Commit Graph

815 Commits (temp_tables)

Author SHA1 Message Date
Marco Slot 6fc545c87b Support temporary distributed tables 2017-08-25 11:56:46 +02:00
Marco Slot 0aadbb1760 Convert multi-row INSERT target list to Vars 2017-08-25 10:55:56 +02:00
Marco Slot 1920390688 Multi-row INSERTs no longer throw errors in isolation tests 2017-08-25 10:55:56 +02:00
Marco Slot ae00795dab Allow default columns in multi-row INSERTs 2017-08-25 10:55:56 +02:00
Marco Slot c97692f382 Fix multi-row INSERT with RETURNING on reference tables 2017-08-24 10:42:12 +02:00
Marco Slot dbf18df995 Don't error out if BuildGlobalWaitGraph fails to connect 2017-08-23 19:08:03 +02:00
Burak Yucesoy 5be6eb9ef6 Increase coverage of isolation tests - Part 2
With this PR we add isolation tests for

COPY to reference table vs. other operations
COPY to partitioned table vs. other operations
Multi row INSERTs vs other operations
INSERT/SELECT vs. other operations
UPSERT vs. other operations
DELETE vs. other operations
TRUNCATE vs. other operations
DROP vs. other operations
DDL vs. other operations

other operations consist of basic SQL operations (like SELECT,
INSERT, DELETE, UPSERT, COPY TRUNCATE, CREATE INDEX) as well
as some Citus functionalities (like master_modify_multiple_shards,
master_apply_delete_command, citus_total_relation_size etc.)
2017-08-23 18:23:36 +03:00
Onder Kalaci c7bb29b69e Prevent maintanince deamon crashes due to dead processes
If after the distributed deadlock detection decides to cancel
a backend, the backend has been terminated/killed/cancelled
externally, we might be accessing to a NULL pointer. This commit
prevents that case by ignoring the current distributed deadlock.
2017-08-23 15:44:09 +03:00
Marco Slot 641420d79f Remove source node argument from dump_local_wait_edges 2017-08-23 13:14:00 +02:00
Jason Petersen 8cb69e3a14 Add alias for target in multi-row INSERTs
This is necessary for multi-row INSERTs for the same reasons we use it
in e.g. UPSERTs: if the range table list has more than one entry, then
PostgreSQL's deparse logic requires that vars be prefixed by the name
of their corresponding range table entry. This of course doesn't affect
single-row INSERTs, but since multi-row INSERTs have a VALUE RTE, they
were affected.

The piece of ruleutils which builds range table names wasn't modified
to handle shard extension; instead UPSERT/INSERT INTO ... SELECT added
an alias to the RTE. When present, this alias is favored. Doing the
same in the multi-row INSERT case fixes RETURNING for such commands.
2017-08-23 10:24:00 +02:00
Marco Slot 4d7927b672 Execute multi-row INSERTs sequentially 2017-08-23 10:04:57 +02:00
Marco Slot cf375d6a66 Consider dropped columns that precede the partition column in COPY 2017-08-22 13:02:35 +02:00
Marco Slot bd6bf29983 Don't add procs multiple times in BuildWaitGraphForSourceNode 2017-08-21 16:48:30 +02:00
Onder Kalaci 6532b69873 Kill the maintenance daemon on DROP DATABASE 2017-08-18 16:03:08 +03:00
Metin Doslu 0d052e9864 Fix a crash on zero-shard tables 2017-08-18 13:53:59 +03:00
Önder Kalacı b82f886ad3 Merge branch 'master' into improve_deadlock_detection 2017-08-18 13:07:18 +03:00
Marco Slot 7523753a73 Clear metadata OID cache prior to deadlock detection 2017-08-18 11:20:24 +02:00
Andres Freund b936bde936 Take AccessShareLock on the extension prior to running deadlock detection 2017-08-18 11:20:24 +02:00
Onder Kalaci 20679c9e8b Relax assertion on deadlock detection considering
self deadlocks.
2017-08-18 11:16:38 +03:00
Onder Kalaci 550a5578d8 Skip deadlock detection on the workers
Do not run distributed deadlock detection
on the worker nodes to prevent errornous
decisions to kill the deadlocks.
2017-08-17 19:43:38 +03:00
Burak Yucesoy ae32d786cf Add new isolation tests 2017-08-17 17:46:03 +03:00
Marco Slot 1eca53ad40 Exit maintenanced on database crash 2017-08-16 18:29:44 +02:00
Marco Slot 9e7b1fb858 Return readable nodes in master_get_active_worker_nodes 2017-08-16 11:28:47 +02:00
Hadi Moshayedi e5fbcf37dd Add Savepoint Support (#1539)
This change adds support for SAVEPOINT, ROLLBACK TO SAVEPOINT, and RELEASE SAVEPOINT.

When transaction connections are not established yet, savepoints are kept in a stack and sent to the worker when the connection is later established. After establishing connections, savepoint commands are sent as they arrive.

