Implement "citus local table" creation logic and a udf #3852

onurctirtir · 2020-05-20T16:45:50Z

(Replaces #3729)

This pr is opened against single-placement-table/master branch, the main branch we are developing citus local tables. We will merge the things as a whole when the end-to-end implementation is complete.

This pr implements lazy shard creation for citus local tables. That means:

Simply rename the given table to citus local table shard.
Suffix the existing objects on the shard relation with shardId.
Re-create those objects on the shell table (indexes, foreign key constraints etc.)
Insert metadata for the table.

TODO:

(in another pr):

Policy support
~~(pr is open: Bring deparse logic for policies from enterprise #3896, should add some tests for citus local tables here or in another pr against single-placement-table/master)~~
(won't support policies on citus local tables in community)
Utility command processing (includes foreign key support)
Planner changes

src/backend/distributed/commands/create_distributed_table.c

onderkalaci · 2020-05-27T14:17:46Z

src/backend/distributed/commands/create_distributed_table.c

+ *  - its distribution method will be DISTRIBUTE_BY_NONE,
+ *  - its replication model will be ReplicationModel,
+ *  - its replication factor will be set to 1.
+ * On the contrary of reference tables, a citus local table has only one placement


this line of the comment is slightly confusing: On the contrary of reference tables

I get what you mean, but the sentence should be separated. Similar to ref. tables, it has 1 placement, in addition to that the restriction is the placement is only allowed on the coordinator.

onderkalaci · 2020-05-27T14:27:21Z

src/backend/distributed/commands/create_distributed_table.c

+
+	EnsureTableOwner(relationOid);
+
+	/* we allow creating citus local tables only from relations with RELKIND_RELATION */


I think we should be allowing any kind of relation

Actually EnsureRelationKindSupported allows us to create the citus table if the table is

a regular table or

a foreign table or

a partitioned table and

not an inherited table.

This check is to prevent views, sequences etc.

onderkalaci · 2020-05-27T14:34:54Z

src/backend/distributed/commands/create_distributed_table.c

+	 * the nodes such that we can create foreign keys and joins work immediately
+	 * after creation.
+	 */
+	EnsureReferenceTablesExistOnAllNodes();


Could we ever be in a state where reference tables are not replicated?

As #3637 claims, yes

src/backend/distributed/commands/create_distributed_table.c

src/backend/distributed/master/worker_node_manager.c

src/backend/distributed/master/master_create_shards.c

src/backend/distributed/master/master_node_protocol.c

SaitTalhaNisanci

I have done an initial review, I will also do tests locally for different scenarios. Submitting the review for faster iteration.

src/test/regress/sql/citus_local_tables.sql

src/backend/distributed/master/worker_node_manager.c

src/backend/distributed/metadata/metadata_cache.c

src/backend/distributed/utils/colocation_utils.c

src/test/regress/bin/normalize.sed

SaitTalhaNisanci · 2020-06-02T12:26:22Z

src/test/regress/sql/citus_local_tables.sql

+-- any foreign key between citus local tables and other tables cannot be set for now
+-- most should error out (for now with meaningless error messages)
+
+-- between citus local tables


We can also have tests where a table references a some other table before creating the citus local table.

As now we don't support initial fkeys with other tables, I will add some tests with self referencing fkeys

onderkalaci

We seem to have a minor bug with schemas:

CREATE SCHEMA "CiTUS!LocalTables";

-- create table with weird names
CREATE TABLE "CiTUS!LocalTables"."LocalTabLE.1!?!"(id int, "TeNANt_Id" int);
SELECT create_citus_local_table('"CiTUS!LocalTables"."LocalTabLE.1!?!"');
ERROR:  syntax error at or near ".1"

And the backtrace:

* thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 3.1
  * frame #0: 0x000000010293e467 postgres`errstart(elevel=20, filename="scan.l", lineno=1149, funcname="scanner_yyerror", domain=0x0000000000000000) at elog.c:263:14
    frame #1: 0x00000001029a3da5 postgres`scanner_yyerror.cold.2 + 53
    frame #2: 0x000000010266c577 postgres`scanner_yyerror(message=<unavailable>, yyscanner=<unavailable>) at scan.l:1145:3
    frame #3: 0x000000010266aef2 postgres`base_yyerror(yylloc=<unavailable>, yyscanner=<unavailable>, msg=<unavailable>) at gram.y:15510:2
    frame #4: 0x0000000102667765 postgres`base_yyparse(yyscanner=0x00007fdb49013f10) at gram.c:45676:7
    frame #5: 0x000000010266ddb2 postgres`raw_parser(str=<unavailable>) at parser.c:53:13
    frame #6: 0x00000001028566f4 postgres`pg_parse_query(query_string="ALTER TABLE \"CiTUS!LocalTables\".\"LocalTabLE.1!?!\" RENAME TO LocalTabLE.1!?!_102087;") at postgres.c:641:23
    frame #7: 0x000000010521a74f citus.so`ParseTreeRawStmt(ddlCommand="ALTER TABLE \"CiTUS!LocalTables\".\"LocalTabLE.1!?!\" RENAME TO LocalTabLE.1!?!_102087;") at worker_data_fetch_protocol.c:574:24
    frame #8: 0x000000010521a2d9 citus.so`ParseTreeNode(ddlCommand=<unavailable>) at worker_data_fetch_protocol.c:560:24
    frame #9: 0x000000010519b359 citus.so`RenameRelationToShardRelation(shellRelationId=<unavailable>, shardId=<unavailable>) at create_citus_local_table.c:260:20
    frame #10: 0x000000010519b091 citus.so`CreateCitusLocalTableShard(relationId=1913196, shardId=102087) at create_citus_local_table.c:231:2
    frame #11: 0x000000010519afab citus.so`CreateCitusLocalTable(shellRelationId=1913196) at create_citus_local_table.c:182:2
    frame #12: 0x000000010519ae97 citus.so`create_citus_local_table(fcinfo=<unavailable>) at create_citus_local_table.c:90:2
    frame #13: 0x00000001026f82c1 postgres`ExecInterpExpr(state=0x00007fdb4706ef98, econtext=0x00007fdb4706ec88, isnull=0x00007ffeed68ae97) at execExprInterp.c:649:8
    frame #14: 0x00000001027272cd postgres`ExecEvalExprSwitchContext(state=0x00007fdb4706ef98, econtext=0x00007fdb4706ec88, isNull=<unavailable>) at executor.h:307:13
    frame #15: 0x0000000102727280 postgres`ExecProject(projInfo=<unavailable>) at executor.h:341:9
    frame #16: 0x00000001026fe890 postgres`ExecutePlan(estate=0x00007fdb4706e918, planstate=0x00007fdb4706eb70, use_parallel_mode=<unavailable>, operation=CMD_SELECT, sendTuples=<unavailable>, numberTuples=0, direction=ForwardScanDirection, dest=0x00007fdb4a81bda8, execute_once=<unavailable>) at execMain.c:1646:10
    frame #17: 0x00000001026fe762 postgres`standard_ExecutorRun(queryDesc=0x00007fdb4701ed18, direction=<unavailable>, count=0, execute_once=<unavailable>) at execMain.c:364:3
    frame #18: 0x00000001051ca81e citus.so`CitusExecutorRun(queryDesc=0x00007fdb4701ed18, direction=ForwardScanDirection, count=0, execute_once=<unavailable>) at multi_executor.c:203:4
    frame #19: 0x00000001053ab521 pg_stat_statements.so`pgss_ExecutorRun(queryDesc=0x00007fdb4701ed18, direction=ForwardScanDirection, count=0, execute_once=<unavailable>) at pg_stat_statements.c:891:4
    frame #20: 0x000000010285b9a2 postgres`PortalRunSelect(portal=0x00007fdb47833118, forward=<unavailable>, count=0, dest=<unavailable>) at pquery.c:929:4
    frame #21: 0x000000010285b524 postgres`PortalRun(portal=0x00007fdb47833118, count=9223372036854775807, isTopLevel=<unavailable>, run_once=<unavailable>, dest=0x00007fdb4a81bda8, altdest=0x00007fdb4a81bda8, completionTag="") at pquery.c:770:18
    frame #22: 0x0000000102858be8 postgres`exec_simple_query(query_string="SELECT create_citus_local_table('\"CiTUS!LocalTables\".\"LocalTabLE.1!?!\"');") at postgres.c:1215:10
    frame #23: 0x0000000102858395 postgres`PostgresMain(argc=<unavailable>, argv=<unavailable>, dbname=<unavailable>, username=<unavailable>) at postgres.c:0
    frame #24: 0x00000001027e2fc6 postgres`BackendRun(port=0x0000000102ba4a7a) at postmaster.c:4448:2
    frame #25: 0x00000001027e2889 postgres`BackendStartup(port=<unavailable>) at postmaster.c:4139:3
    frame #26: 0x00000001027e2065 postgres`ServerLoop at postmaster.c:1704:7
    frame #27: 0x00000001027e0117 postgres`PostmasterMain(argc=3, argv=0x00007fdb46c03210) at postmaster.c:1377:11
    frame #28: 0x000000010274a1b6 postgres`main(argc=3, argv=0x00007fdb46c03210) at main.c:228:3
    frame #29: 0x00007fff631f63d5 libdyld.dylib`start + 1
(lldb) c
Process 12418 resuming

