This is a backport from master 2a7b2bf6

The dbms_pipe_session_{A,B} tests are flaky in CI as it can so happen
that session B calls receiveFrom() before session A can even call
createImplicitPipe(). This leads to flaky test failures such as:

--- /tmp/build/e18b2f02/gpdb_src/gpcontrib/orafce/expected/dbms_pipe_session_B.out	2020-04-20 17:02:27.270832458 +0000
+++ /tmp/build/e18b2f02/gpdb_src/gpcontrib/orafce/results/dbms_pipe_session_B.out	2020-04-20 17:02:27.278832994 +0000
@@ -7,14 +7,6 @@

 -- Receives messages sent via an implicit pipe
 SELECT receiveFrom('named_pipe');
-NOTICE:  RECEIVE 11: Message From Session A
-NOTICE:  RECEIVE 12: 01-01-2013
-NOTICE:  RECEIVE 13: Tue Jan 01 09:00:00 2013 PST
-NOTICE:  RECEIVE 23: \201
-NOTICE:  RECEIVE 24: (2,rob)
-NOTICE:  RECEIVE 9: 12345
-NOTICE:  RECEIVE 9: 12345.6789
-NOTICE:  RECEIVE 9: 99999999999
  receivefrom
 -------------

@@ -152,12 +144,13 @@
 ORDER BY 1;
       name      | items | limit | private |      owner
 ----------------+-------+-------+---------+-----------------
+ named_pipe     |     9 |    10 | f       |
  pipe_name_3    |     1 |       | f       |
  private_pipe_1 |     0 |    10 | t       | pipe_test_owner
  private_pipe_2 |     9 |    10 | t       | pipe_test_owner
  public_pipe_3  |     0 |    10 | f       |
  public_pipe_4  |     0 |    10 | f       |
-(5 rows)
+(6 rows)

This commit introduces an explicit sleep at the start of session B to
give session A a better chance to run.
Co-authored-by: NJesse Zhang <jzhang@pivotal.io>
(cherry picked from commit 985c5e2b)

Commit 13a1f66f forgot to modify the corresponding LockTagTypeNames array. 
This commit fixes this.

Previously, gpinitsystem did not allow the user to specify a hostname
and address for each segment in the input file used with -I; it only
accepted one value per segment and used it for both hostname and
address.

This commit changes the behavior so that the user can specify both
hostname and address. If the user specifies only the address (such as by
using an old config file), it will preserve the old behavior and set
both hostname and address to that value. It also adds a few tests around
input file parsing so SET_VAR is more resilient to further refactors.

The specific changes involved are the following:

1) Change SET_VAR to be able to parse either the old format (address
only) or new format (host and address) of the segment array
representation.
2) Move SET_VAR from gpinitsystem to gp_bash_functions.sh and remove the
redundant copy in gpcreateseg.sh.
3) Remove a hardcoded "~0" in QD_PRIMARY_ARRAY in gpinitsystem,
representing a replication port value, that was left over from 5X.
4) Improve the check for the number of fields in the segment array
representation.

Also, Remove use of ignore warning flag and use [[ ]] the check for IGNORE_WARNING

Previously, gpintsystem was incorrectly filling the hostname field of each
segment in gp_segment_configuration with the segment's address. This commit
changes it to correctly resolve hostnames and update the catalog accordingly.

This reverts commit 12ef7352, Revert "gpinitsystem: update catalog with
correct hostname".
   Commit message from 12ef7352:
   The commit requires some additional tweaks to the input file logic for
   backwards compatibility purposes, so we're reverting this until the full
   fix is ready.

This commit enables parallel writes for Foreign Data Wrapper. This
feature is currently missing from the FDW framework, whilst parallel
scans are supported, parallel writes are missing. FDW parallel writes
are analogous to writing to writable external tables that run on all
segments.

One caveat is that in the external table framework, writable tables
support a distribution policy:

    CREATE WRITABLE EXTERNAL TABLE foo (id int)
    LOCATION ('....')
    FORMAT 'CSV'
    DISTRIBUTED BY (id);

In foreign tables, the distribution policy cannot be defined during the
table definition, so we assume random distribution for all foreign
tables.

