ginRedoRecompress() replays actions over compressed segments of posting list
in-place. However, it might lead to write past pg_upper, because intermediate
state during playing the changes can take more space than both original state
and final state. This commit fixes that by refuse from in-place modification.
Instead page tail is copied once modification is started, and then it's used
as the source of original segments. Backpatch to 9.4 where posting list
compression was introduced.
Reported-by: Sivasubramanian Ramasubramanian
Discussion: https://postgr.es/m/1536091151804.6588%40amazon.com
Author: Alexander Korotkov based on patch from and ideas by Sivasubramanian Ramasubramanian
Review: Sivasubramanian Ramasubramanian
Backpatch-through: 9.4
PostgreSQL 9.4 introduces posting list compression in GIN. This feature
supports online upgrade, so that after pg_upgrade uncompressed posting
lists are compressed on-the-fly. Underlying code appears to always
expect at least one item on uncompressed posting list page. But there
could be completely empty pages, because VACUUM never deletes leftmost
and rightmost pages from posting trees. This commit fixes that.
Reported-by: Sivasubramanian Ramasubramanian
Discussion: https://postgr.es/m/1531867212836.63354%40amazon.com
Author: Sivasubramanian Ramasubramanian, Alexander Korotkov
Backpatch-through: 9.4
I broke this with my WAL format refactoring patch. Before that, the metapage
was read from disk, and modified in-place regardless of the LSN. That was
always a bit silly, as there's no need to read the old page version from
disk disk when we're overwriting it anyway. So that was changed in 9.5, but
I failed to add a GinInitPage call to initialize the page-headers correctly.
Usually you wouldn't notice, because the metapage is already in the page
cache and is not zeroed.
One way to reproduce this is to perform a VACUUM on an already vacuumed
table (so that the vacuum has no real work to do), immediately after a
checkpoint, and then perform an immediate shutdown. After recovery, the
page headers of the metapage will be incorrectly all-zeroes.
Reported by Jeff Janes
Use "a" and "an" correctly, mostly in comments. Two error messages were
also fixed (they were just elogs, so no translation work required). Two
function comments in pg_proc.h were also fixed. Etsuro Fujita reported one
of these, but I found a lot more with grep.
Also fix a few other typos spotted while grepping for the a/an typos.
For example, "consists out of ..." -> "consists of ...". Plus a "though"/
"through" mixup reported by Euler Taveira.
Many of these typos were in old code, which would be nice to backpatch to
make future backpatching easier. But much of the code was new, and I didn't
feel like crafting separate patches for each branch. So no backpatching.
Each WAL record now carries information about the modified relation and
block(s) in a standardized format. That makes it easier to write tools that
need that information, like pg_rewind, prefetching the blocks to speed up
recovery, etc.
There's a whole new API for building WAL records, replacing the XLogRecData
chains used previously. The new API consists of XLogRegister* functions,
which are called for each buffer and chunk of data that is added to the
record. The new API also gives more control over when a full-page image is
written, by passing flags to the XLogRegisterBuffer function.
This also simplifies the XLogReadBufferForRedo() calls. The function can dig
the relation and block number from the WAL record, so they no longer need to
be passed as arguments.
For the convenience of redo routines, XLogReader now disects each WAL record
after reading it, copying the main data part and the per-block data into
MAXALIGNed buffers. The data chunks are not aligned within the WAL record,
but the redo routines can assume that the pointers returned by XLogRecGet*
functions are. Redo routines are now passed the XLogReaderState, which
contains the record in the already-disected format, instead of the plain
XLogRecord.
The new record format also makes the fixed size XLogRecord header smaller,
by removing the xl_len field. The length of the "main data" portion is now
stored at the end of the WAL record, and there's a separate header after
XLogRecord for it. The alignment padding at the end of XLogRecord is also
removed. This compansates for the fact that the new format would otherwise
be more bulky than the old format.
Reviewed by Andres Freund, Amit Kapila, Michael Paquier, Alvaro Herrera,
Fujii Masao.
Every redo routine uses the same idiom to determine what to do to a page:
check if there's a backup block for it, and if not read, the buffer if the
block exists, and check its LSN. Refactor that into a common function,
XLogReadBufferForRedo, making all the redo routines shorter and more
readable.
This has no user-visible effect, and makes no changes to the WAL format.