This is blocking my testing at the moment. These are the basic tests that I'm considering to expand: https://gist.github.com/onderkalaci/a9856bfc8b242ee154e55b6486a48524

So, maybe make sure that we support all before I restart testing. (there might be some syntax errors in the file)

src/backend/distributed/commands/create_citus_local_table.c

onderkalaci · 2020-06-04T08:31:12Z

src/backend/distributed/commands/create_citus_local_table.c

+	EnsureTableOwner(relationId);
+
+	/* we allow creating citus local tables only from relations with RELKIND_RELATION */
+	EnsureRelationKindSupported(relationId);


Can't we simply move if (PartitionTable(relationId) || PartitionedTable(relationId)) check before this one? I'm OK not having a new function because we're giving the correct errors.

onderkalaci · 2020-06-04T08:50:45Z

src/backend/distributed/commands/create_citus_local_table.c

+ *  - the foreign keys referencing to the relation.
+ */
+static void
+CreateCitusLocalTableShard(Oid relationId, uint64 shardId)


Shall we call this function as RenamePostgresTableToCitusShard to make it more explicit?

src/backend/distributed/commands/create_citus_local_table.c

src/backend/distributed/master/worker_node_manager.c

marcocitus · 2020-06-05T09:44:42Z

src/backend/distributed/sql/udfs/create_citus_local_table/latest.sql

+	RETURNS void
+	LANGUAGE C STRICT
+	AS 'MODULE_PATHNAME', $$create_citus_local_table$$;
+COMMENT ON FUNCTION create_citus_local_table(table_name regclass)


missing pg_catalog.

onderkalaci · 2020-06-05T13:59:58Z

src/backend/distributed/commands/create_citus_local_table.c

+	 * Include DEFAULT clauses for columns getting their default values from
+	 * a sequence.
+	 */
+	bool includeSequenceDefaults = true;


Hmm, with this one we are slightly diverging from the regular shard creation. On hash distributed tables, we don't create the sequences on the shards ( see CreateShardsOnWorkers), we only have sequences on the coordinator shell tables. And, we evaluate sequences on the coordinator.

As far as I can tell, this is OK. Let's sign-up @marcocitus's office hour next week, and we explain the latest implementation and can ask these types of edge cases to him.

Well, it turns out that when there is a sequence in the table, we cannot enable MX:

Tables in this gist: https://gist.github.com/onderkalaci/a9856bfc8b242ee154e55b6486a48524:

select start_metadata_sync_to_node(nodename, nodeport) FROM pg_dist_node; ERROR: relation "CiTUS!LocalTables.LocalTabLE.1!?!_serial_data_seq" does not exist CONTEXT: while executing command on localhost:9700

I think that is because of the above note. getOwnedSequences returns the sequence for the shard but not for the shell table. That's why we probably need to drop the sequence on the shard

I will look into that Monday

Wonder whether we should make a pass over pg_depend to update dependencies to the shell table OID.

Semi-related: 9ed67a0

onderkalaci 4 days ago Member
Well, it turns out that when there is a sequence in the table, we cannot enable MX:

For the sequences implied by column definitions automatically (serial, bigserial etc), I implemented the following patch: fbc981f . This commit drops default column definitions from shard relation and grants ownerships of owned sequences to the corresponding columns of the shell table.

