Commit b3f300b9 introduced the novel idea tracking oldest xmin
among all distributed snapshots on QEs.  However, the idea is not
applicable to QD because all distributed transactions can be found in
ProcArray on QD.  Local oldest xmin is therefore the oldest xmin among
all distributed snapshots on QD.  This patch fixes the maintenance of
oldest xmin on QD by avoiding DistributedLog_AdvanceOldestXmin() and all
the heavy-lifting that it performs.  Calling this on QD was also hitting
the "local snapshot's xmin is older than recorded distributed
oldestxmin" error occasionally in CI.

Before this, in order to safely determine if a tuple can be vacuumed away,
you would need an active distributed snapshot. Even if an XID was older
than the locally-computed OldestXMin value, the XID might still be visible
to some distributed snapshot that's active in the QD.

This commit introduces a mechanism to track the "oldest xmin" across any
distributed snapshots. That makes it possible to calculate an "oldest xmin"
value in a QE, that covers any such distributed snapshots, even if the
distributed transaction doesn't currently have an active connection to this
QE. Every distributed snapshot contains such an "oldest xmin" value, but
now we track the latest such value that we've seen in this QE, in shared
memory. Therefore, it's not always 100% up-to-date, but it will reflect
the situation as of the latest query that was dispatched from QD to this
QE.

The value returned by GetOldestXmin(), as well as RecentGlobalXmin, now
includes any distributed transactions. So the value can now be used to
determine which tuples are dead, like in upstream, without doing the extra
check with the localXidSatisfiesAnyDistributedSnapshot() function. This
allows reverting some changes in heap_tuple_freeze.

This allows utility-mode VACUUMs, launched independently in QE nodes, to
reclaim space. Previously, it could not remove any dead tuples that were
ever visible to anyone, because it could not determine whether they might
still be needed by some distributed transaction.

All in-progress transactions, even those in DTX_STATE_ACTIVE_NOT_DISTRIBUTED
state, must be included in a distributed snapshot. All transactions begin
in DTX_STATE_ACTIVE_NOT_DISTRIBUTED state, and can become distributed later
on.

This bug was introduced in commit ff97c70b, which added the ill-advised
optimization to skip DTX_STATE_ACTIVE_NOT_DISTRIBUTED transactions.

This showed up occasionally in the regression tests as a failure in the
'oidjoins' test, as a failure like this:

@@ -230,9 +230,17 @@
 SELECT	ctid, attrelid
 FROM	pg_catalog.pg_attribute fk
 WHERE    attrelid != 0 AND
     NOT EXISTS(SELECT 1 FROM pg_catalog.pg_class pk WHERE pk.oid = fk.attrelid);
- ctid | attrelid
-------+----------
-(0 rows)
+  ctid   | attrelid
+---------+----------
+ (20,10) |    17107
+ (20,11) |    17107
+ (20,12) |    17107
+ (20,13) |    17107
+ (20,14) |    17107
+ (20,15) |    17107
+ (20,8)  |    17107
+ (20,9)  |    17107
+(8 rows)

The plan for that is a hash anti-join, with 'pg_class' in the inner side,
and 'pg_attribute' on the outer side. If a table was created concurrently
with that query (with oid 17107 in the above case), you could get the
failure. What happens is that the concurrent CREATE TABLE transaction was
assigned a distributed XID that was incorrectly not included in the
distributed snapshot that the query took. Hence, the transaction became
visible to the query immediately, as soon as it committed. If the CREATE
TABLE transaction committed between the full scan of pg_class, and the
scan on pg_attribute, the query would not see the just-inserted pg_class
row, but would see the pg_attribute rows.

Entry DB process share snapshot with QD process, but it didn't update
the TransactionXmin. The assert in SubTransGetData() will fail in some
cases:
   Assert(TransactionIdFollowsOrEquals(xid, TransactionXmin));

1.QD takes a snapshot which contains a in-progress transaction A.
2.Transaction A commits.
3.QD creates a gang of entry DB. The entry DB process takes a snapshot
in InitPostgres and updates TransactionXmin.
4.The entry DB process scans the tuple inserted by the transaction A,
and find it's in the snapshot but its xid is larger than TransactionXmin.

1. move TMGXACT to PGPROC.
2. create distributed snapshot and create checkpoint will
traverse procArray and acquire ProcArrayLock. shmControlLock
is only used for serialize recoverTM().
3. get rid of shmGxactArray and maintain an array of TMGXACT_LOG
for recovery.

Author: Gang Xiong <gxiong@pivotal.io>
Author: Asim R P <apraveen@pivotal.io>
Author: Ashwin Agrawal <aagrawal@pivotal.io>

If the caller specifies DF_WITH_SNAPSHOT, so that the command is dispatched
to the segments with a snapshot, but it currently has no active snapshot in
the QD itself, that seems like a mistake.