Reviewed by Andres Freund, Alvaro Herrera, Michael Paquier.
log_newpage is used by many indexams, in addition to heap, but for
historical reasons it's always been part of the heapam rmgr. Starting with
9.3, we have another WAL record type for logging an image of a page,
XLOG_FPI. Simplify things by moving log_newpage and log_newpage_buffer to
xlog.c, and switch to using the XLOG_FPI record type.
Bump the WAL version number because the code to replay the old HEAP_NEWPAGE
records is removed.
To-be-deleted list pages contain no useful information, as they are being
deleted, but we must still protect the writes from being torn by a crash
after a partial write. To do that, re-initialize the pages on WAL replay.
Jeff Janes caught this with a test program to test partial writes.
Backpatch to all supported versions.
In writeListPage, never take a full-page image of the page, because we
have all the information required to re-initialize in the WAL record
anyway. Before this fix, a full-page image was always generated, unless
full_page_writes=off, because when the page is initialized its LSN is
always 0. In stable-branches, keep the code to restore the backup blocks
if they exist, in case that the WAL is generated with an older minor
version, but in master Assert that there are no full-page images.
In the redo routine, add missing "off++". Otherwise the tuples are added
to the page in reverse order. That happens to be harmless because we
always scan and remove all the tuples together, but it was clearly wrong.
Also, it was masked by the first bug unless full_page_writes=off, because
the page was always restored from a full-page image.
Backpatch to all supported versions.
This allows squeezing out the unused space in full-page writes. And more
importantly, it can be a useful debugging aid.
In hindsight we should've done this back when GIN was added - we wouldn't
need the 'maxoff' field in the page opaque struct if we had used pd_lower
and pd_upper like on normal pages. But as long as there can be pages in the
index that have been binary-upgraded from pre-9.4 versions, we can't rely
on that, and have to continue using 'maxoff'.
Most of the code churn comes from renaming some macros, now that they're
used on internal pages, too.
This change is completely backwards-compatible, no effect on pg_upgrade.
Forgot to set the incomplete-split flag on the left page half, in redo of a
page split.
Spotted this by comparing the page contents on master and standby, after
inserting/applying each WAL record.
Inserting (in retail) into the new 9.4 format GIN posting tree created much
larger WAL records than in 9.3. The previous strategy to WAL logging was
basically to log the whole page on each change, with the exception of
completely unmodified segments up to the first modified one. That was not
too bad when appending to the end of the page, as only the last segment had
to be WAL-logged, but per Fujii Masao's testing, even that produced 2x the
WAL volume that 9.3 did.
The new strategy is to keep track of changes to the posting lists in a more
fine-grained fashion, and also make the repacking" code smarter to avoid
decoding and re-encoding segments unnecessarily.
We don't take a full-page image of the GIN metapage; instead, the WAL record
contains all the information required to reconstruct it from scratch. But
to avoid torn page hazards, we must re-initialize it from the WAL record
every time, even if it already has a greater LSN, similar to how normal full
page images are restored.
This was highly unlikely to cause any problems in practice, because the GIN
metapage is small. We rely on an update smaller than a 512 byte disk sector
to be atomic elsewhere, at least in pg_control. But better safe than sorry,
and this would be easy to overlook if more fields are added to the metapage
so that it's no longer small.
Reported by Noah Misch. Backpatch to all supported versions.
The maxoff field is not used in the new, compressed page format. Let's
reset it when converting an old-format page to the new format. The code
won't care either way, but this makes it possible to use the field for
something else in the future.
GIN posting lists are now encoded using varbyte-encoding, which allows them
to fit in much smaller space than the straight ItemPointer array format used
before. The new encoding is used for both the lists stored in-line in entry
tree items, and in posting tree leaf pages.
To maintain backwards-compatibility and keep pg_upgrade working, the code
can still read old-style pages and tuples. Posting tree leaf pages in the
new format are flagged with GIN_COMPRESSED flag, to distinguish old and new
format pages. Likewise, entry tree tuples in the new format have a
GIN_ITUP_COMPRESSED flag set in a bit that was previously unused.
This patch bumps GIN_CURRENT_VERSION from 1 to 2. New indexes created with
version 9.4 will therefore have version number 2 in the metapage, while old
pg_upgraded indexes will have version 1. The code treats them the same, but
it might be come handy in the future, if we want to drop support for the
uncompressed format.
Alexander Korotkov and me. Reviewed by Tomas Vondra and Amit Langote.
