An upcoming patch that adds deduplication to the nbtree AM will rely on
_bt_keep_natts_fast() understanding that differences in TOAST input
state can never affect its answer. In particular, two opclass-equal
datums (with opclasses deemed safe for deduplication) should never be
treated as unequal by _bt_keep_natts_fast() due to TOAST input
differences.
This also seems like a good idea on general principle. nbtsplitloc.c
will now occasionally make better decisions about where to split a leaf
page. The behavior of _bt_keep_natts_fast() is now somewhat closer to
the behavior of _bt_keep_natts().
Discussion: https://postgr.es/m/CAH2-Wzn3Ee49Gmxb7V1VJ3-AC8fWn-Fr8pfWQebHe8rYRxt5OQ@mail.gmail.com
nbtree index builds once stashed the "minimum key" for a page, which was
used as the basis of the pivot tuple that gets placed in the next level
up (i.e. the tuple that stores the downlink to the page in question).
It doesn't quite work that way anymore, so the "minimum key" terminology
now seems misleading (these days the minimum key is actually a straight
copy of the high key from the left sibling, which is a distinct thing in
subtle but important ways). Rename this concept to "low key". This
name is a lot clearer given that there is now a sharp distinction
between pivot and non-pivot tuples. Also remove comments that describe
obsolete details about how the minimum key concept used to work.
Rather than generating the minus infinity item for the leftmost page on
a level by copying the new item and truncating that copy, simply
allocate a small buffer. The old approach confusingly created the
impression that the new item had some kind of significance. This was
another artifact of how things used to work before commits 8224de4f and
dd299df8.
When maintaining or merging patches, one of the most common sources
for conflicts are the list of objects in makefiles. Especially when
the split across lines has been changed on both sides, which is
somewhat common due to attempting to stay below 80 columns, those
conflicts are unnecessarily laborious to resolve.
By splitting, and alphabetically sorting, OBJS style lines into one
object per line, conflicts should be less frequent, and easier to
resolve when they still occur.
Author: Andres Freund
Discussion: https://postgr.es/m/20191029200901.vww4idgcxv74cwes@alap3.anarazel.de
Assert that _bt_binsrch() binary searches with scantid set in insertion
scankey cannot be performed on leaf pages. Leaf-level binary searches
where scantid is set must use _bt_binsrch_insert() instead.
_bt_binsrch_insert() is likely to have additional responsibilities in
the future, such as searching within GIN-style posting lists using
scantid. It seems like a good idea to tighten things up now.
Commit efada2b8e9, which made the nbtree page deletion algorithm more
robust, removed the concept of a half-dead internal page. Remove a
comment about half dead parent pages that was overlooked.
The Postgres approach to coupling locks during an ascent of the tree is
slightly different to the approach taken by Lehman and Yao. Add a new
paragraph to the "Differences to the Lehman & Yao algorithm" section of
the nbtree README that explains the similarities and differences.
Adjust the struct comment that describes how page splits use their
descent stack to cascade up the tree from the leaf level.
In passing, fix up some unrelated nbtree comments that had typos or were
obsolete.
The initial value of the nbtree stack downlink block number field
recorded during an initial descent of the tree wasn't actually used.
Both _bt_getstackbuf() callers overwrote the value with their own value.
Remove the block number field from the stack struct, and add a child
block number argument to _bt_getstackbuf() in its place. This makes the
overall design of _bt_getstackbuf() clearer.
Author: Peter Geoghegan
Reviewed-By: Anastasia Lubennikova
Discussion: https://postgr.es/m/CAH2-Wzmx+UbXt2YNOUCZ-a04VdXU=S=OHuAuD7Z8uQq-PXTYUg@mail.gmail.com
Commit d2086b08b0 removed almost all cases where nbtree must release a
read buffer lock and acquire a write buffer lock instead, so remaining
cases in which that's still necessary are not notable enough to appear
in the nbtree README.
More importantly, holding on to a buffer pin in cases where nbtree must
trade a read lock for a write lock is very unlikely to save any I/O.
This seems to have been a long overlooked throwback to a time when
nbtree cared about write-ordering dependencies, and performed
synchronous buffer writes. It hasn't worked that way in many years.
Use the PageIndexTupleOverwrite() bufpage.c routine within nbtree
instead of deleting a tuple and re-inserting its replacement. This
makes the intent of affected code slightly clearer. It also makes
CREATE INDEX slightly faster, since there is no longer a need to shift
every leaf page's line pointer array back and forth during index builds.
