mirror of
https://github.com/MariaDB/server.git
synced 2026-01-13 06:03:12 +00:00
7b54e04965
Atomic CREATE OR REPLACE allows to keep an old table intact if the
command fails or during the crash. That is done by renaming the
original table to temporary name, as a backup and restoring it if the
CREATE fails. When the command is complete and logged the backup
table is deleted.
Atomic replace algorithm
Two DDL chains are used for CREATE OR REPLACE:
ddl_log_state_create (C) and ddl_log_state_rm (D).
1. (C) Log rename of ORIG to TMP table (Rename TMP to original).
2. Rename orignal to TMP.
3. (C) Log CREATE_TABLE_ACTION of ORIG (drops ORIG);
4. Do everything with ORIG (like insert data)
5. (D) Log drop of TMP
6. Write query to binlog (this marks (C) to be closed in
case of failure)
7. Execute drop of TMP through (D)
8. Close (C) and (D)
If there is a failure before 6) we revert the changes in (C)
Chain (D) is only executed if 6) succeded (C is closed on
crash recovery).
Foreign key errors will be found at the 1) stage.
Additional notes
- CREATE TABLE without REPLACE and temporary tables is not affected
by this commit.
set @@drop_before_create_or_replace=1 can be used to
get old behaviour where existing tables are dropped
in CREATE OR REPLACE.
- CREATE TABLE is reverted if binlogging the query fails.
- Engines having HTON_EXPENSIVE_RENAME flag set are not affected by
this commit. Conflicting tables marked with this flag will be
deleted with CREATE OR REPLACE.
- Replication execution is not affected by this commit.
- Replication will first drop the conflicting table and then
creating the new one.
- CREATE TABLE .. SELECT XID usage is fixed and now there is no need
to log DROP TABLE via DDL_CREATE_TABLE_PHASE_LOG (see comments in
do_postlock()). XID is now correctly updated so it disables
DDL_LOG_DROP_TABLE_ACTION. Note that binary log is flushed at the
final stage when the table is ready. So if we have XID in the
binary log we don't need to drop the table.
- Three variations of CREATE OR REPLACE handled:
1. CREATE OR REPLACE TABLE t1 (..);
2. CREATE OR REPLACE TABLE t1 LIKE t2;
3. CREATE OR REPLACE TABLE t1 SELECT ..;
- Test case uses 6 combinations for engines (aria, aria_notrans,
myisam, ib, lock_tables, expensive_rename) and 2 combinations for
binlog types (row, stmt). Combinations help to check differences
between the results. Error failures are tested for the above three
variations.
- expensive_rename tests CREATE OR REPLACE without atomic
replace. The effect should be the same as with the old behaviour
before this commit.
- Triggers mechanism is unaffected by this change. This is tested in
create_replace.test.
- LOCK TABLES is affected. Lock restoration must be done after new
table is created or TMP is renamed back to ORIG
- Moved ddl_log_complete() from send_eof() to finalize_ddl(). This
checkpoint was not executed before for normal CREATE TABLE but is
executed now.
- CREATE TABLE will now rollback also if writing to the binary
logging failed. See rpl_gtid_strict.test
backup ddl log changes
- In case of a successfull CREATE OR REPLACE we only log
the CREATE event, not the DROP TABLE event of the old table.
ddl_log.cc changes
ddl_log_execute_action() now properly return error conditions.
ddl_log_disable_entry() added to allow one to disable one entry.
The entry on disk is still reserved until ddl_log_complete() is
executed.
On XID usage
Like with all other atomic DDL operations XID is used to avoid
inconsistency between master and slave in the case of a crash after
binary log is written and before ddl_log_state_create is closed. On
recovery XIDs are taken from binary log and corresponding DDL log
events get disabled. That is done by
ddl_log_close_binlogged_events().