In qdSerializeDtxContextInfo(), the comment talked about which snapshot to
use when the transaction has already been aborted. I didn't quite
understand that. I don't think the function is used to dispatch the "ABORT"
statement itself, and we shouldn't be dispatching anything else in an
already-aborted transaction.

This makes it more clear which snapshot is dispatched along with the
command. In theory, the latest or serializable snapshot can be different
from the one being used when the command is dispatched, although I'm not
sure if there are any such cases in practice.

In the upcoming 8.4 merge, there are more changes coming up to snapshot
management, which make it more difficult to get hold of the latest acquired
snapshot in the transaction, so changing this now will ease the pain of
merging that.

I don't know why, but after making the change in qdSerializeDtxContextInfo,
I started to get a lot of "Too many distributed transactions for snapshot
(maxCount %d, count %d)" errors. Looking at the code, I don't understand
how it ever worked. I don't see any no guarantee that the array in
TempQDDtxContextInfo or TempDtxContextInfo was pre-allocated correctly.
Or maybe it got allocated big enough to hold max_prepared_xacts, which
was always large enough, but it seemed rather haphazard to me. So in
the spirit of "if you don't understand it, rewrite it until you do", I
changed the way the allocation of the inProgressXidArray array works.
In statically allocated snapshots, i.e. SerializableSnapshot and
LatestSnapshot, the array is malloc'd. In a snapshot copied with
CopySnapshot(), it is points to a part of the palloc'd space for the
snapshot. Nothing new so far, but I changed CopySnapshot() to set
"maxCount" to -1 to indicate that it's not malloc'd. Then I modified
DistributedSnapshot_Copy and DistributedSnapshot_Deserialize to not give up
if the target array is not large enough, but enlarge it as needed. Finally,
I made a little optimization in GetSnapshotData() when running in a QE, to
move the copying of the distributed snapshot data to outside the section
guarded by ProcArrayLock. ProcArrayLock can be heavily contended, so that's
a nice little optimization anyway, but especially now that
DistributedSnapshot_Copy() might need to realloc the array.

ereport() has one subtle but important difference to elog: it doesn't
evaluate its arguments, if the log level says that the message doesn't
need to be printed. This makes a small but measurable difference in
performance, if the arguments contain more complicated expressions, like
function calls.

While performance testing a workload with very short queries, I saw some
CPU time being used in DtxContextToString. Those calls were coming from the
arguments to elog() statements, and the result was always thrown away,
because the log level was not high enough to actually log anything. Turn
those elog()s into ereport()s, for speed.

The problematic case here was a few elogs containing DtxContextToString
calls, in hot codepaths, but I changed a few surrounding ones too, for
consistency.

Simplify the mock test, to not bother mocking elog(), while we're at it.
The real elog/ereport work just fine in the mock environment.

This commit restore TCP interconnect and fix some hang issues.

* restore TCP interconnect code
* Add GUC called gp_interconnect_tcp_listener_backlog for tcp to control the backlog param of listen call
* use memmove instead of memcpy because the memory areas do overlap.
* call checkForCancelFromQD() for TCP interconnect if there are no data for a while, this can avoid QD from getting stuck.
* revert cancelUnfinished related modification in 8d251945, otherwise some queries will get stuck
* move and rename faultinjector "cursor_qe_reader_after_snapshot" to make test cases pass under TCP interconnect.

Originally, the reader kept copies of subtransaction information in
two places.  First, it copied SharedLocalSnapshotSlot to share between
writer and reader.  Second, reader kept another copy in subxbuf for
better performance.  Due to lazy xid, subtransaction information can
change in the writer asynchronously with respect to the reader.  This
caused reader's subtransaction information out of date.

This fix removes those copies of subtransaction information in the
reader and adds a reference to the writer's PGPROC to
SharedLocalSnapshotSlot.  Reader should refer to subtransaction
information through writer's PGPROC and pg_subtrans.

Also added is a lwlock per shared snapshot slot.  The lock protects
shared snapshot information between a writer and readers belonging to
the same session.

Fixes github issues #2269 and #2284.
Signed-off-by: NAsim R P <apraveen@pivotal.io>

Before this commit, snapshot stored information of distributed in-progress
transactions populated during snapshot creation and its corresponding localXids
found during tuple visibility check later (used as cache) by reverse mapping
using single tightly coupled data structure DistributedSnapshotMapEntry. Storing
the information this way possed couple of problems:

1] Only one localXid can be cached for a distributedXid. For sub-transactions
same distribXid can be associated with multiple localXid, but since can cache
only one, for other local xids associated with distributedXid need to consult
the distributed_log.

2] While performing tuple visibility check, code must loop over full size of
distributed in-progress array always first to check if cached localXid can be
utilized to avoid reverse mapping.