Parallel writes are enabled when the foreign table's exec_location is
set to FTEXECLOCATION_ALL_SEGMENTS only. For foreign tables that run on
master or any segment, the current policy behavior remains.

* docs - clarify text for resource group CPU core usage in catalog tables.

* docs - minor edits

* docs - update for gpstart when standby master is not available

* docs - minor edits.

* Update gpstart.xml
Co-authored-by: NDavid Yozie <dyozie@pivotal.io>

* Update docs around ssl_ciphers default, default behavior, TLS 1.2 recommendations

* Update cipher string per Stanley's feedback

Michael Paquier

Original Postgres commit:
https://github.com/postgres/postgres/commit/3153b1a52f8f2d1efe67306257aec15aaaf9e94c

Previously, some functions returned various fixed strings and others
failed with a cache lookup error.  Per discussion, standardize on
returning NULL.  Although user-exposed "cache lookup failed" error
messages might normally qualify for bug-fix treatment, no back-patch;
the risk of breaking user code which is accustomed to the current
behavior seems too high.

Michael Paquier

Original Postgres commit:
https://github.com/postgres/postgres/commit/976b24fb477464907737d28cdf18e202fa3b1a5b

According to plannode.h "plan_node_id" should be unique across
entire final plan tree. But ORCA DXL to PlanStatement translator
returns uninitialized zero values for BitmapOr and BitmapAnd nodes.
This behaviour differs from Postgres planner and from all other
node translations in this class. It was fixed.

(cherry picked from commit 53a0b781)

The gpinitsystem will fail if the gpinitsystem logs contain errors or
warnings from previous tests.

ALTER DATABASE SET ... FROM CURRENT dispatches incorrect statement to the
segments. Reported in https://github.com/greenplum-db/gpdb/issues/9823

Implements an algorithm in MakeHistArrayCmpAnyFilter() using CStatsPredArrayCmp:
1. Construct a histogram with the same bucket boundaries as present in the
   base_histogram.
   This is better than using a singleton bucket per point, because it that
   case, the frequency of each bucket is so small, it is often less than
   CStatistics::Epsilon, and may be considered as 0, leading to
   cardinality misestimation. Using the same buckets as base_histogram
   also aids in joining histogram later.
2. Compute the frequency for each bucket based on the number of points (NDV)
   present within each bucket boundary. NB: the points must be de-duplicated
   beforehand to prevent double counting.
3. Join this "dummy_histogram" with the base_histogram to determine the buckets
   from base_histogram that should be selected (using MakeJoinHistogram)
4. Compute and adjust the resultant scale factor for the filter.
Co-authored-by: NAshuka Xue <axue@pivotal.io>
Co-authored-by: NShreedhar Hardikar <shardikar@pivotal.io>

Functions renamed:
- CHistogram::Buckets -> GetNumBuckets
- CHistogram::ParseDXLToBucketsArray -> GetBuckets

Implemented DbgPrint for:
- CBucket
- CHistogram
Co-authored-by: NAshuka Xue <axue@pivotal.io>
Co-authored-by: NShreedhar Hardikar <shardikar@pivotal.io>

The reason is that new created reader gang would fail on QE due to missing
writer gang process in locking code, and retry would fail again with the same
reason, since the cached writer gang is still used because QD does not know &
check the real libpq network status. See below for the repro case.

Fixing this by checking the error message and then reset all gangs if seeing
the error message, similar to the code logic that checks the startup/recovery
message in gang create function. We could have other fixes, e.g. checking the
writer gang network status, etc but those fixes seem to be ugly after trying.

create table t1(f1 int, f2 text);
<kill -9 one idle QE>

insert into t1 values(2),(1),(5);
ERROR:  failed to acquire resources on one or more segments
DETAIL:  FATAL:  reader could not find writer proc entry, lock [0,1260] AccessShareLock 0 (lock.c:874)
 (seg0 192.168.235.128:7002)

insert into t1 values(2),(1),(5);
 ERROR:  failed to acquire resources on one or more segments
 DETAIL:  FATAL:  reader could not find writer proc entry, lock [0,1260] AccessShareLock 0 (lock.c:874)
  (seg0 192.168.235.128:7002)

<-- Above query fails again.