On linking two chains together
Chains are executed in the ascending order of entry_pos of execute
entries. But entry_pos assignment order is undefined: it may assign
bigger number for the first chain and then smaller number for the
second chain. So the execution order in that case will be reverse:
second chain will be executed first.
To avoid that we link one chain to another. While the base chain
(ddl_log_state_create) is active the secondary chain
(ddl_log_state_rm) is not executed. That is: only one chain can be
executed in two linked chains.
The interface ddl_log_link_chains() was defined in "MDEV-22166
ddl_log_write_execute_entry() extension".
Atomic info parameters in HA_CREATE_INFO
Many functions in CREATE TABLE pass the same parameters. These
parameters are part of table creation info and should be in
HA_CREATE_INFO (or whatever). Passing parameters via single
structure is much easier for adding new data and
refactoring.
InnoDB changes
Added ha_innobase::can_be_renamed_to_backup() to check if
a table with foreign keys can be renamed.
Aria changes:
- Fixed issue in Aria engine with CREATE + locked tables
that data was not properly commited in some cases in
case of crashes.
Other changes:
- Removed some auto variables in log.cc for better code readability.
- Fixed old bug that CREATE ... SELECT would not be able to auto repair
a table that is part of the SELECT.
- Marked MyISAM that it does not support ROLLBACK (not required but
done for better consistency with other engines).
Known issues:
- InnoDB tables with foreign key definitions are not fully supported
with atomic create and replace:
- ha_innobase::can_be_renamed_to_backup() can detect some cases
where InnoDB does not support renaming table with foreign key
constraints. In this case MariaDB will drop the old table before
creating the new one.
The detected cases are:
- The new and old table is using the same foreign key constraint
name.
- The old table has self referencing constraints.
- If the old and new table uses the same name for a constraint the
create of the new table will fail. The orignal table will be
restored in this case.
- The above issues will be fixed in a future commit.
- CREATE OR REPLACE TEMPORARY table is not full atomic. Any conflicting
table will always be dropped before creating a new one. (Old behaviour).
Bug fixes related to this MDEV:
MDEV-36435 Assertion failure in finalize_locked_tables()
MDEV-36439 Assertion `thd_arg->lex->sql_command != SQLCOM_CREATE_SEQUENCE...
MDEV-36498 Failed CoR in non-atomic mode no longer generates DROP in RBR...
MDEV-36508 Temporary files #sql-create-....frm occasionally stay after
crash recovery
MDEV-38479 Crash in CREATE OR REPLACE SEQUENCE when new sequence cannot
be created
MDEV-36497 Assertion failure after atomic CoR with Aria under lock in
transactional context
InnoDB related changes:
- ha_innodb::rename_table() does not handle foreign key constraint
when renaming an normal table to internal tempory tables. This
causes problems for CREATE OR REPLACE as the old constraints causes
failure when creating a new table with the same constraints.
This is fixed inside InnoDB by not threating tempfiles (#sql-create-..),
created as part of CREATE OR REPLACE, as temporary files.
- In ha_innobase::delete_table(), ignore checking of constraints when
dropping a #sql-create temporary table.
- In tablename_to_filename() and filename_to_tablename(), don't do
filename conversion for internal temporary tables (#sql-...)
Other things:
- maria_create_trn_for_mysql() does not register a new transaction
handler for commits. This was needed to ensure create or replace
will not end with an active transaction.
- We do not get anymore warnings about "Engine not supporting atomic
create" when doing a legal CREATE OR REPLACE on a table with
foreign key constraints.
- Updated VIDEX engine flags to disable CREATE SEQUENCE.
Reverted commits:
MDEV-36685 "CREATE-SELECT may lose in binlog side-effects of
stored-routine" as it did not take into account that it safe to clear
binlogs if the created table is non transactional and there are no
other non transactional tables used.
- This was done because it caused extra logging when it is not needed
(not using any non transactional tables) and it also did not solve
side effects when using statement based loggging.
File : README
Author : Richard A. O'Keefe.