However, we will still have two problems around sequences as I mentioned in #3885 & #3884.
They point to existing problems around the other citus table types, and obviously will apply to citus local tables too.

I wonder if I should work on implementing a solution (like the one in 9ed67a0 done by Marco) in another pr before merging this pr, or should we leave it to another time ?

(cc: @marcocitus & cc: @onderkalaci )

or should we leave it to another time ?

I think we can leave it to a later time

SaitTalhaNisanci · 2020-06-05T14:54:15Z

src/backend/distributed/commands/create_citus_local_table.c

+	 */
+	EnsureReferenceTablesExistOnAllNodes();
+
+	List *shellTableDDLEvents = GetShellTableDDLEventsForCitusLocalTable(relationId);


Thanks for using blocks here, I think we can further use more blocks by separating the part until this like into another function and after the call InsertMetadataForCitusLocalTable to another function as pre create table operations and post create table operations. I think there might be common operations between different table types and we might need to call the methods from those places as well, but still I suggest creating these methods at least for citus local tables in this PR so that this function is even in better form and later in a separate PR we can use the methods(with small changes) in other places as well.

src/backend/distributed/commands/create_citus_local_table.c

onderkalaci · 2020-06-05T14:18:08Z

src/backend/distributed/commands/create_citus_local_table.c

+	 * Include DEFAULT clauses for columns getting their default values from
+	 * a sequence.
+	 */
+	bool includeSequenceDefaults = true;


Well, it turns out that when there is a sequence in the table, we cannot enable MX:

Tables in this gist: https://gist.github.com/onderkalaci/a9856bfc8b242ee154e55b6486a48524:

select start_metadata_sync_to_node(nodename, nodeport) FROM pg_dist_node; ERROR: relation "CiTUS!LocalTables.LocalTabLE.1!?!_serial_data_seq" does not exist CONTEXT: while executing command on localhost:9700

I think that is because of the above note. getOwnedSequences returns the sequence for the shard but not for the shell table. That's why we probably need to drop the sequence on the shard

onderkalaci · 2020-06-05T15:25:07Z

src/backend/distributed/commands/create_citus_local_table.c

+
+	FinalizeCitusLocalTableCreation(relationId);
+
+	relation_close(relation, ExclusiveLock);


Why are we releasing the lock? I think we should close with NoLock, keep the lock until the end of the transaction. (Still, adding this to discussion with Marco)

Now as we pushed all the operations done on relation into that function (CreateCitusLocalTable), isn't it good to release the lock at the end ?

I think Talha's test reveals the issue. Holding the lock until the transaction ends is necessary

open two backends
create a table 'create table local(a int);' in the first one
run select create_citus_local_table('local') but stop it right after ConvertLocalTableToShard is executed, so at this point the postgres table is converted to a shard but shell table(empty table) is not created yet.
in the other backend run 'select create_citus_local_table('local');', now this will block in Relation relation = relation_open(relationId, ExclusiveLock); at CreateCitusLocalTable
Continue in the first one, so it creates the citus local table
Now as the lock is released from the first backend, the second one continues and it doesn't fail and in pg_dist_partition we have an extra entry now:

(please retry this test after changing to NoLock )

I will try and update here

As I also tried locally, " not releasing the lock" doesn't help to solve the problem that the @SaitTalhaNisanci mentioned. This is because, either we release the lock at the end of xact or when the all the creation logic is done in CreateCitusLocalTable, the other backend would still try to create a citus local table from the relationId and this is indeed valid as that relation is still not a citus relation, but it became a shard relation.

So, I propose to check for the the relationId with RelationIsAKnownShard function to error out.

Does it sound good ?

src/backend/distributed/master/worker_node_manager.c

src/backend/distributed/commands/create_citus_local_table.c

This reverts commit 48480d3.

onurctirtir · 2020-06-23T09:12:54Z

Now I can see codecov report and it shows %99, I think that is pretty good :)

SaitTalhaNisanci

I did some tests and seemed to work as expected. As we will do more tests when merging to master, I am approving.