Now, decoupled the distributed in-progress with local xids cache separately. So,
this allows us to store multiple xids per distributedXid. Also, allows to
optimize scanning localXid only if tuple xid is relevant to it and also scanning
size only equivalent to number of elements cached instead of size of distributed
in-progress always even if nothing was cached.

Along the way, refactored relevant code a bit as well to simplify further.

For vacuum, page pruning and freezing to perform its job correctly on QE's, it
needs to know globally what's the lowest dxid till any transaction can see in
full cluster. Hence QD must calculate and send that info to QE. For this purpose
using logic similar to one for calculating globalxmin by local snapshot. TMGXACT
for global transactions serves similar to PROC and hence its leveraged to
provide us lowest gxid for its snapshot. Further using its array, shmGxactArray,
can easily find the lowest across all global snapshots and pass down to QE via
snapshot.

Adding unit test for createDtxSnapshot along with the change.

Coverity reported: Either the check against null is unnecessary, or there may be
a null pointer dereference. In ProcArrayEndTransaction: Pointer is checked
against null but then dereferenced anyway.

While its not an issue and for commit case, pointer is never null, but simplify
the code and stop using pointer itself here.

Commit fb86c90d "Simplify management of
distributed transactions." cleanedup lot of code for LocalDistribXactData and
introduced LocalDistribXactData in PROC for debugging purpose. But it's only
correctly maintained for QE's, QD never populated LocalDistribXactData in
MyProc. Instead TMGXACT also had LocalDistribXactData which was just set
initially for QD but never updated later and confused more than serving the
purpose. Hence removing LocalDistribXactData from TMGXACT, as it already has
other fields which provide required information. Also, cleaned-up QD related
states as even in PROC only QE uses LocalDistribXactData.

As part of 8.3 merge, upstream commit 295e6398
"Implement lazy XID allocation" was merged. But transactionIds were still
allocated in StartTransaction as code changes required to make it work for GPDB
with distrbuted transaction was pending, thereby feature remained as
disabled. Some progress was made by commit
a54d84a3 "Avoid assigning an XID to
DTX_CONTEXT_QE_AUTO_COMMIT_IMPLICIT queries." Now this commit addresses the
pending work needed for handling deferred xid allocation correctly with
distributed transactions and fully enables the feature.

Important highlights of changes:

1] Modify xlog write and xlog replay record for DISTRIBUTED_COMMIT. Even if
transacion is read-only for master and no xid is allocated to it, it can still
be distributed transaction and hence needs to persist itself in such a case. So,
write xlog record even if no local xid is assigned but transaction is
prepared. Similarly during xlog replay of the XLOG_XACT_DISTRIBUTED_COMMIT type,
perform distributed commit recovery ignoring local commit. Which also means for
this case don't commit to distrbuted log, as its only used to perform reverse
map of localxid to distributed xid.

2] Remove localXID from gxact, as its no more needed to be maintained and used.

3] Refactor code for QE Reader StartTransaction. There used to be wait-loop with
sleep checking to see if SharedLocalSnapshotSlot has distributed XID same as
that of READER to assign reader some xid as that of writer, for SET type
commands till READER actually performs GetSnapShotData(). Since now a) writer is
not going to have valid xid till it performs some write, writers transactionId
turns out InvalidTransaction always here and b) read operations like SET doesn't
need xid, any more hence need for this wait is gone.

4] Thow error if using distributed transaction without distributed xid. Earlier
AssignTransactionId() was called for this case in StartTransaction() but such
scenario doesn't exist hence convert it to ERROR.

5] QD earlier during snapshot creation in createDtxSnapshot() was able to assign
localXid in inProgressEntryArray corresponding to distribXid, as localXid was
known by that time. That's no more the case and localXid mostly will get
assigned after snapshot is taken. Hence now even for QD similar to QE's snapshot
creation time localXid is not populated but later found in
DistributedSnapshotWithLocalMapping_CommittedTest(). There is chance to optimize
and try to match earlier behavior somewhat by populating gxact in
AssignTransactionId() once locakXid is known but currently seems not so much
worth it as QE's anyways have to perform the lookups.

Leverage the fact that inProgressEntryArray is sorted based on distribXid while
creating the snapshot in createDtxSnapshot. So, can break out fast in function
DistributedSnapshotWithLocalMapping_CommittedTest().

`MyProc->inCommit` is to protect checkpoint running during inCommit
transactions.

However, `MyProc->lxid` has to be valid because `GetVirtualXIDsDelayingChkpt()`
and `HaveVirtualXIDsDelayingChkpt()` require `VirtualTransactionIdIsValid()` in
addition to `inCommit` to block the checkpoint process.

In this fix, we defer clearing `inCommit` and `lxid` to `CommitTransaction()`.

Originally checkpoint is checking for xid, however, xid is used to control the
transaction visibility and it's crucial to clean this xid if process is done
with commit and before release locks.