Updated: 30 April 1984
Purpose: Explain the new strings package.
The UNIX string libraries (described in the string(3) manual page)
differ from UNIX to UNIX (e.g. strtok is not in V7 or 4.1bsd). Worse,
the sources are not in the public domain, so that if there is a string
routine which is nearly what you want but not quite you can't take a
copy and modify it. And of course C programmers on non-UNIX systems
are at the mercy of their supplier.
This package was designed to let me do reasonable things with C's
strings whatever UNIX (V7, PaNiX, UX63, 4.1bsd) I happen to be using.
Everything in the System III manual is here and does just what the S3
manual says it does. There are also lots of new goodies. I'm sorry
about the names, but the routines do have to work on asphyxiated-at-
birth systems which truncate identifiers. The convention is that a
routine is called
str [n] [c] <operation>
If there is an "n", it means that the function takes an (int) "length"
argument, which bounds the number of characters to be moved or looked
at. If the function has a "set" argument, a "c" in the name indicates
that the complement of the set is used. Functions or variables whose
names start with _ are support routines which aren't really meant for
general use. I don't know what the "p" is doing in "strpbrk", but it
is there in the S3 manual so it's here too. "istrtok" does not follow
this rule, but with 7 letters what can you do?
I have included new versions of atoi(3) and atol(3) as well. They
use a new primitive str2int, which takes a pair of bounds and a radix,
and does much more thorough checking than the normal atoi and atol do.
The result returned by atoi & atol is valid if and only if errno == 0.
There is also an output conversion routine int2str, with itoa and ltoa
as interface macros. Only after writing int2str did I notice that the
str2int routine has no provision for unsigned numbers. On reflection,
I don't greatly care. I'm afraid that int2str may depend on your "C"
compiler in unexpected ways. Do check the code with -S.
Several of these routines have "asm" inclusions conditional on the
VaxAsm option. These insertions can make the routines which have them
quite a bit faster, but there is a snag. The VAX architects, for some
reason best known to themselves and their therapists, decided that all
"strings" were shorter than 2^16 bytes. Even when the length operands
are in 32-bit registers, only 16 bits count. So the "asm" versions do
not work for long strings. If you can guarantee that all your strings
will be short, define VaxAsm in the makefile, but in general, and when
using other machines, do not define it.
To use this library, you need the "strings.a" library file and the
"strings.h" and "ctypes.h" header files. The other header files are
for compiling the library itself, though if you are hacking extensions
you may find them useful. General users really shouldn't see them.
I've defined a few macros I find useful in "strings.h"; if you have no
need for "index", "rindex", "streql", and "beql", just edit them out.
On the 4.1bsd system I am using declaring all these functions 'extern'
does not mean that they will all be loaded; but only the ones you use.
When using lesser systems you may find it necessary to break strings.h
up, or you could get by with just adding "extern" declarations for the
functions you want as you need them. Many of these functions have the
same names as functions in the "standard C library", by design as this
is a replacement/reimplementation of part of that library. So you may
have to talk the loader into loading this library first. Again, I've
found no problems on 4.1bsd.
You may wonder at my failure to provide manual pages for this code.
For the things in V7, 4.?, or SIII, you should be able to use whichever
manual page came with that system, and anything I might write would be
so like it as to raise suspicions of violating AT&T copyrights. In the
sources you will find comments which provide far more documentation for
these routines than AT&T ever provided for their strings stuff, I just
don't happen to have put it in nroff -man form. Had I done so, the .3
files would have outbulked the .c files!
These files are in the public domain. This includes getopt.c, which
is the work of Henry Spencer, University of Toronto Zoology, who says of
it "None of this software is derived from Bell software. I had no access
to the source for Bell's versions at the time I wrote it. This software
is hereby explicitly placed in the public domain. It may be used for
any purpose on any machine by anyone." I would greatly prefer it if *my*
material received no military use.