If node 1 has file1, and node 2 has file2 after PARTITION step, it is
enough to either move file1 from node1 to node2 or vice versa. So we
prune one of them.
In the MERGE step, if the schema for a given jobID doesn't exist, the
node tries to use the `public` schema if it is a superuser, which is
actually added for testing in the past.
So when we were not sending schema creation comands for each job ID to
the coordinator(because we were using ALL_WORKERS flag, and it doesn't
include the coordinator), we would basically not have any schemas for
repartitions in the coordinator. The PARTITION step would be executed on
the coordinator (because the tasks are generated in the planner part)
and it wouldn't give us any error because it doesn't have anything to do
with the temporary schemas(that we didn't create). But later two things
would happen:
- If by chance the fetch is pruned on the coordinator side, we the other
nodes would fetch the partitioned files from the coordinator and execute
the query as expected, because it has all the information.
- If the fetch tasks are not pruned in the coordinator, in the MERGE
step, the coordinator would either error out saying that the necessary
schema doesn't exist, or it would try to create the temporary tables
under public schema ( if it is a superuser). But then if we had the same
task ID with different jobID it would fail saying that the table already
exists, which is an error we were getting.
In the first case, the query would work okay, but it would still not do
the cleanup, hence we would leave the partitioned files from the
PARTITION step there. Hence ensure_no_intermediate_data_leak would fail.
To make things more explicit and prevent such bugs in the future,
ALL_WORKERS is named as ALL_NON_COORD_WORKERS. And a new flag to return
all the active nodes is added as ALL_DATA_NODES. For repartition case,
we don't use the only-reference table nodes but this version makes the
code simpler and there shouldn't be any significant performance issue
with that.
Rename TargetWorkerSet enums to make them more explicit about what they
mean. Ideally it would be good to treat everything as a node without the
'worker' concept because it makes things complicated. Another
improvement could be to rename TargetWorkerSet as TargetNodeSet but it
goes to renaming many occurrences of Worker, which is probably too big
for this PR.
#3866 removed the shard ID hash in metadata_cache.c to simplify cache management,
but we observed a significant performance regression that was being masked by the
performance improvement provided by #3654 in our benchmarks, but #3654 only
applies to specific workloads.
This PR brings back the shard ID cache as it existed before #3866 with some extra
measures to handle invalidation. When we load a table entry, we overwrite
ShardIdCacheEntry->tableEntry pointers for all the shards in that table, though
it's possible that the table no longer contains the old shard ID or the table
entry is never reloaded, which would leave a dangling pointer once the table
entry is freed. To handle that case, we remove all shard ID cache entries that
point exactly to that table entry when a table is freed (at the end of the
transaction or any call to CitusTableCacheFlushInvalidatedEntries).
Co-authored-by: SaitTalhaNisanci <s.talhanisanci@gmail.com>
Co-authored-by: Marco Slot <marco.slot@gmail.com>
Co-authored-by: Jelte Fennema <github-tech@jeltef.nl>
This is so we don't need to calculate it twice in
insert_select_executor.c and multi_explain.c, which can
cause discrepancy if an update in one of them is not
reflected in the other site.
* Not set TaskExecution with adaptive executor
Adaptive executor is using a utility method from task tracker for
repartition joins, however adaptive executor doesn't need taskExecution.
It is only used by task tracker. This causes a problem when explain
analyze is used because what taskExecution is pointing to might be
random.
We solve this by not setting taskExecution from adaptive executor. So it
will stay NULL as set by CreateTask.
* use same memory context as task for taskExecution
Co-authored-by: Jelte Fennema <github-tech@jeltef.nl>
As suggested by @marcocitus in https://github.com/citusdata/citus/pull/3911#issuecomment-643978531, there was
a regression in #3893. If another backend would write a file during deletion of
the intermediate results directory, this file would not necessarily be deleted.
The approach used in `CitusRemoveDirectory` is to try recursive removal of the
directory again if it has failed. This does not work here, since when a file
can not be removed for other reasons (e.g. `EPERM`) it will not throw an error
anymore. So then we would get into an infinite removal loop. Instead I now
`rename` the directory before removing it. That way other backends will not
write files to it anymore.