SaitTalhaNisanci · 2020-06-25T08:23:12Z

src/test/regress/sql/citus_local_tables.sql

+CREATE TABLE citus_local_table_1 (a int primary key);
+
+-- this should fail as coordinator is removed from pg_dist_node
+SELECT create_citus_local_table('citus_local_table_1');


This fails with the call ErrorIfCoordinatorNotAddedAsWorkerNode. So the only single line that is not covered in this PR is in CoordinatorNode :

ereport(ERROR, (errmsg("couldn't find the coordinator node in the metadata " "as it is not added as a worker")));

Currently it seems a bit hard to hit this line. I have tried the following:

in one backend stopped the execution right after ErrorIfCoordinatorNotAddedAsWorkerNode while creating a citus local table

in another backend tried to remove the coordinator from the cluster

As expected the remove operation failed with:

[local] talha@talha:9700-30867=# SELECT 1 FROM master_remove_node('localhost', 9700); ERROR: you cannot remove the primary node of a node group which has shard placements Time: 28518.667 ms (00:28.519)

So this is expected as the second backend is trying to get an ExclusiveLock on pg_dist_node in ModifiableWorkerNode, which is held by the first backend.

SaitTalhaNisanci · 2020-06-25T08:40:43Z

src/test/regress/sql/citus_local_tables.sql

+-- cannot create citus local table from an existing citus table
+SELECT create_distributed_table('distributed_table', 'a');
+
+-- this will error out


maybe move the above comment here -- cannot create citus local table from an existing citus table

SaitTalhaNisanci · 2020-06-25T08:41:02Z

src/test/regress/sql/citus_local_tables.sql

+-- this will error out
+SELECT create_citus_local_table('distributed_table');
+
+-- partitiond table tests --


partitiond -> partitioned

SaitTalhaNisanci · 2020-06-25T08:45:05Z

src/test/regress/sql/citus_local_tables.sql

+BEGIN;
+  CREATE TABLE citus_local_table PARTITION OF partitioned_table FOR VALUES FROM (20) TO (30);
+
+  -- this should fail


maybe move cannot create citus local table as a partition of a local table here

SaitTalhaNisanci · 2020-06-25T08:46:10Z

src/test/regress/sql/citus_local_tables.sql

+
+  SELECT create_citus_local_table('citus_local_table');
+
+  -- this should fail


I think in general instead of writing this should fail, writing the reason directly right above the command could be easier to understand later.

SaitTalhaNisanci · 2020-06-25T08:57:52Z

src/test/regress/spec/isolation_create_citus_local_table.spec

@@ -0,0 +1,75 @@
+setup


Thanks for adding these tests.

SaitTalhaNisanci · 2020-06-25T09:36:08Z

src/test/regress/bin/normalize.sed

@@ -34,6 +34,9 @@ s/ keyval(1|2|ref)_[0-9]+ / keyval\1_xxxxxxx /g
 # shard table names for custom_aggregate_support
 s/ daily_uniques_[0-9]+ / daily_uniques_xxxxxxx /g

+# shard table names for isolation_create_citus_local_table
+s/ "citus_local_table_[0-9]_[0-9]+" / "citus_local_table_[0-9]_xxxxxxx" /g


what if the first part is more than 1 digit? then would it make sense to use [0-9]+ in the first part as well?

yes, it makes sense

SaitTalhaNisanci · 2020-06-25T10:07:18Z

src/include/distributed/commands.h

+ */
+typedef enum ExtractForeignKeyConstrainstMode
+{
+	/* extract the foreign key OIDs where the table is the referencing one */


Minor:
extract -> include to be aligned with the variable name

I'm less inclined to make this change in this pr as the only related change done in this pr is to make its usage public

SaitTalhaNisanci · 2020-06-25T12:34:34Z

src/backend/distributed/commands/create_citus_local_table.c

+	 * create_citus_local_table executions would not error out for that relation.
+	 * Hence we need to error out for shard relations too.
+	 */
+	ErrorIfRelationIsAKnownShard(relationId);


If we remove 'citus local table' from these messages, it might make sense to put these methods to a common place as non-static so that they can be used in other places as well.