However, checkpoint need to wait for the `AtExat_smgr()` to cleanup persistent
table information, which happened after release locks, where `xid` is already
cleaned.

Hence, we use VXID, which doesn't have visibility impact.

NOTE: Upstream PostgreSQL commit f21bb9cf for the
similar fix.

Minimizes merge conflicts in the future.

As part of 8.3 merge via this upstream commit
92c2ecc1, code to ignore lazy vacuum from
calculating RecentXmin and RecentGlobalXmin was introduced.

In GPDB as part of lazy vacuum, reindex is performed for bitmap indexes, which
generates tuples in pg_class with lazy vacuum's transaction ID. Ignoring lazy
vacuum from RecentXmin and RecentGlobalXmin during GetSnapshotData caused
incorrect setting of hintbits to `HEAP_XMAX_INVALID` for tuple intended to be
deletd by lazy vacuum and breaking HOT chain. This transaction visibility issue
was encountered in CI many times with parallel schedule `bitmap_index, analyze`
failing with error `could not find pg_class tuple for index` at commit time of
lazy vacuum. Hence this commit stops tracking lazy vacuum in MyProc and
performing any specific action related to same.

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

We used to have a separate array of LocalDistributedXactData instances, and
a reference in PGPROC to its associated LocalDistributedXact. That's
unnecessarily complicated: we can store the LocalDistributedXact information
directly in the PGPROC entry, and get rid fo the auxiliary array and the
bookkeeping needed to manage that array.

This doesn't affect the backend-private cache of committed Xids that also
lives in cdblocaldistribxact.c.

Now that the PGPROC->localDistributedXactData fields are never accessed
by other backends, don't protect it with ProcArrayLock anymore. This makes
the code simpler, and potentially improves performance too (ProcArrayLock
can be very heavily contended on a busy system).

Callers to FaultInjector_InjectFaultIfSet() which don't pass neither
databasename nor tablename and that use DDLNotSpecified can instead
use the convenient macro SIMPLE_FAULT_INJECTOR() which cuts down on
the boilerplate in the code. This commit does not bring any changes
in functionality, merely readability.

Not urgent to do right now, but makes merging and diffing easier.

As promised in previous commit. Upstream patch:

commit bd0a2609
Author: Tom Lane <tgl@sss.pgh.pa.us>
Date:   Fri Jun 1 19:38:07 2007 +0000

    Make CREATE/DROP/RENAME DATABASE wait a little bit to see if other backends
    will exit before failing because of conflicting DB usage.  Per discussion,
    this seems a good idea to help mask the fact that backend exit takes nonzero
    time.  Remove a couple of thereby-obsoleted sleeps in contrib and PL
    regression test sequences.

to have different values in different processes of the primary server.
Also put it into the "Streaming Replication" GUC category; it doesn't belong
in "Standby Servers" because you use it on the master not the standby.
In passing also correct guc.c's idea of wal_keep_segments' category.

without them, related to previous commit. Report by Bruce Momjian.

of requirements and documentation on LogStandbySnapshot(). Fixes
two minor bugs reported by Tom Lane that would lead to an incorrect
snapshot after transaction wraparound. Also fix two other problems
discovered that would give incorrect snapshots in certain cases.
ProcArrayApplyRecoveryInfo() substantially rewritten. Some minor
refactoring of xact_redo_apply() and ExpireTreeKnownAssignedTransactionIds().

confusion with streaming-replication settings.  Also, change its default
value to "off", because of concern about executing new and poorly-tested
code during ordinary non-replicating operation.  Per discussion.

In passing do some minor editing of related documentation.

and be more tense about the locking requirements for it, to improve performance
in Hot Standby mode.  In passing fix a few bugs and improve a number of
comments in the existing HS code.

Simon Riggs, with some editorialization by Tom

Clarify comments, downgrade a message to DEBUG and remove some
debug counters. Direct from ideas by Heikki Linnakangas.

to handling of btree delete records mean that all snapshot
conflicts on standby now have a valid, useful latestRemovedXid.
Our earlier approach using LW_EXCLUSIVE was useful when we didnt
always have a valid value, though is no longer useful or necessary.
Asserts added to code path to prove and ensure this is the case.
This will reduce contention and improve performance of larger Hot
Standby servers.

This prevents a rare, yet possible race condition at the exact moment
of transition from recovery to normal running.

through normal backends. Makes code clearer also, since we
avoid various Assert()s. Performance of snapshots taken
during recovery no longer depends upon number of read-only
backends.

assertion failure reported by Erik Rijkers, but this alone doesn't explain
the failure.

woken by alarm we send SIGUSR1 to all backends requesting that they
check to see if they are blocking Startup process. If so, they throw
ERROR/FATAL as for other conflict resolutions. Deadlock stop gap
removed. max_standby_delay = -1 option removed to prevent deadlock.