Author: Peter Geoghegan, Anastasia Lubennikova
Reviewed-By: Anastasia Lubennikova
Discussion: https://postgr.es/m/CAH2-Wz=Zk=B9+Vwm376WuO7YTjFc2SSskifQm4Nme3RRRPtOSQ@mail.gmail.com
Commit 857f9c36cd, which taught nbtree VACUUM to avoid unnecessary
index scans, bumped the nbtree version number from 2 to 3, while adding
the ability for nbtree indexes to be upgraded on-the-fly. Various
assertions that assumed that an nbtree index was always on version 2 had
to be changed to accept any supported version (version 2 or 3 on
Postgres 11).
However, a few assertions were missed in the initial commit, all of
which were in code paths that cache a local copy of the metapage
metadata, where the index had been expected to be on the current version
(no longer version 2) as a generic sanity check. Rather than simply
update the assertions, follow-up commit 0a64b45152 intentionally made
the metapage caching code update the per-backend cached metadata version
without changing the on-disk version at the same time. This could even
happen when the planner needed to determine the height of a B-Tree for
costing purposes. The assertions only fail on Postgres v12 when
upgrading from v10, because they were adjusted to use the authoritative
shared memory metapage by v12's commit dd299df8.
To fix, remove the cache-only upgrade mechanism entirely, and update the
assertions themselves to accept any supported version (go back to using
the cached version in v12). The fix is almost a full revert of commit
0a64b45152 on the v11 branch.
VACUUM only considers the authoritative metapage, and never bothers with
a locally cached version, whereas everywhere else isn't interested in
the metapage fields that were added by commit 857f9c36cd. It seems
unlikely that this bug has affected any user on v11.
Reported-By: Christoph Berg
Bug: #15896
Discussion: https://postgr.es/m/15896-5b25e260fdb0b081%40postgresql.org
Backpatch: 11-, where VACUUM was taught to avoid unnecessary index scans.
This is numbered take 7, and addresses a set of issues around:
- Fixes for typos and incorrect reference names.
- Removal of unneeded comments.
- Removal of unreferenced functions and structures.
- Fixes regarding variable name consistency.
Author: Alexander Lakhin
Discussion: https://postgr.es/m/10bfd4ac-3e7c-40ab-2b2e-355ed15495e8@gmail.com
The logic just added by commit e3899ffd falls back on a 50:50 page split
in the event of a new item that's just to the right of our provisional
"many duplicates" split point. Fix a comment that incorrectly claimed
that the new item had to be just to the left of our provisional split
point.
Backpatch: 12-, just like commit e3899ffd.
Specific ever-decreasing insertion patterns could cause successive
unbalanced nbtree page splits. Problem cases involve a large group of
duplicates to the left, and ever-decreasing insertions to the right.
To fix, detect the situation by considering the newitem offset before
performing a split using nbtsplitloc.c's "many duplicates" strategy. If
the new item was inserted just to the right of our provisional "many
duplicates" split point, infer ever-decreasing insertions and fall back
on a 50:50 (space delta optimal) split. This seems to barely affect
cases that already had acceptable space utilization.
An alternative fix also seems possible. Instead of changing
nbtsplitloc.c split choice logic, we could instead teach _bt_truncate()
to generate a new value for new high keys by interpolating from the
lastleft and firstright key values. That would certainly be a more
elegant fix, but it isn't suitable for backpatching.
Discussion: https://postgr.es/m/CAH2-WznCNvhZpxa__GqAa1fgQ9uYdVc=_apArkW2nc-K3O7_NA@mail.gmail.com
Backpatch: 12-, where the nbtree page split enhancements were introduced.
This is still using the 2.0 version of pg_bsd_indent.
I thought it would be good to commit this separately,
so as to document the differences between 2.0 and 2.1 behavior.
Discussion: https://postgr.es/m/16296.1558103386@sss.pgh.pa.us
It's not safe for nbtree VACUUM to attempt to delete a target page whose
right sibling is already half-dead, since that would fail the
cross-check when VACUUM attempts to re-find a downlink to the right
sibling in the parent page. Logic to prevent this from happening was
added by commit 8da3183780, which addressed a bug in the overhaul of
page deletion that went into PostgreSQL 9.4 (commit efada2b8e9).
VACUUM was made to check the right sibling page, and back off when it
happened to be half-dead already.
However, it is only truly necessary to do the right sibling check on the
leaf level, since that transitively determines if the deletion target's
parent's right sibling page is itself undergoing deletion. Remove the
internal page level check, and add a comment explaining why the leaf
level check alone suffices.
