Commit Graph

1323 Commits (a658f5ecdaeac0d65df9d17fdcd4ab72912e898c)

Author SHA1 Message Date
Önder Kalacı 4a09bb4948 Merge pull request #1516 from citusdata/improve_lw_locks
Improve locking semantics for the backend management
2017-08-09 17:33:15 +03:00
Onder Kalaci b5ea3ab6a3 Improve locking semantics for backend management
We use the backend shared memory lock for preventing
new backends to be part of a new distributed transaction
or an existing backend to leave a distributed transaction
while we're reading the all backends' data.

The primary goal is to provide consistent view of the
current distributed transactions while doing the
deadlock detection.
2017-08-09 17:17:12 +03:00
Brian Cloutier 2e0916e15a Add master_add_secondary_node() UDF 2017-08-09 17:10:48 +03:00
Marco Slot 3f338a3fc6 Merge pull request #1407 from citusdata/fix_node_locking
Fix pg_dist_node locking
2017-08-09 16:32:28 +04:00
Marco Slot 08ed6d8269 Prevent pg_dist_node changes during master_create_empty_shard 2017-08-09 14:22:09 +02:00
Murat Tuncer 5cb9466255 Rebase node metadata isolation tests 2017-08-09 14:22:09 +02:00
Marco Slot 3a0571e69b Remove LockMetadataSnapshot 2017-08-09 14:09:54 +02:00
Marco Slot ad0fdf57ca Add add/remove node rollback isolation tests 2017-08-09 14:09:54 +02:00
Marco Slot c2f8bafa05 Fix shard creation vs. pg_dist_node change locking 2017-08-09 14:09:54 +02:00
Marco Slot 868ee6be83 Fix and simplify pg_dist_node locking 2017-08-09 14:09:54 +02:00
Burak Yücesoy 01d8926228 Merge pull request #1509 from citusdata/create_distributed_partitions
Add support for distributed partitioned tables
2017-08-09 14:27:10 +03:00
Burak Yucesoy ab5f97861b Add regression tests for distributed partitioned tables 2017-08-09 10:01:35 +03:00
Burak Yucesoy 8455d1a4ef Ensure we are allowing partitioned tables at all appropriate places 2017-08-09 10:01:35 +03:00
Burak Yucesoy 2eee556738 Add distributed partitioned table support for COPY
For partitioned tables, PostgreSQL opens partition and its partitions
in BeginCopyFrom and it expects its caller to close those relations.
However, we do not have quick access to opened relations and performing
special operations for partitioned tables isn't necessary in coordinator
node. Therefore before calling BeginCopyFrom, we change relkind of those
partitioned tables to RELKIND_RELATION. This prevents PostgreSQL to open
its partitions as well.
2017-08-09 10:01:35 +03:00
Burak Yucesoy 31f3221342 Add distributed partitioned table support to router plannable queries
In standart_planner, PostgreSQL expands partitioned tables to their
partitions and call our restriction hook for each partition. It also,
for some queries, skips the partitioned table itself completely. This
behaviour makes it difficult to prune shards and decide whether query
is router plannable or not. To prevent this behaviour, we change inh
flag of partitioned tables to false in the query tree. In this case,
PostgreSQL treats those partitioned tables as regular relations and
does not expand them.

This behaviour is inline with our expectations, because we do not want
to treat partitioned tables differently on coordinator. Although we are
not entirely comfortable with modifying query tree, other solutions to
this problem is overly complicated.
2017-08-09 10:01:35 +03:00
Burak Yucesoy fddf9b3fcc Add distributed partitioned table support distributed table creation
With this PR, Citus starts to support all possible ways to create
distributed partitioned tables. These are;

- Distributing already created partitioning hierarchy
- CREATE TABLE ... PARTITION OF a distributed_table
- ALTER TABLE distributed_table ATTACH PARTITION non_distributed_table
- ALTER TABLE distributed_table ATTACH PARTITION distributed_table

We also support DETACHing partitions from partitioned tables and propogating
TRUNCATE and DDL commands to distributed partitioned tables.