Replace it with an approach similar to what GiST uses: when a page is split,
the left sibling is marked with a flag indicating that the parent hasn't been
updated yet. When the parent is updated, the flag is cleared. If an insertion
steps on a page with the flag set, it will finish split before proceeding
with the insertion.
The post-recovery cleanup mechanism was never totally reliable, as insertion
to the parent could fail e.g because of running out of memory or disk space,
leaving the tree in an inconsistent state.
This also divides the responsibility of WAL-logging more clearly between
the generic ginbtree.c code, and the parts specific to entry and posting
trees. There is now a common WAL record format for insertions and deletions,
which is written by ginbtree.c, followed by tree-specific payload, which is
returned by the placetopage- and split- callbacks.
Separate the insertion payload from the more static portions of GinBtree.
GinBtree now only contains information related to searching the tree, and
the information of what to insert is passed separately.
Add root block number to GinBtree, instead of passing it around all the
functions as argument.
Split off ginFinishSplit() from ginInsertValue(). ginFinishSplit is
responsible for finding the parent and inserting the downlink to it.
This creates a new gin-btree callback function for creating a downlink for
a page. Previously, ginxlog.c duplicated the logic used during normal
operation.
It makes for cleaner code to have separate Get/Add functions for PostingItems
and ItemPointers. A few callsites that have to deal with both types need to
be duplicated because of this, but all the callers have to know which one
they're dealing with anyway. Overall, this reduces the amount of casting
required.
Extracted from Alexander Korotkov's larger patch to change the data page
format.
Remove use of PageSetTLI() from all page manipulation functions
and adjust README to indicate change in the way we make changes
to pages. Repurpose those bytes into the pd_checksum field and
explain how that works in comments about page header.
Refactoring ahead of actual feature patch which would make use
of the checksum field, arriving later.
Jeff Davis, with comments and doc changes by Simon Riggs
Direction suggested by Robert Haas; many others providing
review comments.
This gets rid of XLByteLT, XLByteLE, XLByteEQ and XLByteAdvance.
These were useful for brevity when XLogRecPtrs were split in
xlogid/xrecoff; but now that they are simple uint64's, they are just
clutter. The only downside to making this change would be ease of
backporting patches, but that has been negated by other substantive
changes to the involved code anyway. The clarity of simpler expressions
makes the change worthwhile.
Most of the changes are mechanical, but in a couple of places, the patch
author chose to invert the operator sense, making the code flow more
logical (and more in line with preceding comments).
Author: Andres Freund
Eyeballed by Dimitri Fontaine and Alvaro Herrera
Most of the replay functions for WAL record types that modify more than
one page failed to ensure that those pages were locked correctly to ensure
that concurrent queries could not see inconsistent page states. This is
a hangover from coding decisions made long before Hot Standby was added,
when it was hardly necessary to acquire buffer locks during WAL replay
at all, let alone hold them for carefully-chosen periods.
The key problem was that RestoreBkpBlocks was written to hold lock on each
page restored from a full-page image for only as long as it took to update
that page. This was guaranteed to break any WAL replay function in which
there was any update-ordering constraint between pages, because even if the
nominal order of the pages is the right one, any mixture of full-page and
non-full-page updates in the same record would result in out-of-order
updates. Moreover, it wouldn't work for situations where there's a
requirement to maintain lock on one page while updating another. Failure
to honor an update ordering constraint in this way is thought to be the
cause of bug #7648 from Daniel Farina: what seems to have happened there
is that a btree page being split was rewritten from a full-page image
before the new right sibling page was written, and because lock on the
original page was not maintained it was possible for hot standby queries to
try to traverse the page's right-link to the not-yet-existing sibling page.
To fix, get rid of RestoreBkpBlocks as such, and instead create a new
function RestoreBackupBlock that restores just one full-page image at a
time. This function can be invoked by WAL replay functions at the points
where they would otherwise perform non-full-page updates; in this way, the
physical order of page updates remains the same no matter which pages are
replaced by full-page images. We can then further adjust the logic in
individual replay functions if it is necessary to hold buffer locks
for overlapping periods. A side benefit is that we can simplify the
handling of concurrency conflict resolution by moving that code into the
record-type-specfic functions; there's no more need to contort the code
layout to keep conflict resolution in front of the RestoreBkpBlocks call.
In connection with that, standardize on zero-based numbering rather than
one-based numbering for referencing the full-page images. In HEAD, I
removed the macros XLR_BKP_BLOCK_1 through XLR_BKP_BLOCK_4. They are
still there in the header files in previous branches, but are no longer
used by the code.