The extra check is also unnecessary due to the fact that internal pages
that are marked half-dead are generally considered corrupt. Commit
efada2b8e9 established the principle that there should never be
half-dead internal pages (internal pages pending deletion are possible,
but that status is never directly represented in the internal page).
VACUUM will complain about corruption when it encounters half-dead
internal pages, so VACUUM is bound to raise an error one way or another
when an nbtree index has a half-dead internal page (contrib/amcheck will
also report that the page is corrupt).
It's possible that a pg_upgrade'd 9.3 database will still have half-dead
internal pages, so it may seem like there is an argument for leaving the
check in place to reliably get a cleaner error message that advises the
user to REINDEX. However, leaf pages are also deleted in the first
phase of deletion prior to PostgreSQL 9.4, so I believe we won't even
attempt to re-find the parent page anyway (we won't have the fully
deleted leaf page as the right sibling of our target page, so we won't
even try to find a downlink for it).
Discussion: https://postgr.es/m/CAH2-Wzm_ntmqJjWLRyKzimFmFvk+BnVAvUpaA4s1h9Ja58woaQ@mail.gmail.com
Remove a Berkeley-era comment above _bt_insertonpg() that admonishes the
reader to grok Lehman and Yao's paper before making any changes. This
made a certain amount of sense back when _bt_insertonpg() was
responsible for most of the things that are now spread across
_bt_insertonpg(), _bt_findinsertloc(), _bt_insert_parent(), and
_bt_split(), but it doesn't work like that anymore.
I believe that this comment alludes to the need to "couple" or "crab"
buffer locks as we ascend the tree as page splits cascade upwards. The
nbtree README already explains this in detail, which seems sufficient.
Besides, the changes to page splits made by commit 40dae7ec53 altered
the exact details of how buffer locks are retained during splits; Lehman
and Yao's original algorithm seems to release the lock on the left child
page/buffer slightly earlier than _bt_insertonpg()/_bt_insert_parent()
can.
Commit fab25024, which taught nbtree to choose candidate split points
more carefully, had _bt_findsplitloc() record all possible split points
in an initial pass over a page that is about to be split. The order
that candidate split points were processed and stored in was assumed to
match the offset number order of split points on an imaginary version of
the page that contains the same items as the original, but also fits
newitem (the item that provoked the split precisely because it didn't
fit).
However, the order of split points in the final array was not quite what
was expected: the split point that makes newitem the firstright item
came after the split point that makes newitem the lastleft item -- not
before. As a result, _bt_findsplitloc() could get confused about the
leftmost and rightmost tuples among all possible split points recorded
for the page. This seems to have no appreciable impact on the quality
of the final split point chosen by _bt_findsplitloc(), but it's still
wrong.
To fix, switch the order in which newitem candidate splits are recorded
in. This also makes it possible to describe candidate split points in
terms of which pair of adjoining tuples enclose the split point within
_bt_findsplitloc(), making it clearer why it's generally safe for
_bt_split() to expect lastleft and firstright tuples.
The term "item pointer" should not be used to refer to ItemIdData
variables, since that is needlessly ambiguous. Only
ItemPointerData/ItemPointer variables should be called item pointers.
To fix, establish the convention that ItemIdData variables should always
be referred to either as "item identifiers" or "line pointers". The
term "item identifier" already predominates in docs and translatable
messages, and so should be the preferred alternative there.
Discussion: https://postgr.es/m/CAH2-Wz=c=MZQjUzde3o9+2PLAPuHTpVZPPdYxN=E4ndQ2--8ew@mail.gmail.com
Commit 8fa30f906b reduced the elevel of a number of "can't happen"
_bt_split() errors from PANIC to ERROR. At the same time, the new right
page buffer for the split could continue to be acquired well before the
critical section. This was possible because it was relatively
straightforward to make sure that _bt_split() could not throw an error,
with a few specific exceptions. The exceptional cases were safe because
they involved specific, well understood errors, making it possible to
consistently zero the right page before actually raising an error using
elog(). There was no danger of leaving around a junk page, provided
_bt_split() stuck to this coding rule.
Commit 8224de4f, which introduced INCLUDE indexes, added code to make
_bt_split() truncate away non-key attributes. This happened at a point
that broke the rule around zeroing the right page in _bt_split(). If
truncation failed (perhaps due to palloc() failure), that would result
in an errant right page buffer with junk contents. This could confuse
VACUUM when it attempted to delete the page, and should be avoided on
general principle.