This change fixes #1493 .
2017-08-15 13:02:28 -04:00
Onder Kalaci 205501532a Add version check to the maintenance daemon
We should prevent running the deadlock detection if
there is a major version change. Otherwise, the daemon
may access to obsolete metadata catalog tables.
2017-08-15 18:47:13 +03:00
Marco Slot 3ff46245b3 Make sure we don't use 2PC in copy from worker 2017-08-15 13:44:20 +02:00
Marco Slot 4614814de1 Enable 2PC for INSERT...SELECT via coordinator 2017-08-15 13:44:20 +02:00
Marco Slot fa70089766 Enable 2PC during distributed table creation 2017-08-15 13:44:20 +02:00
Marco Slot 9232823070 Abort on failure on master connection during copy from worker 2017-08-15 13:44:20 +02:00
Marco Slot df7723cde5 Should not commit on aborted non-critical connections 2017-08-15 13:44:20 +02:00
Eren Başak 77626c4238 Fix NULL nodeClusterString crush on pg_worker_list.conf migrations 2017-08-14 18:13:53 +03:00
Eren Başak b3d2f9ba71 Fix pg_worker_list use-after-free bug
This change fixes a use-after-free bug while renaming obsolete
`pg_worker_list.conf` file, which causes Citus to crash during upgrade
(or even extension creation) if `pg_worker_list.conf` exists.
2017-08-14 18:13:53 +03:00
Burak Yucesoy 45b273321f Add tests for locking operations on partitioned tables 2017-08-14 14:55:45 +03:00
Burak Yucesoy dfdfb44ebf Acquire shard resource locks on parent tables while operating on partitions 2017-08-14 14:44:30 +03:00
Burak Yucesoy a321e750c0 Acquire relation locks on partitions while operation on parent table 2017-08-14 14:44:30 +03:00
Burak Yucesoy 52b9e35d50 Add relationIdList field to the Job struct 2017-08-14 14:06:22 +03:00
Onder Kalaci 4f668ad38b Make the test outputs consistent
by using VACUUM ANALYZE on the tables.
2017-08-12 13:29:25 +03:00
Onder Kalaci 0ba2f9e4e4 Add regression tests for distributed deadlock detection 2017-08-12 13:29:25 +03:00
Onder Kalaci 5b48de7430 Improve deadlock detection for MX
We added a new field to the transaction id that is set to true only
for the transactions initialized on the coordinator. This is only
useful for MX in order to distinguish the transaction that started
the distributed transaction on the coordinator where we could
 have the same transactions' worker queries on the same node.
2017-08-12 13:28:37 +03:00
Onder Kalaci 59133415b0 Add logging infrasture for distributed deadlock detection
We added a new GUC citus.log_distributed_deadlock_detection
which is off by default. When set to on, we log some debug messages
related to the distributed deadlock to the server logs.
2017-08-12 13:28:37 +03:00
Onder Kalaci e5d5bdff51 Enable distributed deadlock detection on the maintenance deamon
With this commit, the maintenance deamon starts to check for
distributed deadlocks.

We also introduced a GUC variable (distributed_deadlock_detection_factor)
whose value is multiplied with Postgres' deadlock_timeout. Setting
it to -1 disables the distributed deadlock detection.
2017-08-12 13:28:37 +03:00
Onder Kalaci 66936053a0 Improve error messages when a backend is cancelled by deadlock detection
We send SIGINT to a backend that is cancelled due to a deadlock. That
approach ends up being a very confusing error message.

With this commit we intercept the error messages and show a more
meaningful error message to the user.
2017-08-12 13:28:37 +03:00
Onder Kalaci be4fc45c03 Deprecate enable_deadlock_prevention flag
Now that we already have the necessary infrastructure for detecting
distributed deadlocks. Thus, we don't need enable_deadlock_prevention
which is purely intended for preventing some forms of distributed
deadlocks.
2017-08-12 13:28:37 +03:00
Onder Kalaci a333c9f16c Add infrastructure for distributed deadlock detection
This commit adds all the necessary pieces to do the distributed
deadlock detection.

Each distributed transaction is already assigned with distributed
transaction ids introduced with
3369f3486f. The dependency among the
distributed transactions are gathered with
80ea233ec1.

With this commit, we implement a DFS (depth first seach) on the
dependency graph and search for cycles. Finding a cycle reveals
a distributed deadlock.

Once we find the deadlock, we examine the path that the cycle exists
and cancel the youngest distributed transaction.

Note that, we're not yet enabling the deadlock detection by default
with this commit.
2017-08-12 13:28:37 +03:00
Marco Slot 59e626d158 Add regression tests for follower clusters 2017-08-12 12:05:56 +02:00
Marco Slot 55992d4bc0 Disallow task-tracker queries on follower clusters 2017-08-12 11:47:31 +02:00
velioglu 100739f62a Change citus subversion 2017-08-11 11:57:57 +03:00
Marco Slot 53584affa8 Fix locking in create_distributed_table 2017-08-11 11:34:33 +03:00
velioglu 7c65001e23 Do not delete row from colocation table within drop table 2017-08-11 11:34:33 +03:00
velioglu b0efffae1c Correct planner and add more tests 2017-08-11 10:16:13 +03:00