Cherry-picked from 24f16417 and a0a5b4d5

commit d453a4aa implemented that for the crash
recovery case (not marking the node down and then not promoting the mirror). It
seems that we should do that for the usual "starting up" case also(i.e.
CAC_STARTUP), besides for the existing "in recovery mode" case (i.e.
CAC_RECOVERY).

We've seen that fts promotes the "starting up" primary during isolation2
testing due to 'pg_ctl restart'. In this patch we check recovery progress for
both CAC_STARTUP an CAC_RECOVERY during fts probe and thus can avoid this.
Reviewed-by: NAshwin Agrawal <aagrawal@pivotal.io>

cherry-picked from d71b3afd

On master the commit message was eliminated by mistake. Added back on gpdb6.

In TCP interconnect, the sender used to force an EOS messages to the
receiver in two cases:
1. cancelUnfinished is true in mppExecutorFinishup.
2. an error occurs.

For case1, the comment says: to finish a cursor, the QD used to send
a cancel to the QEs, QEs then set the cancelUnfinished flag and did
a normal executor finish up. We now use QueryFinishPending mechanism
to stop a cursor, so case1 logic is invalid for a long time.

For case2, the purpose is: when an error occurs, we force an EOS to
the receiver so the receiver didn't report an interconnect error and
QD then will check the dispatch results and report the errors in the
QEs. From the view of interconnect, we have selectedd to the end of
the query and no error in the interconnect, this logic has two
problems:
1. it doesn't work for initplan, initplan will not check the dispatch
results and throw the errors, so when an error occurs in the QEs for
the initplan, the QD cannot notice that.
2. it doesn't work for cursors, for example:
   DECLARE c1 cursor for select i from t1 where i / 0 = 1;
   FETCH all from c1;
   FETCH all from c1;
All FETCH commands don't report errors which is not expected.

This commit removed the forceEos mechanism, for the case2, the
receiver will report an interconnect error without forceEos, this is
ok because when multiple errors reports from QEs, the QD is inclined
to report non-interconnect error.

This flag is duplicated with 'forceEOS', 'forceEOS' can also tell
whether errors occur or not.

flushBuffer() is used to send packets through TCP interconnect, before
sending, it first check whether receiver stopped or teared down the
interconnect, however, there is window between checking and sending, the
receiver may tear down the interconnect and close the peer, so send()
will report an error, to resolve this, we recheck whether the receiver
stopped or teared down the interconnect in this window and don't error
out in that case.
Reviewed-by: NJinbao Chen <jinchen@pivotal.io>
Reviewed-by: NHao Wu <hawu@pivotal.io>

We hit interconnect hung issue many times in many cases, all have
the same pattern: the downstream interconnect motion senders keep
sending the tuples and they are blind to the fact that upstream
nodes have finished and quitted the execution earlier, the QD
then get enough tuples and wait all QEs to quit which cause a
deadlock.

Many nodes may quit execution earlier, eg, LIMIT, HashJoin, Nest
Loop, to resolve the hung issue, they need to stop the interconnect
stream explicitly by calling ExecSquelchNode(), however, we cannot
do that for rescan cases in which data might lose, eg, commit
2c011ce4. For rescan cases, we tried using QueryFinishPending to
stop the senders in commit 02213a73 and let senders check this
flag and quit, that commit has its own problem, firstly, QueryFini
shPending can only set by QD, it doesn't work for INSERT or UPDATE
cases, secondly, that commit only let the senders detect the flag
and quit the loop in a rude way (without sending the EOS to its
receiver), the receiver may still be stuck inreceiving tuples.

This commit revert the QueryFinishPending method firstly.

To resolve the hung issue, we move TeardownInterconnect to the
ahead of cdbdisp_checkDispatchResult so it guarantees to stop
the interconnect stream before waiting and checking the status
of QEs.