In addition, fix some other bugs identified in the course of making these
changes:
spgRedoAddNode could fail to update the parent downlink at all, if the
parent tuple is in the same page as either the old or new split tuple and
we're not doing a full-page image: it would get fooled by the LSN having
been advanced already. This would result in permanent index corruption,
not just transient failure of concurrent queries.
Also, ginHeapTupleFastInsert's "merge lists" case failed to mark the old
tail page as a candidate for a full-page image; in the worst case this
could result in torn-page corruption.
heap_xlog_freeze() was inconsistent about using a cleanup lock or plain
exclusive lock: it did the former in the normal path but the latter for a
full-page image. A plain exclusive lock seems sufficient, so change to
that.
Also, remove gistRedoPageDeleteRecord(), which has been dead code since
VACUUM FULL was rewritten.
Back-patch to 9.0, where hot standby was introduced. Note however that 9.0
had a significantly different WAL-logging scheme for GIST index updates,
and it doesn't appear possible to make that scheme safe for concurrent hot
standby queries, because it can leave inconsistent states in the index even
between WAL records. Given the lack of complaints from the field, we won't
work too hard on fixing that branch.
XLOG_GIN_UPDATE_META_PAGE and XLOG_GIN_DELETE_LISTPAGE records were printed
with a list link field labeled as "blkno", which was confusing, especially
when the link was empty (InvalidBlockNumber). Print the metapage block
number instead, since that's what's actually being updated. We could
include the link values too as a separate field, but not clear it's worth
the trouble.
Back-patch to 8.4 where the dubious code was added.
In commit 4016bdef8a I fixed a bunch of
ginxlog.c bugs having to do with not handling XLogReadBuffer failures
correctly. However, in ginRedoUpdateMetapage and ginRedoDeleteListPages,
I unaccountably thought that failure to read the metapage would be
impossible and just put in an elog(PANIC) call. This is of course wrong:
failure is exactly what will happen if the index got dropped (or rebuilt)
between creation of the WAL record and the crash we're trying to recover
from. I believe this explains Nicholas Wilson's recent report of these
errors getting reached.
Also, fix memory leak in forgetIncompleteSplit. This wasn't of much
concern when the code was written, but in a long-running standby server
page split records could be expected to accumulate indefinitely.
Back-patch to 8.4 --- before that, GIN didn't have a metapage.
A simple thinko in ginRedoUpdateMetapage, namely failing to increment a
loop counter, led to inserting records into the last pending-list page in
the wrong order (the opposite of that intended). So far as I can tell,
this would not upset the code that eventually flushes pending items into
the main part of the GIN index. But it did break the code that searched
the pending list for matches, resulting in transient failure to find
matching entries during index lookups, as illustrated in bug #6307 from
Maksym Boguk.
Back-patch to 8.4 where the incorrect code was introduced.
Per my recent proposal(s). Null key datums can now be returned by
extractValue and extractQuery functions, and will be stored in the index.
Also, placeholder entries are made for indexable items that are NULL or
contain no keys according to extractValue. This means that the index is
now always complete, having at least one entry for every indexed heap TID,
and so we can get rid of the prohibition on full-index scans. A full-index
scan is implemented much the same way as partial-match scans were already:
we build a bitmap representing all the TIDs found in the index, and then
drive the results off that.
Also, introduce a concept of a "search mode" that can be requested by
extractQuery when the operator requires matching to empty items (this is
just as cheap as matching to a single key) or requires a full index scan
(which is not so cheap, but it sure beats failing or giving wrong answers).
The behavior remains backward compatible for opclasses that don't return
any null keys or request a non-default search mode.
Using these features, we can now make the GIN index opclass for anyarray
behave in a way that matches the actual anyarray operators for &&, <@, @>,
and = ... which it failed to do before in assorted corner cases.
This commit fixes the core GIN code and ginarrayprocs.c, updates the
documentation, and adds some simple regression test cases for the new
behaviors using the array operators. The tsearch and contrib GIN opclass
support functions still need to be looked over and probably fixed.
Another thing I intend to fix separately is that this is pretty inefficient
for cases where more than one scan condition needs a full-index search:
we'll run duplicate GinScanEntrys, each one of which builds a large bitmap.
There is some existing logic to merge duplicate GinScanEntrys but it needs
refactoring to make it work for entries belonging to different scan keys.
