It hasn't done anything since 2010. If I'm reading the commit log
correctly, it was added and deprecated only a few months apart, and
probably hasn't done anything in any released version.

Originally, we don't hold AccessExclusiveLock while renaming relations to provide
supportability for altering relation name while directly modifying system catalogs.

However, this breaks the isolation semantics of ALTER TABLE. Hence, we add the lock
to make it consistent with upstream behavior.  The original behavior is now
possible with the GUC.
Signed-off-by: NXin Zhang <xzhang@pivotal.io>

There is already a preset GUC for the PostgreSQL version number in
machine readable form: server_version_num, and human readable form:
server_version. This adds a Greenplum counterparts for these as
gp_server_version_num and gp_server_version.

Getting mirror or primary in change tracking and recovering the same is done in
many other tests, no point testing the same this old way using guc to inject
fault.

QE reader leverages SharedLocalSnapshot to perform visibility checks. QE writer
is responsible to keep the SharedLocalSnapshot up to date. Before this fix,
SharedLocalSnapshot was only updated by writer while acquiring the snapshot. But
if transaction id is assigned to subtransaction after it has taken the snapshot,
it was not reflected. Due to this when QE reader called
TransactionIdIsCurrentTransactionId, it may get sometimes false based on timings
for subtransaction ids used by QE writer to insert/update tuples. Hence to fix
the situation, SharedLocalSnapshot is now updated when assigning transaction id
and deregistered if subtransaction aborts.

Also, adding faultinjector to suspend cursor QE reader instead of guc/sleep used
in past. Moving cursor tests from bugbuster to ICG and adding deterministic test
to exercise the behavior.

Fixes #1276, reported by @pengzhout

Added a guc, `optimizer_penalize_broadcast_threshold` to set the threshold on
maximum number of rows to broadcast.
Signed-off-by: NOmer Arap <oarap@pivotal.io>

Refactor the phase 2 retry logic of distributed transaction so that the retry happens
immediately after failure instead of happening inside EndCommand(). The patch also
increases the number of retries in case of failure to 2 and introduces a guc called
dtx_phase2_retry_count to control the number of retries.

This guc allow us to change the cost of sort and tuning the query plan
generated by Orca for better performance.

Fix wording for gp_email_to GUC and clarify the expected separator
for the list (as it has changed).  Also use email rather than e-mail
to be consistent with the online Greenplum documentation.

Segments that are still in an old format are treated as read-only. All new
data go to new segments, in new format. This allows us to eventually get
rid of the old format completely.

This is hypothetical until we have pg_upgrade working for GPDB 4.3 -> 5.0
upgrade, as you can't have old-format tables or segments at all in a
cluster that's initialized with 5.0. Stay tuned for pg_upgrade, but this
is preparatory work for that.

* Adding support for interrupt processing before reserving more vmem.

* Process pending interrupts before reserving VMEM.

* Adding guc to control vmem tracker checking for interrupts before reserving more vmem.

Persistent tables historically implemented different mechanism
to maintain free tuples by using on-disk freelist chain of itself.
It proved extremely hard to maintain this free-list and required
lot of supporting code to maintain and validate its integrity.
Hence this commit leverages the native heap delete and vacuum
framework to manage tuple deletion for persistent tables as well.

Introduce CodegenConfig to expose api for enabling/disabling generator
using guc. Gucs for exec_variable_list, expression eval, advance aggregate and slot_getattr
are added.
Signed-off-by: NFoyzur Rahman <foyzur@gmail.com>

Update comments for codegen enroll

Using INT_MAX is problematic in the case of gp_fts_probe_timeout
since calculations on it can overflow at INT_MAX. Using INT_MAX
gp_fts_probe_interval won't overflow with current usage but running
the FTS probe every INT_MAX seconds is nonsensical.

As part of creating a new relfile node, for example during reindex, the
persistent TID and the serial number in the relation descriptor used to be
updated with new values immediately.

This commit prevents such updates leaving the description unchanged.  Instead,
the relation descriptor will be udpated upon next cache invalidation.

The new ICG test is added to its own file and not in gp_create_table.sql
because reindex of pg_class index seems to affect planner, causing other tests
to fail with "insufficient memory reserved for statement" error.

We have observed that the persistent TID and serial number associated with a
relfilenode in a relation descriptor may be updated to a new value before
relfilenode is updated.  If an XLog record is written with such a relation
descriptor, then the XLog record fails to apply during recovery.

This commit adds a check to validate sanity of persistent information in a
relation descriptor by fetching persistent information using relfilenode from
gp_relation_node.

The validation is controlled by a new GUC, gp_validate_pt_info_relcache.

use asynchronous way to create gang and dispatch command by default.

Currently, when the stats are not existing for a relation, planner
relies on default values instead of running pg_relation_size to save
time. However, this causes planner to produce bad plans. Earlier, the
code had the feature to collect stats for planner if the table did not
had any stats, but it was removed.

This fix re-introduces the feature with the GUC to control whether to use
default values or run pg_relation_size to get actual stats if stats are
not present. By default, the GUC is turned off.

[#129570829]

This failure was introduced when we set the new GUC
(codegen_optimization_level) during init.
Fixes issue from this commit :
f160fa5a.

Used when compiling generated code. EXPLAIN codegen also
runs optimize with this optimize level, making it easier to see
the features the compiler optimizes.

Looking at old git history, this was added back in 2009. The related
ticket on adding it said:

    Add GUC that make every buffer page from the buffer pool flushed on
    eviction.

    Note that this will NOT necessarily flush all buffer pages when the
    postmaster is shutdown. I think this is acceptable for our purposes.

    (Our purpose is to make sure that overwrites of the buffer pages are not
    lost and instead are always written to disk so we can catch errors)

I'm not sure what errors that was meant to catch, or how, but I don't
think we have any regression tests or anything that uses it anymore.
Let's remove it, to make merging with upstream easier.

The previous commits have removed all usage of CaQL. It's no longer
needed.

Currenty, there is no easy way to evict entry from appendonly writer hash
table, most of current tests reduce the value of max_appendonly_tables which
needs database restart to take effect and then need to actually create that
many concurrent AO table inserts to evict the hash table entry. Using this
GUC can easily tests the eviction behavior.

This fixes the naming in c85f858e

A new feature of ORCA is to more efficiently handle array constraints.
It includes a new preprocessing stage, and a new way of internally
representing array constraints. This feature can be enabled by use of
this GUC.

A new feature of ORCA is to more efficiently handle array constraints.
It includes a new preprocessing stage, and a new way of internally
representing array constraints. This feature can be enabled by use of
this GUC.

This GUC is disabled by default.

Currently, when users run an `UNION ALL` query, such as
```sql
SELECT a FROM foo UNION ALL SELECT b FROM bar
```,
GPDB will parallelize the execution across all the segments, but the
table scans on `foo` and `bar` are not executed parallel. With this GUC
enabled, we add redistribution motion node under APPEND(UNION) operator,
which makes all the children of APPEND(UNION) operator execute in
parallel in every segment.

This closes #836
Signed-off-by: NMarc Spehlmann <cramja@gmail.com>

reorder header file

Currently gp_connections_per_thread=0 means using main thread to
do dispatch instead of creating new threads. It's not working if
segments report error before sending back any data.
For example:
create table foo(id int);
insert into foo values(0);
set gp_connections_per_thread=0;
select 100/id from foo;

We disable it for now and will reenable it when we have the new
implementation of non-threaded dispatcher.