Commit Graph

3744 Commits (fix-repartition-join-single-node)

Author SHA1 Message Date
Andres Freund a6150c2916 Lower "waiting for activity on tasks took longer than" log level.
It's perfectly normal to wait longer in several circumstances, and the
output can lead to spurious regression output changes.
2016-10-03 13:07:43 -07:00
Marco Slot e6ecbc2063 Merge pull request #831 from citusdata/ultimate_citus_improvement
Change logicalrelid type to regclass
2016-10-03 20:34:57 +02:00
Marco Slot a4efb60b54 Change logicalrelid type in pg_dist_partition and pg_dist_shard to regclass 2016-10-03 20:27:16 +02:00
Marco Slot fa2f5087ad Merge pull request #832 from citusdata/bugfix/remove_eventinvoke_trigger
Remove EventInvokeTrigger from regression test output
2016-10-03 20:26:19 +02:00
Marco Slot fc93974238 Remove EventInvokeTrigger from regression test output 2016-10-03 20:21:15 +02:00
Marco Slot 3d1e2c1d3a Merge pull request #819 from robin900/handle-repartitions-by-typname
During repartitions use partitionColumnType as ::regtype so that UDTs work
2016-10-03 19:50:45 +02:00
Robin Thomas c507a0df1c During repartitions, the partitionColumnType argument sent to workers
is now a `::regtype` using the qualified name of the column type,
not the column type OID which may differ between master/worker nodes.
Test coverage of a hash reparitition using a UDT as the join column.

Note that the UDFs `worker_hash_partition_table` and `worker_range_partition_table`
are unchanged, and rightly expect an OID for the column type; but the
planner code building the commands now allows for `::regtype` casting
to do its magic.

Fixes citusdata/citus#111.
2016-10-03 13:41:20 -04:00
Marco Slot 4d60aa2d53 Merge pull request #808 from robin900/partial-index-tests
Added test coverage for partial unique indexes, exclusion constraints
2016-10-03 17:32:16 +02:00
Robin Thomas b1493e299e Added test coverage for partial unique indexes and exclude constraints. 2016-10-03 10:47:30 -04:00
Eren Başak 83ef3d0820 Merge pull request #825 from citusdata/fix_command_counter_increment_wrong_place
Fix command counter increment bug
2016-10-03 17:16:58 +03:00
Eren Basak ac3a4eee21 Fix command counter increment bug
Fixes citusdata/citus#714

On `InsertShardRow`, we previously called `CommandCounterIncrement()` before
`CitusInvalidateRelcacheByRelid(relationId);`. This might prevent to skip
invalidation of the distributed table in the next access within the same session.
2016-10-03 17:00:27 +03:00
Eren Başak 7e7b0f3491 Merge pull request #742 from citusdata/feature/task_tracker_folders
Differentiate worker and master job temporary folders - MX Backport
2016-10-03 14:29:54 +03:00
Onder Kalaci a533b8e7c1 Differentiate worker and master job temporary folders
This commit enables to create different worker and master temporary folders.
This change is important for citus-mx on task-tracker execution. In simple words,
on citus-mx, the worker could actually be reponsible for the master tasks as well.
Prior to this change, both master and worker logic on task-tracker executor was
accessing and using the same files for different purposes which was dangerous on
certain cases (i.e., when task_tracker_delay is low).
2016-10-03 14:24:08 +03:00
Jason Petersen a8701841b5 Merge pull request #824 from citusdata/use_lock_tranches
Move task tracker lwlocks into their own tranche

cr: @anarazel
2016-09-30 16:11:11 -06:00
Andres Freund 77efe7fcd4
Move task tracker lwlocks into their own tranche.
RequestAddinLWLocks()/LWLockAssign() are gone in 9.6. Luckily all citus
supported postgres versions support tranches, so use those.
2016-09-30 16:06:49 -06:00
Jason Petersen d52ebffce5 Merge pull request #823 from citusdata/update_postgresql_files
Update PostgreSQL-sourced files with latest changes

cr: @anarazel
2016-09-30 16:06:28 -06:00
Jason Petersen f59cf2b818
Remove references to 9.4
Some still lingered.
2016-09-29 17:35:19 -06:00
Jason Petersen 37631cd132
Remove alternate multi_hash test file
This was made irrelevant by Citus v5.1.0.
2016-09-29 16:43:19 -06:00
Jason Petersen 3046c2b62c
Remove references to PostgreSQL 9.4 support files
No longer extant.
2016-09-29 15:54:38 -06:00
Jason Petersen 5634b027b5
Remove gitattributes for csql files
This was missed before.
2016-09-29 15:54:38 -06:00
Jason Petersen 6671cf5171
Remove unused dumputils.h header
Believe this was used by csql, which is now gone.
2016-09-29 15:54:38 -06:00
Jason Petersen 1c560dfa9c
Update ruleutils_95 with latest PostgreSQL changes
Hand-applied changes from a diff I generated between 9.5.0 and 9.5.4.
2016-09-29 15:54:38 -06:00
Marco Slot 9a5b844a81 Merge pull request #815 from citusdata/bugfix/count_null
Make count return 0 if all shards are pruned away
2016-09-29 20:32:46 +02:00
Marco Slot c4bc0742a7 Make count return 0 if all shards are pruned away
Before this change, count on a distributed returned NULL if all shards
were pruned away, because on the master we replace with count(..) call
with a sum(..) call to sum the counts from the shards. However, sum
returns NULL when there are no rows, whereas count is expected to return
0.
2016-09-29 20:27:26 +02:00
Jason Petersen 9c19ad8b78 Merge pull request #818 from citusdata/fix_xact_callbacks
Directly register transaction callbacks in PG_init