We sort the workerList because adaptive connection management
(e.g., OPTIONAL_CONNECTION) requires any concurrent executions
to wait for the connections in the same order to prevent any
starvation. If we don't sort, we might end up with:
Execution 1: Get connection for worker 1, wait for worker 2
Execution 2: Get connection for worker 2, wait for worker 1
and, none could proceed. Instead, we enforce every execution establish
the required connections to workers in the same order.
In #3901 the "Data received from worker(s)" sections were added to EXPLAIN
ANALYZE. After merging @pykello posted some review comments. This addresses
those comments as well as fixing a other issues that I found while addressing
them. The things this does:
1. Fix `EXPLAIN ANALYZE EXECUTE p1` to not increase received data on every
execution
2. Fix `EXPLAIN ANALYZE EXECUTE p1(1)` to not return 0 bytes as received data
allways.
3. Move `EXPLAIN ANALYZE` specific logic to `multi_explain.c` from
`adaptive_executor.c`
4. Change naming of new explain sections to `Tuple data received from node(s)`.
Firstly because a task can reference the coordinator too, so "worker(s)" was
incorrect. Secondly to indicate that this is tuple data and not all network
traffic that was performed.
5. Rename `totalReceivedData` in our codebase to `totalReceivedTupleData` to
make it clearer that it's a tuple data counter, not all network traffic.
6. Actually add `binary_protocol` test to `multi_schedule` (woops)
7. Fix a randomly failing test in `local_shard_execution.sql`.
Shard id to index mapping stored in cache entry as there may now be multiple entries alive for a given relation
insert_select_executor: revert copying cache entry, which was a hack added to avoid memory safety issues
Sadly this does not actually work yet for binary protocol data, because
when doing EXPLAIN ANALYZE we send two commands at the same time. This
means we cannot use `SendRemoteCommandParams`, and thus cannot use the
binary protocol. This can still be useful though when using the text
protocol, to find out that a lot of data is being sent.
This can save a lot of data to be sent in some cases, thus improving
performance for which inter query bandwidth is the bottleneck.
There's some issues with enabling this as default, so that's currently not done.
We still recursively plan some cases, eg:
- INSERTs
- SELECT FOR UPDATE when reference tables in query
- Everything must be same single shard & replication model
We wrap worker tasks in worker_save_query_explain_analyze() so we can fetch
their explain output later by a call worker_last_saved_explain_analyze().
Fixes#3519Fixes#2347Fixes#2613Fixes#621
This is a different version of #3634. It also removes SwallowErrors, but
instead of modifying our own functions to not throw errors, it uses the
postgres built in `PathNameDeleteTemporaryDir` function. This function
does not throw errors.
Since this change is for a bugfix, I tried to minimize the changes.
PRs with the following changes would be good to do separately from this
PR:
1. Use PathName(Create|Open|Delete)Temporary(File|Dir) to open and
remove all files/dirs instead of our own custom file functions.
2. Prefix our outmost files/directories with `PG_TEMP_FILE_PREFIX` so
that they are identified by Postgres as temporary files, which will be
removed at postmaster start. This way we do not have to do this cleanup
ourselves.
3. Store the files in the temporary table space if it exists.
Fixes#3634Fixes#3618
If we want to get necessary lockmode for a relation RangeVar within
a query, we can get the lockmode easily from the RangeVar itself (if
pg version >= 12).
However, if we want to decide the lockmode appropriate for the
"query", we can derive this information by using GetQueryLockMode
according to the code comment from RangeTblEntry->rellockmode.
Implements a new `TupleDestination` interface to allow custom tuple processing per task.
This can be specially useful if a task contains multiple queries. An example of this EXPLAIN
ANALYZE, where it needs to add some UDF calls to the query to fetch the explain output
from worker after fetching the actual query results.