For UDPIFC, TeardownInterconnect() remove the ic entries, any
packets for this interconnect context will be treated as 'past'
packets and be acked with STOP flag.

For TCP, TeardownInterconnect() close all connection with its
children, the children will treat any readable data in the
connection as a STOP message include the closure operation.

A test case is not included, both commit 2c011ce4 and 02213a73
contain one.

QD backend should not forget whether a sub transaction performed writes

QD backend process can avoid two-phase commit overhead if it knows that no QEs
involved in this transaction or any of its sub transactions performed any
writes. Previously, if a sub transaction performed write on one or more QEs, it
was remembered in that sub transaction's global state. However, the sub
transaction state was lost after sub transaction commit. That resulted in QD
not performing two-phase commit at the end of top transaction.

In fact, regardless of the transaction nesting level, we only need to remember
whether a write was performed by a sub transaction. Therefore, use a backend
global variable, instead of current transaction state to record this
information.
Reviewed-by: NGang Xiong <gxiong@pivotal.io>
Reviewed-by: NHao Wu <gfphoenix78@gmail.com>
Reviewed-by: NAsim R P <apraveen@pivotal.io>

Currently this UDF might report a false positive if the node is still starting
up after timeout since currently pg_ctl returns 0 for this case. This behavior
is changed in upstream with the below patch:

commit f13ea95f
Author: Tom Lane <tgl@sss.pgh.pa.us>
Date:   Wed Jun 28 17:31:24 2017 -0400

    Change pg_ctl to detect server-ready by watching status in postmaster.pid.

We've seen some test flakiness due to this issue since pg_ctl restart needs
more time sometimes on pipeline (by default pg_ctl timeout is 60 seconds).
Yesterday I found on a hang job that a primary needs ~ 4 minutes to get the
recovery finished during 'pg_ctl restart' (It's test
ao_same_trans_truncate_crash which enables fsync. Even it launches a checkpoint
before pg_ctl restart, pg_ctl restarts still needs a lot of time).

Enlarge the timeout of pg_ctl to 600 seconds now and add a pg_ctl stdout
checking before returning OK in the UDF (this check could be removed after PG
12 merge finishes so I added a FIXME there).

Here is the output of the pg_ctl experiment:

$ pg_ctl -l postmaster.log -D /data/gpdb7/gpAux/gpdemo/datadirs/dbfast1/demoDataDir0 -w -m immediate restart -t 1
waiting for server to shut down.... done
server stopped
waiting for server to start.... stopped waiting
server is still starting up
$ echo $?
0
Reviewed-by: NAsim R P <apraveen@pivotal.io>

Cherry-picked from 934d87c6

* docs - updates for PostGIS 2.5.4

-Updated version 2.1.5 -> 2.5.4
-Updated references to PostGIS topics
-Removed geometry and geography limitation
-Added enhancement for postgis_manager.sh script - support installing PostGIS in a non-default schema
-Added examples of updating postgis.enable_outdb_rasters and postgis.gdal_enabled_drivers for a GPDB session
-Remove ANALYZE limitation for user-defined data types
-Added  _postgis_index_extent function is not supported
-Added note: <-> operator returns centroid/centroid distance
-changed name of PostGIS aggregate function ST_MemCollect to ST_Collect

* docs - fix typo

* docs - minor changes to topic hierarchy.

* docs - one more minor hierarchy change.

* docs - update external URLs

* docs - fix typo, minor cleanup

* docs - found strange character

* docs - add Note about host requirements to segment mirror overview

-also edited some related topics and changed some links
-fixed a bad link to the gpcheck utiltiy (removed in GPDB 6)

* docs - minor edit
Co-authored-by: NDavid Yozie <dyozie@pivotal.io>

* docs - add information about view dependency

-example queries to find dependency information
-catalog table information
-best practices

* docs - minor edits and fixes.

* docs - fix minor typo.

* docs - another minor edit.

Similar to 3bdece19 and 51257848, an Append node doesn't always have
a Flow attached to it. Teach EXPLAIN code to dig into its children.