SaitTalhaNisanci · 2020-06-25T12:35:23Z

src/backend/distributed/commands/create_citus_local_table.c

+
+
+/*
+ * ErrorIfCoordinatorNotAddedAsWorkerNode error out if coordinator is not added


error -> errors

SaitTalhaNisanci · 2020-06-25T14:55:56Z

src/backend/distributed/operations/worker_node_manager.c

-		ereport(ERROR, (errmsg("couldn't find the coordinator node in the metadata "
-							   "as it is not added as a worker")));
-	}
+	ErrorIfCoordinatorNotAddedAsWorkerNode();


For future reference, this line doesn't error out with the current tests.

(A detailed commit message will be added later)

…-cache-entry-rebased DESCRIPTION: Introduce citus local tables The commits in this pr are merged from other sub-pr's: * community/#3852: Brings lazy&fast table creation logic for create_citus_local_table udf * community/#3995: Brings extended utility command support for citus local tables * community/#4133: Brings changes in planner and in several places to integrate citus local tables into our distributed execution logic We are introducing citus local tables, which a new table type to citus. To be able to create a citus local table, first we need to add coordinator as a worker node. Then, we can create a citus local table via SELECT create_citus_local_table(<tableName>). Calling this udf from coordinator will actually create a single-shard table whose shard is on the coordinator. Also, from the citus metadata perspective, for citus local tables: * partitionMethod is set to DISTRIBUTE_BY_NONE (like reference tables) and * replicationModel is set to the current value of citus.replication_model, which already can't be equal to REPLICATION_MODEL_2PC, which is only used for reference tables internally. Note that currently we support creating citus local tables only from postgres tables living in the coordinator. That means, it is not allowed to execute this udf from worker nodes or it is not allowed to move shard of a citus local table to any other nodes. Also, run-time complexity of calling create_citus_local_table udf does not depend on the size of the relation, that means, creating citus local tables is actually a non-blocking operation. This is because, instead of copying the data to a new shard, this udf just does the following: * convert input postgres table to the single-shard of the citus local table by suffixing the shardId to it's name, constraints, indexes and triggers etc., * create a shell table for citus local table in coordinator and in mx-worker nodes when metadata sycn is enabled. * create necessary objects on shell table. Here, we should also note we can execute queries/dml's from mx worker nodes as citus local tables are already first class citus tables. Even more, we brought trigger support for citus local tables. That means, we can define triggers on citus local tables so that users can define trigger objects to perform execution of custom functions that might even modify other citus tables and other postgres tables. Other than trigger support, citus local tables can also be involved in foreign key relationships with reference tables. Here the only restriction is, foreign keys from reference tables to citus local tables cannot have behaviors other than RESTRICT & NO ACTION behavior. Other than that, foreign keys between citus local tables and reference tables just work fine. All in all, citus local tables are actually just local tables living in the coordinator, but natively accessible from other nodes like other first class citus tables and this enables us to set foreign keys constraints between very big coordinator tables and reference tables without having to do any data replication to worker nodes for local tables.

onurctirtir self-assigned this May 20, 2020

onderkalaci reviewed May 27, 2020

View reviewed changes

onurctirtir force-pushed the single-placement-table/udf2 branch from 8eb5401 to 4b6ae1e Compare May 31, 2020 20:13

onurctirtir changed the base branch from single-placement-table/master to copy_paste_pg_get_triggerdef May 31, 2020 20:13

onurctirtir force-pushed the single-placement-table/udf2 branch from 4b6ae1e to 7cf6515 Compare May 31, 2020 21:25

Base automatically changed from copy_paste_pg_get_triggerdef to master June 1, 2020 07:35

onurctirtir force-pushed the single-placement-table/udf2 branch from 7cf6515 to 96452cb Compare June 1, 2020 07:40

onurctirtir changed the base branch from master to single-placement-table/master June 1, 2020 07:42

onurctirtir marked this pull request as ready for review June 1, 2020 07:43

onurctirtir requested a review from SaitTalhaNisanci June 1, 2020 07:43

onurctirtir force-pushed the single-placement-table/udf2 branch 4 times, most recently from d08e70c to 8476a4d Compare June 1, 2020 12:47

