Also improved the atomicity and affected row count logic for
insert/replace with sqlite.
Also remove unused "fileHandle" code from insert().
Change-Id: If7b9148fd44f3a958899885753c7c86ba66bf193
LoadBalancer uses Database::getDomainId() for deciding which keys to use
in the foreign connection handle arrays. This method should reflect any
changes made to the DB selection.
If the query fails, then do not change domain field. This is the sort of
approach that LoadBalancer is expects in openForeignConnection(). Also,
throw an exception when selectDB() fails.
The db/schema/prefix fields of Database no longer exist in favor of just
using the newer currentDomain field.
Also:
* Add IDatabase::selectDomain() method and made selectDB() wrap it.
* Extract the DB name from sqlite files if not explicitly provided.
* Fix inconsistent open() return values from Database subclasses.
* Make a relationSchemaQualifier() method to handle the concern of
omitting schema names in queries. The means that getDomainId() can
still return the right value, rather than confusingly omitt the schema.
* Make RevisionStore::checkDatabaseWikiId() account for the domain schema.
Unlike d2a4d614fc, this does not incorrectly assume the storage is
always for the current wiki domain. Also, LBFactorySingle sets the local
domain so it is defined even in install.php.
* Make RevisionStoreDbTestBase actually set the LoadBalancer local domain.
* Make RevisionTest::testLoadFromTitle() account for the domain schema.
Bug: T193565
Change-Id: I6e51cd54c6da78830b38906b8c46789c79498ab5
It can be true for successful write queries, not just false.
f3a197e49b introduced a caller which calls numRows() on the return
value of CREATE TEMPORARY TABLE queries, and it improved guarding of
numRows() in the PostgreSQL and SQLite cases accordingly, but it
neglected MySQL.
Bug: T201900
Change-Id: I8ae754a2518d9e47b093c31c20d98daaba913513
This is not called externally and there is little reason for that to
change. The current caller pattern is to use factory(), possibly with
initConnection() afterwards, or to use a LoadBalancer to begin with.
Change-Id: Ib1fdd5c960f1ed877fcd17bcb99b999d5d894716
Find: /isset\(\s*([^()]+?)\s*\)\s*\?\s*\1\s*:\s*/
Replace with: '\1 ?? '
(Everywhere except includes/PHPVersionCheck.php)
(Then, manually fix some line length and indentation issues)
Then manually reviewed the replacements for cases where confusing
operator precedence would result in incorrect results
(fixing those in I478db046a1cc162c6767003ce45c9b56270f3372).
Change-Id: I33b421c8cb11cdd4ce896488c9ff5313f03a38cf
When reading through DatabaseMysqlBase::open(), it was not
obvious that execution would not continue after the conditional
`!$this->conn` block, given it ends in a method call, without
return or throw. I considered adding a return statement after it
for clarity, but it seems in this case it might make more sense
to throw directly given $error here has already gone through a
fallback to getLastError() a few lines up.
Replace the other three calls to reportConnectionError() as well,
which previously passed a useful string that was overwritten
with lastError(). Instead, log both. And make their call to
queryLogger->error() match the previous ones to have an 'error' as well.
This leaves reportConnectionError() as being unused, except for
a call from LoadBalancer. That call was problematic because
it was inside a conditional for IDatabase, but the method isn't
part of that interface. Replace it with a direct throw as well.
Deprecate the method as its now unused in core, and also remove its
'# New method' comment which hasn't made sense since r75341 (16cded8b32).
Change-Id: I0f2ef00ba44bf7090a3ce54edeb8c7e8e543e46a
If it is known that master writes will use GTIDs with a certain
domain and server ID, ignore other ones on masterPosWait().
This restores ceb7d61ee except it uses the same server variables
and MySQLMasterPos preserves the active server/domain information
across serialization.
Change-Id: I1a4f143adcbec642966d7d1a55edb0f414a7f0c4
Handle all errors in query() that might have caused rollback by
putting the Database handle into an error state that can only be
resolved by cancelAtomic() or rollback(). Other queries will be
rejected until then.
This results in more immediate exceptions in some cases where
atomic section mismatch errors would have been thrown, such as a
an error bubbling up from a child atomic section. Most cases were
a try/catch block assumes that only the statement was rolled back
now result in an error and rollback.
Callers using try/catch to handle key conflicts should instead use
SELECT FOR UPDATE to find conflicts beforehand, or use IGNORE, or
the upsert()/replace() methods. The try/catch pattern is unsafe and
no longer allowed, except for some common errors known to just
rollback the statement. Even then, such statements can come from
child atomic sections, so committing would be unsafe. Luckily, in
such cases, there will be a mismatch detected on endAtomic() or a
dangling section detected in close(), resulting in rollback.