This seems to be a recurring issue, but since it's been refactored away in
'master', I guess we'll just whack these on 6X_STABLE as they're reported.
Downgrade it to a WARNING, though; an assertion failure seems too harsh
for this. Now you get the warning even on non-assertion builds, but I
think that's fine.

Fixes https://github.com/greenplum-db/gpdb/issues/9819Reviewed-by: NZhenghua Lyu <zlv@pivotal.io>

At several places we just invoke the function add_size
but does not assign its return value to any variable.
The result is just discarded. This commit fixes this.

* clarifying pg_upgrade note

* adding VACUUM note for gpexpand

* gpexpand edits

* small edit

* edits after review

- add landing page for upgrade (6->6 upgrade, 4/5->6 migration)
- nest in subnav

Half committed transactions in shmCommittedGxactArray are omitted.
The bug could cause data loss/inconsistency. If transaction T1
failed to commit prepared for some reasons, and the transaction T1
has been committed on the master and other segments, but the transaction
T1 isn't appended in the checkpoint record. So the DTX recovery
can't retrieve the transaction and run recovery-commit-prepared,
and the prepared transactions on the segment are aborted.
Co-authored-by: NGang Xiong <gxiong@pivotal.io>

Allocate memory of CdbPgResults.pg_results with palloc0() instead
of calloc(), and free the memory afer use.

The CdbPgResults.pg_results array that is returned from various dispatch
functions is allocated by cdbdisp_returnResults() via calloc(), but in
most cases the memory is not free()-ed after use.

To avoid memory leak, the array is now allocated with palloc0() and
recycled with pfree().

Track which row and which result set is being processed in function context in pg_lock_status(), 
so that an ineffient inner loop can be eliminated.

Author: Fang Zheng <zhengfang.xjtu@gmail.com>

Consider the scenario below

```
create table tenk1 (c1 int, ten int);
create temp sequence ts1;
explain select * from (select distinct ten from tenk1) ss where ten < 10 + nextval('ts1') order by 1;
```

The filter outside the subquery is a candidate to be pushed below the
'distinct' in the sub-query.  But since 'nextval' is a volatile function, we
should not push it.

Volatile functions give different results with each execution. We don't want
aggs to use result of a volatile function before it is necessary. We do it for
all aggs - DISTINCT and GROUP BY.

Also see commit 6327f25d.

1. The CurrentMemoryContext in CreateCheckPoint is long-live
memory context owned by the checkpointer process. The memory context
is reset only if the error occurs. So memory leak will lead to
huge memory leak in the OS.
2. The prepared transactions are not appended as an extension in
the checkpoint wal record, which introduces a bug. It occurs:
  1) T1 does some DML and is prepared.
  2) T2 runs a checkpoint.
  3) The seg0 crashes before the transaction T1 successfully runs `COMMIT PREPARE`, but
     all other segments have successfully committed.
  4) When running local recovery, seg0 doesn't know T1 is prepared
     and needs to be committed again.
This is different from the master branch. Master uses files to record
whether there exists any prepared transaction and what them are.
Reviewed-by: NNing Yu <nyu@pivotal.io>
Reviewed-by: NGang Xiong <gxiong@pivotal.io>

The shared oldestXmin (DistributedLogShared->oldestXmin) may be updated
concurrently. It should be set to a higher value, because a higher xmin
can belong to another distributed log segment, its older segments might
already be truncated.

For Example: txA and txB call DistributedLog_AdvanceOldestXmin concurrently.

```
txA and txB: both hold shared DistributedLogTruncateLock.

txA: set the DistributedLogShared->oldestXmin to XminA. TransactionIdToSegment(XminA) = 0009

txB: set the DistributedLogShared->oldestXmin to XminB. TransactionIdToSegment(XminB) = 0008

txA: truncate segment 0008, 0007...
```

After that, DistributedLogShared->oldestXmin == XminB, it is on removed
segment 0008. Subsequent GetSnapshotData() calls will be failed because SimpleLruReadPage will error out.
Co-authored-by: Ndh-cloud <60729713+dh-cloud@users.noreply.github.com>