This PR also refactors some parts of distributed table creation logic.
2017-08-09 10:01:35 +03:00
Metin Döşlü a650aaa631 Merge pull request #1483 from citusdata/update_subquery_on_where_false
Add support for router UPDATEs and DELETEs with subqueries and joins
2017-08-08 22:24:41 +03:00
Metin Doslu b8a9e7c1bf Add support for UPDATE/DELETE with subqueries 2017-08-08 21:35:08 +03:00
Marco Slot 518750bd51 Merge pull request #1519 from citusdata/connection_per_placement
Avoid connections that accessed non-colocated placements in multi-shard commands
2017-08-08 20:47:16 +04:00
Marco Slot d3e9746236 Avoid connections that accessed non-colocated placements in multi-shard commands 2017-08-08 18:32:34 +02:00
Brian Cloutier 7060ade6fe GetNodeTuple returns NULL it node does not exist
It never throws an error.
2017-08-08 13:12:06 +03:00
Brian Cloutier a3e9bef685 All users of WorkerNodeHash take an AccessShareLock
The metadata cache simulates a SELECT on pg_dist_node. Now the locks it
takes also simulate that SELECT.
2017-08-08 13:12:06 +03:00
Brian Cloutier 5914c992e6 cluster management UDFs see nodes in different clusters
- master_activate_node and master_disable_node correctly toggle
  isActive, without crashing
- master_add_node rejects duplicate nodes, even if they're in different
  clusters
- master_remove_node allows removing nodes in different clusters
2017-08-08 13:12:06 +03:00
Brian Cloutier 3151b52a0b Add citus.cluster_name GUC
- Nodes with a nodecluster which does not match citus.cluster_name
  are excluded from the metadata cache and never seen by another part of
  Citus.
2017-08-08 13:12:06 +03:00
Brian Cloutier 94947c0d54 Refactor: ReplicateShardToAllWorkers more explicitly locks pg_dist_node 2017-08-08 13:12:06 +03:00
Brian Cloutier f87fefa323 Refactor: DistributedTableSize more explicitly only locks pg_dist_node 2017-08-08 13:12:06 +03:00
Brian Cloutier 3769381366 Fix inaccurate comment on SetNodeState 2017-08-08 13:12:06 +03:00
Brian Cloutier bf197e9f0c Add test for super-long cluster names 2017-08-08 11:18:31 +03:00
Brian Cloutier fbecf48a03 Disallow adding primary nodes to non-default clusters 2017-08-08 11:18:31 +03:00
Brian Cloutier 5618e69386 Add pg_dist_node.nodecluster 2017-08-08 11:18:31 +03:00
Brian Cloutier 74ce4faab5 Make multi_cluster_management test more stable 2017-08-08 11:18:31 +03:00
Brian Cloutier e7846ba7d1 Allow metadata sync functions on secondaries
{start,stop}_metadata_sync_to_node now toggle the hasMetadata flag when
run on secondaries but don't attempt to actually sync any metadata.
2017-08-07 18:46:51 +03:00
Marco Slot 7e4b2c1595 Merge pull request #1474 from citusdata/update_where_false
Execute UPDATE/DELETE statements with 0 shards
2017-08-07 17:48:52 +04:00
Marco Slot 4cc7c36596 Simplify metadata lock acquisition for DML 2017-08-07 15:36:58 +02:00
Marco Slot aa7ca81548 Execute UPDATE/DELETE statements with 0 shards 2017-08-07 15:36:58 +02:00
Marco Slot 3248f1a2b7 Merge pull request #1506 from citusdata/descend_function_evaluation
Function evaluation descends into expression trees
2017-08-06 22:24:22 +04:00
Marco Slot bac60bb64f Function evaluation descends into expression trees 2017-08-06 19:53:05 +02:00
Brian Cloutier 37985de85e master_disable_node no longer crashes when given a non-existant node 2017-08-04 11:14:54 +03:00
Hadi Moshayedi 8229a64fe8 Remove distributed tables' dependency on distribution key columns. (#1527)
This change removes distributed tables' dependency on distribution key columns. We already check that we cannot drop distribution key columns in ErrorIfUnsupportedAlterTableStmt() at multi_utility.c, so we don't need to have distributed table to distribution key column dependency to avoid dropping of distribution key column.

Furthermore, having this dependency causes some warnings in pg_dump --schema-only (See #866), which are not desirable.

This change also adds check to disallow drop of distribution keys when citus.enable_ddl_propagation is set to false. Regression tests are updated accordingly.
2017-08-03 10:07:04 -04:00
Murat Tuncer fa18899cf9 Remove serialization/deserialization of multiplan node (#1477)
introduces copy functions for Citus MultiPlan nodes.
uses ExtensibleNode mechanism to store MultiPlan data
drops serialiazation of MultiPlans
2017-08-02 08:24:00 +03:00
Burak Yücesoy f0275fe4ae Merge pull request #1525 from citusdata/refactor_create_distributed_table
Refactor distributed table creation logic
2017-07-31 12:08:18 +03:00
Burak Yucesoy 37b200a52e Fix broken isolation tests
We try to run our isolation tests paralles as much as possible. In
some of those isolation tests we used same table name which causes
problem while running them in paralles. This commit changes table
names in those tests to ensure tests can run in parallel.
2017-07-31 11:11:49 +03:00
Burak Yucesoy 7769f1d012 Refactor distributed table creation logic
This commit is preperation for introducing distributed partitioned
table support. We want to clean and refactor some code in distributed
table creation logic so that we can handle partitioned tables in more
robust way.
2017-07-31 11:11:23 +03:00
Murat Tuncer 520d74b96d Add a regression test for citus.max_task_string_size (#1524) 2017-07-28 10:49:09 -07:00
Brian Cloutier 7d8bcb6a88 These tests sometimes deadlock on travis 2017-07-28 16:02:43 +03:00
Brian Cloutier b20a086a8f master_activate_node UDF also returns noderole 2017-07-28 16:02:43 +03:00
Murat Tuncer 26f020dc6e Make maxTaskStringSize configurable (#1501)
maxTaskStringSize determines the size of worker query string.
It was originally hard coded to a specific value. This has caused
issues at some users. Since it determines initial shared memory
allocation, we did not want to set it to an arbitrary higher number.
Instead made it configurable.

This commit introduces a new GUC variable max_task_string_size

Changes in this variable requires restart to be in effect.
2017-07-27 11:39:12 -07:00
Önder Kalacı 6698ca8d9e Merge pull request #1523 from citusdata/deadlock_detection_main
Convert the global wait edges to adjacency lists
2017-07-27 20:14:49 +03:00
Onder Kalaci 6132d17481 Convert global wait edges to adjacency list
In this commit, we add ability to convert global wait edges
into adjacency list with the following format:
 [transactionId] = [transactionNode->waitsFor {list of waiting transaction nodes}]
2017-07-27 19:53:51 +03:00
Murat Tuncer 8729b7d55a Use cstore_table_size function to determine cstore table size (#1521)
pg_table_size/pg_relation_size variants always return 0 for
cstore tables. We should be using cstore_table_size function
for cstore_tables.
2017-07-27 09:02:07 -07:00