To fix, reorganize _bt_split() so that truncation occurs before the new
right page buffer is even acquired. A junk page/buffer will not be left
behind if _bt_nonkey_truncate()/_bt_truncate() raise an error.
Discussion: https://postgr.es/m/CAH2-WzkcWT_-NH7EeL=Az4efg0KCV+wArygW8zKB=+HoP=VWMw@mail.gmail.com
Backpatch: 11-, where INCLUDE indexes were introduced.
Commit dd299df818, which added suffix truncation to nbtree, simplified
the WAL record format used by page splits. It became necessary to
explicitly WAL-log the new high key for the left half of a split in all
cases, which relieved the REDO routine from having to reconstruct a new
high key for the left page by copying the first item from the right
page. Remove a comment that referred to the previous practice.
It is no longer possible under any circumstances for nbtree code to
reconstruct a strict lower bound key (parent page's pivot tuple key) for
a right sibling page by retrieving the first item in the right sibling
page.
Commit 3f342839 corrected obsolete comments about buffer locks at the
main _bt_insert_parent() call site, but missed similar obsolete comments
above _bt_insert_parent() itself. Both sets of comments were rendered
obsolete by commit 40dae7ec53, which made the nbtree page split
algorithm more robust. Fix the comments that were missed the first time
around now.
In passing, refine a related _bt_insert_parent() comment about
re-finding the parent page to insert new downlink.
Commit dd299df818, which made heap TID a tiebreaker nbtree index
column, introduced new rules on page space management to make suffix
truncation safe. In general, suffix truncation needs to have a small
amount of extra space available on the new left page when splitting a
leaf page. This is needed in case it turns out that truncation cannot
even "truncate away the heap TID column", resulting in a
larger-than-firstright leaf high key with an explicit heap TID
representation.
Despite all this, CREATE INDEX/nbtsort.c did not account for the
possible need for extra heap TID space on leaf pages when deciding
whether or not a new item could fit on current page. This could lead to
"failed to add item to the index page" errors when CREATE
INDEX/nbtsort.c tried to finish off a leaf page that lacked space for a
larger-than-firstright leaf high key (it only had space for firstright
tuple, which was just short of what was needed following "truncation").
Several conditions needed to be met all at once for CREATE INDEX to
fail. The problem was in the hard limit on what will fit on a page,
which tends to be masked by the soft fillfactor-wise limit. The easiest
way to recreate the problem seems to be a CREATE INDEX on a low
cardinality text column, with tuples that are of non-uniform width,
using a fillfactor of 100.
To fix, bring nbtsort.c in line with nbtsplitloc.c, which already
pessimistically assumes that all leaf page splits will have high keys
that have a heap TID appended.
Reported-By: Andreas Joseph Krogh
Discussion: https://postgr.es/m/VisenaEmail.c5.3ee7fe277d514162.16a6d785bea@tc7-visena
Remove a comment that refers to a coding practice that was fully removed
by commit a8b8f4db, which introduced MarkBufferDirty(). It looks like
the comment was even obsolete before then, since it concerns
write-ordering dependencies with synchronous buffer writes.
Commit dd299df8 made nbtree treat heap TID as a tiebreaker column,
establishing the principle that there is only one correct location (page
and page offset number) for every index tuple, no matter what.
Insertions of tuples into non-unique indexes proceed as if heap TID
(scan key's scantid) is just another user-attribute value, but
insertions into unique indexes are more delicate. The TID value in
scantid must initially be omitted to ensure that the unique index
insertion visits every leaf page that duplicates could be on. The
scantid is set once again after unique checking finishes successfully,
which can force _bt_findinsertloc() to step right one or more times, to
locate the leaf page that the new tuple must be inserted on.
Stepping right within _bt_findinsertloc() was assumed to occur no more
frequently than stepping right within _bt_check_unique(), but there was
one important case where that assumption was incorrect: inserting a
"duplicate" with NULL values. Since _bt_check_unique() didn't do any
real work in this case, it wasn't appropriate for _bt_findinsertloc() to
behave as if it was finishing off a conventional unique insertion, where
any existing physical duplicate must be dead or recently dead.
_bt_findinsertloc() might have to grovel through a substantial portion
of all of the leaf pages in the index to insert a single tuple, even
when there were no dead tuples.
To fix, treat insertions of tuples with NULLs into a unique index as if
they were insertions into a non-unique index: never unset scantid before
calling _bt_search() to descend the tree, and bypass _bt_check_unique()
entirely. _bt_check_unique() is no longer responsible for incoming
tuples with NULL values.
