I8ac4bc4d6 caused lock() to be counted as a write operation. Since
acquiring a lock may by design take a long time (e.g. PageEditStash),
this was causing transactions to be flagged as problematic due to the
large amount of time spent in this "write query".
Bug: T251457
Change-Id: Ic54d6c78b43a463c8f6edc6d65baa671a39ee39c
This reduces regex overhead and reliance on brittle assumptions.
This will also be useful for complex write queries involving WITH.
Some RDBMS types allow writes with in the WITH aliases themselves,
in addition to the main query itself. Checking raw SQL strings for
such things would get fairly complex.
Change-Id: I8ac4bc4d671abf02f97e82c5daf7b21271b85e5e
Allow truncation of multiple tables. This also provides for
a way to avoid risky keywords like CASCADE for Postgres.
For Postgres, use RESTART IDENTITY, which has been supported
since Postgres 8.4.
Avoid TRUNCATE/DELETE queries for empty temp tables, which is
useful for integrations tests that frequently call this method.
Reorganize and tweak the regexes in Database::getTempWrites().
It now recognizes multi-table DROP/TRUNCATE (Postgres-style).
Change-Id: Idd49f118b20ea5a0f7a3e8c00369aabcd45dd44e
Make related DatabaseSqlite::listTables() method directly use query()
rather than high-level wrapper.
Change-Id: I253f83091fb62fd0a449ac78d8846fb69bb808f6
It's highly unlikely that that query would ever error, but if it does
(and apparently it does sometimes somehow) we need to either handle the
error or let it be raised.
Since there doesn't seem to be any particularly sane thing to assume if
there is an error, let's go with "let it be raised".
Bug: T247865
Change-Id: I6c0e08df90eb46953ba5eb6b5a3e8c6f52929564
Use it in MediaWikiIntegrationTestCase for resetting tables.
Also create Database::resetSequencesForTable() helper method from
the resetSequenceForTable() methods in the SQLite/Postgres classes.
Change-Id: I20945e20590e69340b1ce75f6bb2f6972375b00c
Add several new internal methods to help with wrangling
the various formats that rows, conditions, options, and
unique key lists can come in. Remove now unused method
isMultiRowArray().
Add various sanity checks and logging for parameters to
upsert(), replace(), insert(), and insertSelect().
Move DatabasePostgresTest to the integration/ directory.
Change-Id: If5988a6f0816e8da2cbf2fd612e1a3e3a2e9c52f
Code cleanup and hardening (see also: T234014) of Database-related
lib code in MediaWiki core.
Bug: T233342
Change-Id: I3c968f4f5300374253dc80d99596cac50fbeb59e
Quotes started being added to integers in r4984 (August 2004). Before
that, is_numeric() was used to determine whether to add quotes, so
quotes were omitted from numeric strings, which is obviously wrong.
The idea here is to use the type of the variable to hint to the database
as to whether quotes are needed. The results are somewhat inconsistent,
since some callers do not convert numeric strings obtained from user
input to integers. That makes it a more conservative change. Callers can
opt out of unquoted integers by casting them to string.
The reason for doing this is that quoting integers turns out to be not
as harmless as originally assumed. We found a case of it confusing the
MariaDB query planner, causing inappropriate indexes to be used.
I also made addQuotes() consistently return a string, instead of
returning an integer for boolean values. This was already the case for
MySQL, but it seems like a good idea everywhere.
Bug: T238378
Change-Id: I70473280f542ee5ecd79e187f580807410fbd548
If a string starts with "DELIMITER" but the word is *not* followed by
a space, the code will set $this->delimiter to null, possibly resulting
in all kinds of hard to track errors.
Even more problematic: the preg_match was *not* case-insensitive, but the
condition was. Which means that all non-uppercase "delimiter …" would
result in the same error.
I'm not sure if it's worth keeping the additional string comparison for
performance reasons. Probably not. That comparison is slow as well, and
preg_match is surprisingly fast (when the pattern is properly written,
which it is).
Change-Id: I33944b7a2410f77e67ce7450af0359a88d39f1aa
Move the DBO_TRX init logic out of Database::__construct() and into
LoadBalancer since the later already handles setting and clearing this
flag based on transaction rounds starting and ending.
Add 'lazyMasterHandle', 'topologyRole', and 'topologicalMaster' parameters
to Database::factory() and inject them via LoadBalancer all at once in order
to avoid worrying about call order. Move some type casting code to
Database::__construct().