cr: @anarazel
2016-09-29 11:52:03 -06:00
Jason Petersen 5b80d4e8dd
Directly register multi-shard callbacks in PG_init
I had changed these callbacks to use the same method I chose for the
router executor (for consistency), but as that method is flawed, we now
want to ensure we directly register them from PG_init as well.
2016-09-29 11:43:19 -06:00
Jason Petersen 5f6264105d
Directly register router xact callbacks in PG_init
Not entirely sure why we went with the shared memory hook approach, but
it causes problems (multiple registration) during crashes. Changing to
a simple direct registration call from PG_init.
2016-09-29 11:43:18 -06:00
Burak Yücesoy e2f720dbe7 Merge pull request #785 from citusdata/colocation_features
Internal co-location API
2016-09-29 12:43:02 +03:00
Burak Yucesoy 1ee39eb098 Internal co-location API
With this commit we introduce internal API for co-location related operations.
2016-09-29 11:56:53 +03:00
Jason Petersen 127c0e513f Merge pull request #814 from citusdata/copy_to_distributed_table_final_goodbye
Remove copy_to_distributed_table

cr: @jasonmp85
2016-09-28 11:32:08 -06:00
Marco Slot 5cdbe2b86c
Remove copy_to_distributed_table 2016-09-28 11:27:54 -06:00
Murat Tuncer 854ed613fb Merge pull request #806 from citusdata/fix_805_where_false
Make join queries with where false clauses router plannable
2016-09-28 18:54:19 +03:00
Murat Tuncer 5b42318ac4 Make where false queries router plannable 2016-09-28 18:49:26 +03:00
Murat Tuncer 116fcdbcc8 Merge pull request #784 from citusdata/fix_594_shard_refresh
Add UDF master_expire_table_cache
2016-09-28 12:13:01 +03:00
Murat Tuncer c16dec88c3 Add UDF master_expire_table_cache 2016-09-28 12:08:37 +03:00
Jason Petersen e8a942485c Merge pull request #812 from citusdata/fix_uniq_constraint_segfault
Fix unique-violation-in-xact segfault

cr: @anarazel
2016-09-27 16:51:48 -06:00
Jason Petersen 0caf0d95f1
Fix unique-violation-in-xact segfault
An interaction between ReraiseRemoteError and DML transaction support
causes segfaults:

  * ReraiseRemoteError calls PurgeConnection, freeing a connection...
  * That connection is still in the xactParticipantHash

At transaction end, the memory in the freed connection might happen to
pass the "is this connection OK?" check, causing us to try to send an
ABORT over that connection. By removing it from the transaction hash
before calling ReraiseRemoteError, we avoid this possibility.
2016-09-27 16:44:03 -06:00
Metin Döşlü 1a04cd123d Merge pull request #770 from citusdata/fix/insert_query_inside_plpgsql
Pass text oid instead of invalid oid for null values
2016-09-27 08:46:04 +03:00
Metin Doslu c9dcad9b05 Pass text oid inteads of invalid oid for null values
Passing invalid oids even for null values in PQsendQueryParams() causes worker
nodes to fail. Therefore, we pass text oid for null values.
2016-09-27 08:15:46 +03:00
Jason Petersen dbc58c60ca Merge pull request #776 from citusdata/feature/no-movement
Support NoMovement direction in router executor

cr: @jasonmp85
2016-09-26 18:32:33 -06:00
Andres Freund 776b3868b9
Support NoMovement direction in router executor
This is mainly interesting because it allows to use RETURN QUERY/RETURN
QUERY EXECUTE and FOR ... IN .. LOOPs in plpgsql.
2016-09-26 18:28:36 -06:00
Jason Petersen 33552c2f13 Merge pull request #721 from citusdata/feature_truncate
Add truncate support for distributed tables

cr: @jasonmp85
2016-09-26 18:27:52 -06:00
Murat Tuncer 32003c4aa1
Add tests with spaces in table names 2016-09-26 18:23:43 -06:00
Murat Tuncer 2f78fb8f1b
Remove extra space 2016-09-26 18:23:43 -06:00
Murat Tuncer 902e68c9ef
Refactor SendQueryToPlacements api 2016-09-26 18:23:43 -06:00
Murat Tuncer 6317bbe9a8
Address feedback 2016-09-26 18:23:42 -06:00
Murat Tuncer 877694296f
Fix regression test failures after rebase 2016-09-26 18:23:42 -06:00
Murat Tuncer 2eec0167be
Add support for truncate statement 2016-09-26 18:23:42 -06:00
Marco Slot ad9d0bb69c Merge pull request #804 from citusdata/bugfix/join_filter_crash
Fix segmentation fault in case of joins with WHERE false
2016-09-26 15:23:57 +02:00
Marco Slot 3318288d75 Fix segmentation fault in case of joins with WHERE 1=0 2016-09-26 15:12:29 +02:00