Discussion: https://postgr.es/m/CAH2-Wzm08nr+JPx4jMOa9CGqxWYDQ-_D4wtPBiKghXAUiUy-nQ@mail.gmail.com
_bt_check_unique() failed to invalidate binary search bounds in the
event of a live conflict following commit e5adcb78. This resulted in
problems after waiting for the conflicting xact to commit or abort. The
subsequent call to _bt_check_unique() would restore the initial binary
search bounds, rather than starting a new search. Fix by explicitly
invalidating bounds when it becomes clear that there is a live conflict
that insertion will have to wait to resolve.
Ashutosh Sharma, with a few additional tweaks by me.
Author: Ashutosh Sharma
Reported-By: Ashutosh Sharma
Diagnosed-By: Ashutosh Sharma
Discussion: https://postgr.es/m/CAE9k0PnQp-qr-UYKMSCzdC2FBzdE4wKP41hZrZvvP26dKLonLg@mail.gmail.com
This uses the progress reporting infrastructure added by c16dc1aca5,
adding support for CREATE INDEX and CREATE INDEX CONCURRENTLY.
There are two pieces to this: one is index-AM-agnostic, and the other is
AM-specific. The latter is fairly elaborate for btrees, including
reportage for parallel index builds and the separate phases that btree
index creation uses; other index AMs, which are much simpler in their
building procedures, have simplistic reporting only, but that seems
sufficient, at least for non-concurrent builds.
The index-AM-agnostic part is fairly complete, providing insight into
the CONCURRENTLY wait phases as well as block-based progress during the
index validation table scan. (The index validation index scan requires
patching each AM, which has not been included here.)
Reviewers: Rahila Syed, Pavan Deolasee, Tatsuro Yamada
Discussion: https://postgr.es/m/20181220220022.mg63bhk26zdpvmcj@alvherre.pgsql
Commit 29b64d1d mishandled skipping over truncated high key attributes
during row comparisons. The row comparison key matching loop would loop
forever when a truncated attribute was encountered for a row compare
subkey. Fix by following the example of other code in the loop: advance
the current subkey, or break out of the loop when the last subkey is
reached.
Add test coverage for the relevant _bt_check_rowcompare() code path.
The new test case is somewhat tied to nbtree implementation details,
which isn't ideal, but seems unavoidable.
To support building indexes over tables of different AMs, the scans to
do so need to be routed through the table AM. While moving a fair
amount of code, nearly all the changes are just moving code to below a
callback.
Currently the range based interface wouldn't make much sense for non
block based table AMs. But that seems aceptable for now.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
Previously the xid horizon was only computed during WAL replay. That
had two major problems:
1) It relied on knowing what the table pointed to looks like. That was
easy enough before the introducing of tableam (we knew it had to be
heap, although some trickery around logging the heap relfilenodes
was required). But to properly handle table AMs we need
per-database catalog access to look up the AM handler, which
recovery doesn't allow.
2) Not knowing the xid horizon also makes it hard to support logical
decoding on standbys. When on a catalog table, we need to be able
to conflict with slots that have an xid horizon that's too old. But
computing the horizon by visiting the heap only works once
consistency is reached, but we always need to be able to detect
conflicts.
There's also a secondary problem, in that the current method performs
redundant work on every standby. But that's counterbalanced by
potentially computing the value when not necessary (either because
there's no standby, or because there's no connected backends).
Solve 1) and 2) by moving computation of the xid horizon to the
primary and by involving tableam in the computation of the horizon.
To address the potentially increased overhead, increase the efficiency
of the xid horizon computation for heap by sorting the tids, and
eliminating redundant buffer accesses. When prefetching is available,
additionally perform prefetching of buffers. As this is more of a
maintenance task, rather than something routinely done in every read
only query, we add an arbitrary 10 to the effective concurrency -
thereby using IO concurrency, when not globally enabled. That's
possibly not the perfect formula, but seems good enough for now.
Bumps WAL format, as latestRemovedXid is now part of the records, and
the heap's relfilenode isn't anymore.
Author: Andres Freund, Amit Khandekar, Robert Haas
Reviewed-By: Robert Haas
Discussion:
https://postgr.es/m/20181212204154.nsxf3gzqv3gesl32@alap3.anarazel.dehttps://postgr.es/m/20181214014235.dal5ogljs3bmlq44@alap3.anarazel.dehttps://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