Add IDatabase::getTopologyRole()/getTopologicalMaster().
Use constants for getLBInfo()/setLBInfo() for better usage tracking and
typo resistance.
Change-Id: I437ce434326601e6ba36d9aedc55db396dfe4452
In masterPosWait(), only $pos will have the known active domain/server set
since it usually comes from getMasterPos(). However, the reference position,
from getReplicaDB(), does not have the active domains set since querying
gtid_domain_id on the replica would be incorrect and getting a connection
to the master could be expensive.
Remove obsolete hacks for jobs that used to store master positions.
Also, use the regular Database::query() method for stylistic consistency.
Bug: T224422
Change-Id: I41bbb9f337e46451aa17788dbd446db4a213a5a7
Avoid nesting of the same getWithSetCallback() cache updates.
Also favor accuracy over initial cache use for the case where
there is already a master connection. Add missing "lockTSE"
flag to protect against stampedes updating stale values.
Change serverIsReadOnly() to use SELECT in mysql instead of
SHOW to avoid internal temporary tables.
Bug: T227838
Change-Id: I2b0d680c9c3bdc7aaa1d1e1d6beb2dd203a815f1
Mainly:
* Use oci_new_connect() for Oracle to avoid broken connection reuse
similar to the PGSQL_CONNECT_FORCE_NEW flag in DatabasePostgres
* Set 'client_min_messages' unconditionally for PostgreSQL
* Factor out Database::getConnectExceptionAndLog() helper method
* Use the same style of query() calls in DatabaseOracle::open() as
the other subclasses
* Make sure the Database driver handle field is null on failure
instead of false for sanity
Also:
* Disallow changing of Database handle DBO_* flags after construction
where it does not make sense to change them
* Do not mention DBO_* flags meant for non-config use in $wgDBservers
* Ignore DBO_PERSISTENT for SQLite if DBO_TRX is also set for sanity
* Remove $wgDBOracleDRCP variable to discourage careless automatic
setting of DBO_PERSISTENT that breaks LoadBalancer assumptions
Change-Id: Iea948f7f872294ea8fc5d897fc10c9d29b7141d5
This avoids having two similar fields that have to stay
in sync. Clean up the related error handling for connections.
If a connection handle is unusable, like when essential SET
queries fail, then destroy it.
Also:
* Avoid use of transactions in DatabasePostgres::determineCoreSchema.
* Make sure all subclasses log on connection failure.
* Add schema sanity checks to mysql/sqlite classes.
* Add IDatabase::QUERY_NO_RETRY flag to simplify reasoning about
queries that already run on open() to begin with.
* Remove unused return value of Database::open.
* Remove deprecated Database::reportConnectionError method.
Change-Id: I97beba7ead1523085bda8784234d00c69ef1accc
Move the constants to the bottom and make more of them private.
Place the configuration fields at the top of the list.
Also, move some related fields closer to each other.
In addition:
* Rename the named lock tracking variable to start with the
prefix "session" for consistency.
* Remove unused $preparedArgs field.
* Rename $sessionVars and $rttEstimate fields.
* Use short field documentation syntax.
* Make transaction callback fields private.
Change-Id: I7d78be6744723f4d7bb32a75154564ee04eca0f6
This shares reconnection and retry logic but lacks some of the
restrictions applied to queries that go through the public query()
interface.
Use this in a few places such as doSelectDomain() for mysql/mssql.
Bug: T212284
Change-Id: Ie7341a0e6c4149fc375cc357877486efe9e56eb9
Change from ResultWrapper to IResultWrapper
Change from mysqli_result to resource
Changed mixed to return hint of interface
Document that fieldInfo returns bool
Change-Id: I5572fd41e0e11a2bc2eb116d0c82327499ecc518
Make IDatabase::lastDoneWrites() reflect creation and changes to
the cloned temporary unit test tables but not other temporary tables.
This effects the LB method hasOrMadeRecentMasterChanges(). Other tables
are assumpted to really just be there for temporary calculations rather
acting as test-only ephemeral versions of permanent tables. Treating
writes to the "fake permanent" temp tables more like real permanent
tables means that the tests better align with production.
At the moment, temporary tables still have to use DB_MASTER, given
the assertIsWritableMaster() check in query(). This restriction
can be lifted at some point, when RDBMs compatibility is robust.
Bug: T218388
Change-Id: I4c0d629da254ac2aaf31aae35bd2efc7bc064ac6