Note that most of gin.h has been split out into a new file gin_private.h,
so that gin.h doesn't export anything that's not supposed to be used by GIN
opclasses or the rest of the backend. I did quite a bit of other code
beautification work as well, mostly fixing comments and choosing more
appropriate names for things.
The GIN code has absolutely no business exporting GIN-specific functions
with names as generic as compareItemPointers() or newScanKey(); that's
just trouble waiting to happen. I got annoyed about this again just now
and decided to fix it. This commit ensures that all global symbols
defined in access/gin/ have names including "gin" or "Gin". There were a
couple of cases, like names involving "PostingItem", where arguably the
names were already sufficiently nongeneric; but I figured as long as I was
risking creating merge problems for unapplied GIN patches I might as well
impose a uniform policy.
I didn't touch any static symbol names. There might be some places
where it'd be appropriate to rename some static functions to match
siblings that are exported, but I'll leave that for another time.
The better estimate requires more statistics than we previously stored:
in particular, counts of "entry" versus "data" pages within the index,
as well as knowledge of the number of distinct key values. We collect
this information during initial index build and update it during VACUUM,
storing the info in new fields on the index metapage. No initdb is
required because these fields will read as zeroes in a pre-existing
index, and the new gincostestimate code is coded to behave (reasonably)
sanely if they are zeroes.
Teodor Sigaev, reviewed by Jan Urbanski, Tom Lane, and Itagaki Takahiro.
The original coding was quite sloppy about handling the case where
XLogReadBuffer fails (because the page has since been deleted). This
would result in either "bad buffer id: 0" or an Assert failure during
replay, if indeed the page were no longer there. In a couple of places
it also neglected to check whether the change had already been applied,
which would probably result in corrupted index contents. I believe that
bug #5703 is an instance of the first problem. These issues could show up
without replication, but only if you were unfortunate enough to crash
between modification of a GIN index and the next checkpoint.
Back-patch to 8.2, which is as far back as GIN has WAL support.
Enabled by recovery_connections = on (default) and forcing archive recovery using a recovery.conf. Recovery processing now emulates the original transactions as they are replayed, providing full locking and MVCC behaviour for read only queries. Recovery must enter consistent state before connections are allowed, so there is a delay, typically short, before connections succeed. Replay of recovering transactions can conflict and in some cases deadlock with queries during recovery; these result in query cancellation after max_standby_delay seconds have expired. Infrastructure changes have minor effects on normal running, though introduce four new types of WAL record.
New test mode "make standbycheck" allows regression tests of static command behaviour on a standby server while in recovery. Typical and extreme dynamic behaviours have been checked via code inspection and manual testing. Few port specific behaviours have been utilised, though primary testing has been on Linux only so far.
This commit is the basic patch. Additional changes will follow in this release to enhance some aspects of behaviour, notably improved handling of conflicts, deadlock detection and query cancellation. Changes to VACUUM FULL are also required.
Simon Riggs, with significant and lengthy review by Heikki Linnakangas, including streamlined redesign of snapshot creation and two-phase commit.
Important contributions from Florian Pflug, Mark Kirkwood, Merlin Moncure, Greg Stark, Gianni Ciolli, Gabriele Bartolini, Hannu Krosing, Robert Haas, Tatsuo Ishii, Hiroyuki Yamada plus support and feedback from many other community members.
multiple index entries in a holding area before adding them to the main index
structure. This helps because bulk insert is (usually) significantly faster
than retail insert for GIN.
This patch also removes GIN support for amgettuple-style index scans. The
API defined for amgettuple is difficult to support with fastupdate, and
the previously committed partial-match feature didn't really work with
it either. We might eventually figure a way to put back amgettuple
support, but it won't happen for 8.4.
catversion bumped because of change in GIN's pg_am entry, and because
the format of GIN indexes changed on-disk (there's a metapage now,
and possibly a pending list).
Teodor Sigaev
be used instead of the normal exclusive lock, and make WAL redo functions
responsible for calling RestoreBkpBlocks(). They know better what kind of a
lock they need.
At the moment, this just moves things around with no functional change, but
makes the hot standby patch that's under review cleaner.
forks. XLogOpenRelation() and the associated light-weight relation cache in
xlogutils.c is gone, and XLogReadBuffer() now takes a RelFileNode as argument,
instead of Relation.
For functions that still need a Relation struct during WAL replay, there's a
new function called CreateFakeRelcacheEntry() that returns a fake entry like
XLogOpenRelation() used to.