Remove caching from DatabaseMyslBase::getServerVariableSettings
in case some SET query changes the values.
Bug: T189999
Change-Id: I532bc5201681a915d0c8aa7a3b1c143b040b142e
* Make onTransactionIdle() wait until any transaction round
is gone, even if there is no SQL transaction active. This
is what onTransactionPreCommitOrIdle() already does.
* Decouple "transaction round mode" (DBO_TRX) from whether a
round is active via a 'trxRoundId' LB info field. If rounds
are enabled, but not is started, then the transaction state
should be interpreted as "idle".
* Improve related documentation.
* Add more related unit tests.
Change-Id: I3ab18f577ec0375897fcb63f18f4ee2deeb436e9
Make sure any value from the last transaction is cleared and not used.
For sanity, make getLagFromPtHeartbeat() use this regardless of the age of
the transaction unless it returns null.
Change-Id: I52df6147f99736ad1a389ae70d347ae968e50c7f
* Filter out GTIDs with a domain that is not the one binlog
events would be written to if the Database handle was given
write queries. Likewise for the MariaDB server_id component.
* Also improve MySQL GTID support to better match that of MariaDB.
This covers position retrieval, replication waiting, and ranges
in GTIDs (which are almost always present).
* Make some MySQLMasterPos variables private by making use of
accesors instead.
* Store the gtids array keyed by domain ID for convenience.
* Clean up dynamic call to static method.
Change-Id: Ic6ab517bc8f200c968ff892ade69ad1b9394ab21
This takes the logic from wasErrorReissuable(), but puts it under
better name. The way that method was used, as well its comments,
were only about connection loss.
Make wasErrorReissuable() check if there was any error that
does not preclude the ability to retry. This matches the actual
name of the method.
Also improve some other related comments.
Change-Id: I68455d803afb2370897fecab0e79aadbb5d1a740
If the $var argument is provided, then it will make the resulting
count exclude rows where the value for that column is NULL.
Also add buildSelectSubquery() method and Subquery
wrapper class for use with select() for calculated tables.
Change-Id: I549d629af99afdf370602de095f7fba6d1546c37
An INSERT SELECT in MySQL/MariaDB is unsafe for replication if a column
is getting values from auto-increment, statement-based replication is in
use, and the default innodb_autoinc_lock_mode is set.
I9173f655 added checks to force non-native insertSelect for the
statement-based replication and innodb_autoinc_lock_mode != 2 case, but
determining whether a column is getting values from auto-increment is
too hard to do automatically there.
Instead, let's add a flag to let the caller hint that the query isn't
getting any auto-increment values. And use it in MysqlUpdater when
appropriate.
Bug: T160993
Change-Id: If70450a64aa3bcbf763c62838bb21306d124ae3d
Named locks are session-level constructs and this transaction agnostic.
Also make lockIsFree() a bit more consistent when the thread has the
lock itself.
Change-Id: Ief51196161bbc50c798740f3c738fd0e39880508
Done using the PhpStorm refactor->rename tool.
Also move "defaultBigSelects" declaration to DatabaseMysqlBase
as no other classes uses that.
Change-Id: I424a2d9815de3a5d4cca2522f3db23a5efe6b592
This regressed in 14ee3f2, trying to select the server config
"dbname" value as the database on connect. If that config is
bogus, then the connection attempt would fail. Instead, always
pass null to the driver's connection function if the DB domain
doesn't matter.
This affected WMF sites during lag checks, on account of the
"serverTemplate" value in $wgLBFactoryConf always using the local
wiki domain (whether the cluster had such a database or not).
Use strlen() as mysqli sees null and "" as the same for $dbname.
Bug: T186764
Change-Id: I6699d17c0125a08415046211fee7906bbaf2c366
The binlog file/pos where only being used in __toString() for the GTID
case. Make that method use the GTID set instead and avoid querying the
old-fashioned binlog fields all together in that case. The STATUS
queries involve some global lock contention.
Bug: T180918
Change-Id: I18123a702e4f554b87bf5f90017b248062e73049
Certain server configurations, including the current MariaDB defaults,
make INSERT SELECT unsafe for replication in MySQL/MariaDB. When the
server configuration is not known to be safe, force the use of the
non-native implementation. Note this only has effect in CLI mode, as
non-CLI mode already forces the non-native implementation since
I2dba6024.
Also, native INSERT SELECT won't be safe with any statement-based
replication method if the order of rows in the SELECT is not
deterministic. Add a warning to the method's documentation pointing this
out.
Change-Id: I9173f6559809bd01830bd0a9f443c7269cc58ce2
Whether the lock() acquisition failing is a huge problem depends
on what the caller is doing. Let the caller do any logging it needs.
Bug: T180793
Change-Id: I3d05138d312d8b973df153bb511e69619d663c9d