SaitTalhaNisanci reviewed Jun 2, 2020

View reviewed changes

onurctirtir force-pushed the single-placement-table/udf2 branch 2 times, most recently from 28479b3 to 1615b56 Compare June 4, 2020 09:35

onderkalaci reviewed Jun 4, 2020

View reviewed changes

onurctirtir force-pushed the single-placement-table/udf2 branch 6 times, most recently from f5210fd to 8710dd0 Compare June 4, 2020 22:30

marcocitus reviewed Jun 5, 2020

View reviewed changes

onurctirtir mentioned this pull request Jun 5, 2020

Append IF NOT EXISTS to deparsed CREATE SERVER commands #3875

Merged

onderkalaci reviewed Jun 5, 2020

View reviewed changes

SaitTalhaNisanci reviewed Jun 5, 2020

View reviewed changes

onderkalaci reviewed Jun 5, 2020

View reviewed changes

onurctirtir force-pushed the single-placement-table/udf2 branch from b2cbffa to e8ec5bc Compare June 5, 2020 16:33

onurctirtir added 5 commits June 23, 2020 01:19

log commands that are executed

37341cd

rename triggers on shard relation

736877a

address some other reviews

12fcc53

improve tests for triggeers

39df007

add test for sequences

da74ce7

onurctirtir force-pushed the single-placement-table/udf2 branch from 6af50b8 to da74ce7 Compare June 23, 2020 08:09

onurctirtir added 2 commits June 23, 2020 11:25

fix access to pg_trigger for pg11

6fa943e

Revert "update normalize rule"

e40826d

This reverts commit 48480d3.

onurctirtir added 2 commits June 23, 2020 12:28

improve comment for getOwnedSequences

025a428

remove redundant check in IsCitusLocalTable

66954e3

SaitTalhaNisanci approved these changes Jun 25, 2020

View reviewed changes

onurctirtir added 4 commits June 25, 2020 16:51

address reviews

10c5a47

address

1753a5f

fix normalization rule

faf6ee6

fix test output

1e2995a

SaitTalhaNisanci reviewed Jun 25, 2020

View reviewed changes

onurctirtir merged commit 1824e2f into single-placement-table/master Jun 25, 2020

onurctirtir added a commit that referenced this pull request Jun 25, 2020

Implement "citus local table" creation logic (#3852)

72d3ba3

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Jun 30, 2020

Implement "citus local table" creation logic (#3852)

6c93207

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Jul 7, 2020

Implement "citus local table" creation logic (#3852)

177fea3

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Jul 29, 2020

Implement "citus local table" creation logic (#3852)

4502cb3

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Aug 14, 2020

Implement "citus local table" creation logic (#3852)

9b536e2

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Aug 31, 2020

Implement "citus local table" creation logic (#3852)

f06fb67

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Sep 1, 2020

Implement "citus local table" creation logic (#3852)

e24df3f

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Sep 3, 2020

Implement "citus local table" creation logic (#3852)

7a245ee

(A detailed commit message will be added later)

onurctirtir added a commit that referenced this pull request Sep 7, 2020

Implement "citus local table" creation logic (#3852)

2b6ec2a

(A detailed commit message will be added later)

onurctirtir mentioned this pull request Sep 8, 2020

Introduce citus local tables #4143

Merged

onurctirtir deleted the single-placement-table/udf2 branch September 15, 2020 14:11


		EnsureTableOwner(relationOid);

		/* we allow creating citus local tables only from relations with RELKIND_RELATION */


		FinalizeCitusLocalTableCreation(relationId);

		relation_close(relation, ExclusiveLock);


		SELECT create_citus_local_table('citus_local_table');

		-- this should fail



		/*
		* ErrorIfCoordinatorNotAddedAsWorkerNode error out if coordinator is not added

Implement "citus local table" creation logic and a udf #3852

Implement "citus local table" creation logic and a udf #3852

Conversation

onurctirtir commented May 20, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

onurctirtir Jun 1, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

SaitTalhaNisanci left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

onderkalaci left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

onurctirtir Jun 5, 2020 • edited

Choose a reason for hiding this comment

onderkalaci Jun 7, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

onurctirtir Jun 8, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

onurctirtir commented Jun 23, 2020

SaitTalhaNisanci left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

onurctirtir commented May 20, 2020 •

edited

onurctirtir Jun 1, 2020 •

edited

onurctirtir Jun 5, 2020 •

edited

onderkalaci Jun 7, 2020 •

edited

onurctirtir Jun 8, 2020 •

edited