SELECT_TASK is renamed to READ_TASK as a SELECT with modifying CTEs will be a MODIFYING_TASK
RouterInsertJob: Assert originalQuery->commandType == CMD_INSERT
CreateModifyPlan: Assert originalQuery->commandType != CMD_SELECT
Remove unused function IsModifyDistributedPlan
DistributedExecution, ExecutionParams, DistributedPlan: Rename hasReturning to expectResults
SELECTs set expectResults to true
Rename CreateSingleTaskRouterPlan to CreateSingleTaskRouterSelectPlan
This PR removes ExecuteUtilityTaskListWithoutResults and uses the same
path for local execution via ExecuteTaskListExtended.
ExecuteUtilityTaskList is added. ExecuteLocalTaskListExtended now has a
parameter for utility commands so that it can call the right method. In
order not to change the existing calls,
ExecuteTaskListExtendedInternal is added, which is the main method that
runs the execution, via local and remote execution.
The reason is that PQconnectPoll() may change the underlying
socket. If we don't rebuild the wait event set, the low level
APIs (such as epoll_ctl()) may fail due to invalid sockets.
Instead, rebuilding ensures that we'll use accurate/active sockets.
* Not append empty task in ExtractLocalAndRemoteTasks
ExtractLocalAndRemoteTasks extracts the local and remote tasks. If we do
not have a local task the localTaskPlacementList will be NIL, in this
case we should not append anything to local tasks. Previously we would
first check if a task contains a single placement or not, now we first
check if there is any local task before doing anything.
* fix copy of node task
Task node has task query, which might contain a list of strings in its
fields. We were using postgres copyObject for these lists. Postgres
assumes that each element of list will be a node type. If it is not a
node type it will error.
As a solution to that, a new macro is introduced to copy a list of
strings.
We had 9+ parameters in some of the functions related to execution.
Execution params is created to simplify this a bit so that we can set
only the fields that we are interested in and it is easier to read.
With this commit, we're introducing a new infrastructure to throttle
connections to the worker nodes. This infrastructure is useful for
multi-shard queries, router queries are have not been affected by this.
The goal is to prevent establishing more than citus.max_shared_pool_size
number of connections per worker node in total, across sessions.
To do that, we've introduced a new connection flag OPTIONAL_CONNECTION.
The idea is that some connections are optional such as the second
(and further connections) for the adaptive executor. A single connection
is enough to finish the distributed execution, the others are useful to
execute the query faster. Thus, they can be consider as optional connections.
When an optional connection is not allowed to the adaptive executor, it
simply skips it and continues the execution with the already established
connections. However, it'll keep retrying to establish optional
connections, in case some slots are open again.
We currently don't use any cursor flags in local execution, but we can
use CURSOR_OPT_PARALLEL_OK flag to potentially benefit from parallelism
when possible.
We have two variables that are related to local execution status.
TransactionAccessedLocalPlacement and
TransactionConnectedToLocalGroup. Only one of these fields should be
set, however we didn't have any check for this contraint and it was
error prone.
What those two variables are used is that we are trying to understand if
we should use local execution, the current session, or if we should be
using a connection to execute the current query, therefore the tasks. In
the enum, now it is more clear what these variables mean.
Also, now we have a method to change the local execution status. The
method will error if we are trying to transition from a state to a wrong
state. This will help us avoid problems.
* use local executon when in a transaction block
When we are inside a transaction block, there could be other methods
that need local execution, therefore we will use local execution in a
transaction block.
* update test outputs with transaction block local execution
* add a test to verify we dont leak intermediate schemas
* test that we don't leak intermediate schemas
We have tests to make sure that we don't intermediate any intermediate
files, tables etc but we don't test if we are leaking schemas. It makes
sense to test this as well.
* remove all repartition schemas in case of error
This solution is not an ideal one but it seems to be doing the job.
We should have a more generic solution for the cleanup but it seems that
putting the cleanup in the abort handler is dangerous and it was
crashing.
It is possible to return an error in ExecuteTaskListExtended after
performing local execution with the current structure. However there is
no point in execution the local tasks if we are going to return an error
later. So the local execution is moved after the error check.
When the file does not exist, it could mean two different things.
First -- and a lot more common -- case is that a failure happened
in a concurrent backend on the same distributed transaction. And,
one of the backends in that transaction has already been roll
backed, which has already removed the file. If we throw an error
here, the user might see this error instead of the actual error
message. Instead, we prefer to WARN the user and pretend that the
file has no data in it. In the end, the user would see the actual
error message for the failure.
Second, in case of any bugs in intermediate result broadcasts,
we could try to read a non-existing file. That is most likely
to happen during development. Thus, when asserts enabled, we throw
an error instead of WARNING so that the developers cannot miss.
When we have a query like the following:
```SQL
WITH a AS (SELECT * FROM foo LIMIT 10) SELECT max(x) FROM a JOIN bar 2 USING (y);
```
Citus currently opens side channels for doing the
`COPY "1_1"` FROM STDIN (format 'result')
before starting the execution of
`SELECT * FROM foo LIMIT 10`
Since we need at least 1 connection per worker to do
`SELECT * FROM foo LIMIT 10`
We need to have 2 connections to worker in order to broadcast the results.
However, we don't actually send a single row over the side channel until the
execution of `SELECT * FROM foo LIMIT 10` is completely done (and connections
unclaimed) and the results are written to a tuple store. We could actually
reuse the same connection for doing the `COPY "1_1"` FROM STDIN (format 'result').
This also fixes the issue that Citus doesn't obey `citus.max_adaptive_executor_pool_size`
when the query includes an intermediate result.
We don't need any side channel connections. That is actually
problematic in the sense that it creates extra connections.
Say, citus.max_adaptive_executor_pool_size equals to 1, Citus
ends up using one extra connection for the intermediate results.
Thus, not obeying citus.max_adaptive_executor_pool_size.
In this PR, we remove the following entities from the codebase
to allow further commits to implement not requiring extra connection
for the intermediate results:
- The connection flag REQUIRE_SIDECHANNEL
- The function GivePurposeToConnection
- The ConnectionPurpose struct and related fields
* explicitly return false if transaction connected to local node
* not set TransactionConnectedToLocalGroup if we are writing to a file
We use TransactionConnectedToLocalGroup to prevent local execution from
happening as that might cause visibility problems. As files are visible
to all transactions, we shouldn't set this variable if we are writing to
a file.
In case we don't care about the tupleStoreState in
ExecuteLocalTaskListExtended, it could be passed as null. In that case
we will get a seg error. This changes it so that a dummy tuple store
will be created when it is null.
Do not use local execution in ExecuteTaskListOutsideTransaction.
As we are going to run the tasks outside transaction, we shouldn't use local execution.
However, there is some problem when using local execution related to
repartition joins, when we solve that problem, we can execute the tasks
coming to this path with local execution.
Also logging the local command is simplified.
normalize job id in worker_hash_partition_table in test outputs.
For shardplacements, we were setting nodeid, nodename, nodeport and
nodegroup manually. This makes it very error prone, and it seems that we
already forgot to set some of them. This would mean that they would have
their default values, e.g group id would be 0 when its group id is not
0.
So the implication is that we would have inconsistent worker metadata.
A new method is introduced, and we call the method to set those fields
now, so that as long as we call this method, we won't be setting
inconsistent metadata.
It probably makes sense to have a struct for these fields. We already
have NodeMetadata but it doesn't have nodename or nodeport. So that
could be done over another refactor to make things simpler.
ExecuteTaskListExtended is the common method for different codepaths,
and instead of writing separate local execution logics in different
codepaths, it makes more sense to have the logic here. We still need to
do some refactoring, this is an initial step.
After this commit, we can run create shard commands locally. There is a
special case with shard creation commands. A create shard command might
have a concatenated query string, however local execution did not know
how to execute a task with multiple query strings. This is also
implemented in this commit. We go over each query in the concatenated
query string and plan/execute them one by one.
A more clean solution to this would be to make sure that each task has a
single query. We currently cannot do that because we need to ensure the
task dependencies. However, it would make sense to do that at some point
and it would simplify the code a lot.
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.
Onder Kalaci
Marco Slot
SaitTalhaNisanci
Önder Kalacı
Philip Dubé
Naisila Puka
Nils Dijk
Onur Tirtir
Jelte Fennema
Hadi Moshayedi
Hanefi Onaldi
citus bot
Marco Slot
Philip Dubé
Philip Dubé