There is one part of the test case that needs to break
and re-establish the circular topology. For this the test
stops the slave threads on a couple of servers and restarts
them with START SLAVE. However, no check is done on the
status of the IO or SQL threads before proceeding with
the subsequent commands.
Because rpl_only_running_threads is set to 1 this can lead
to silently not syncing all slave threads as expected,
ultimately resulting in unexpected results (and consequently
on a failing test run).
We fix this by replacing the START SLAVE instructions with
calls to --source include/start_slave.inc, which will wait
for the slave threads to be running (show 'Yes' in
Slave_IO|SQL_Running fields of SHOW SLAVE STATUS) before
proceeding. Additionally, we change rpl_sync.inc to make the
IO thread report that it is running when its running status
is any other than 'No'.
In SBR, if a statement does not fail, it is always written to the binary
log, regardless if rows are changed or not. If there is a failure, a
statement is only written to the binary log if a non-transactional (.e.g.
MyIsam) engine is updated.
INSERT ON DUPLICATE KEY UPDATE and INSERT IGNORE were not following the
rule above and were not written to the binary log, if then engine was
Innodb.
There are two calls to read_log_event() on master in mysql_binlog_send().
Each call reads 19 bytes in this test case and the error of the second
read_log_event() is reported to the slave.
The second read_log_event() starts from position 94 (75 + 19) to 113
(75 + 19 + 19). Usually, there are two events in the binary log:
. 0 - 3 - Header
. 4 - 105 - Format Descriptor Event
. 106 - 304 - Query Event
and both reads fail because operations are reading from invalid positions
as expected.
However, mysql_binlog_send() does not use the same IO_CACHE that is used to
write into binary log (i.e. mysql_bin_log.log_file) for the hot binary log.
It opens the binary log file directly by calling open_binlog() and creates a
separated IO_CACHE. So there is a possibly that after a master has flushed
the binary log file, the content has been cached by the filesystem, and has
not updated the disk file. If this happens, then a slave will only see part
of the file, and thus the second read_log_event() will report event truncated
error.
To fix the problem, if the first read_log_event() has failed, we ensure that
the second one will try to read from the same position.
rpl_packet got a timeout failure sporadically on PB when stopping
slave. The real reason of this bug is that STOP SLAVE stopped
IO thread first and then stopped SQL thread. It was
possible that IO thread stopped after replicating part of a
transaction which SQL thread was executing. SQL thread would
be hung if the transaction could not be rolled back safely.
After this patch, STOP SLAVE will stop SQL thread first and then stop IO
thread, which guarantees that IO thread will fetch the reset of the
events of the transaction that SQL thread is executing, so that SQL
thread can finish the transaction if it cannot be rolled back safely.
Added below auxiliary files to make the test code neater.
restart_slave_sql.inc
rpl_connection_master.inc
rpl_connection_slave.inc
rpl_connection_slave1.inc
Problem: master executed a statement that would fail on slave
(namely, DROP USER 'create_rout_db'@'localhost').
Then the test did:
--let $rpl_only_running_threads= 1
--source include/rpl_reset.inc
rpl_reset.inc calls rpl_sync.inc, which first checks which of
the threads are running and then syncs those threads that are
running. If the SQL thread fails after the check, the sync will
fail. So there was a race in the test and it failed on some
slow hosts.
Fix: Don't replicate the failing statement.
Normally, auto_increment value is generated for the column by
inserting either NULL or 0 into it. NO_AUTO_VALUE_ON_ZERO
suppresses this behavior for 0 so that only NULL generates
the auto_increment value. This behavior is also followed by
a slave, specifically by the SQL Thread, when applying events
in the statement format from a master. However, when applying
events in the row format, the flag was ignored thus causing
an assertion failure:
"Assertion failed: next_insert_id == 0, file .\handler.cc"
In fact, we never need to generate a auto_increment value for
the column when applying events in row format on slave. So we
don't allow it to happen by using 'MODE_NO_AUTO_VALUE_ON_ZERO'.
Refactoring: Get rid of all the sql_mode checks to rows_log_event
when applying it for avoiding problems caused by the inconsistency
of the sql_mode on slave and master as the sql_mode is not set for
Rows_log_event.
Major replication test framework cleanup. This does the following:
- Ensure that all tests clean up the replication state when they
finish, by making check-testcase check the output of SHOW SLAVE STATUS.
This implies:
- Slave must not be running after test finished. This is good
because it removes the risk for sporadic errors in subsequent
tests when a test forgets to sync correctly.
- Slave SQL and IO errors must be cleared when test ends. This is
good because we will notice if a test gets an unexpected error in
the slave threads near the end.
- We no longer have to clean up before a test starts.
- Ensure that all tests that wait for an error in one of the slave
threads waits for a specific error. It is no longer possible to
source wait_for_slave_[sql|io]_to_stop.inc when there is an error
in one of the slave threads. This is good because:
- If a test expects an error but there is a bug that causes
another error to happen, or if it stops the slave thread without
an error, then we will notice.
- When developing tests, wait_for_*_to_[start|stop].inc will fail
immediately if there is an error in the relevant slave thread.
Before this patch, we had to wait for the timeout.
- Remove duplicated and repeated code for setting up unusual replication
topologies. Now, there is a single file that is capable of setting
up arbitrary topologies (include/rpl_init.inc, but
include/master-slave.inc is still available for the most common
topology). Tests can now end with include/rpl_end.inc, which will clean
up correctly no matter what topology is used. The topology can be
changed with include/rpl_change_topology.inc.
- Improved debug information when tests fail. This includes:
- debug info is printed on all servers configured by include/rpl_init.inc
- User can set $rpl_debug=1, which makes auxiliary replication files
print relevant debug info.
- Improved documentation for all auxiliary replication files. Now they
describe purpose, usage, parameters, and side effects.
- Many small code cleanups:
- Made have_innodb.inc output a sensible error message.
- Moved contents of rpl000017-slave.sh into rpl000017.test
- Added mysqltest variables that expose the current state of
disable_warnings/enable_warnings and friends.
- Too many to list here: see per-file comments for details.
Post-push fixes:
- fixed platform dependent result files
- appeasing valgrind warnings:
Fault injection was also uncovering a previously existing
potential mem leaks. For BUG#46166 testing purposes, fixed
by forcing handling the leak when injecting faults.
When a query fails with a different error on the slave,
the sql thread outputs a message (M) containing:
1. the error message format for the master error code
2. the master error code
3. the error message for the slave's error code
4. the slave error code
Given that the slave has no information on the error message
itself that the master outputs, it can only print its own
version of the message format (but stripped from the
additional data if the message format requires). This may
confuse users.
To fix this we augment the slave's message (M) to explicitly
state that the master's message is actually an error message
format, the one associated with the given master error code
and that the slave server knows about.
when generating new name.
If find_uniq_filename returns an error, then this error is not
being propagated upwards, and execution does not report error to
the user (although a entry in the error log is generated).
Additionally, some more errors were ignored in new_file_impl:
- when writing the rotate event
- when reopening the index and binary log file
This patch addresses this by propagating the error up in the
execution stack. Furthermore, when rotation of the binary log
fails, an incident event is written, because there may be a
chance that some changes for a given statement, were not properly
logged. For example, in SBR, LOAD DATA INFILE statement requires
more than one event to be logged, should rotation fail while
logging part of the LOAD DATA events, then the logged data would
become inconsistent with the data in the storage engine.
There were actually more problems in this area:
Slaves (if any) were unconditionally restarted, this appears unnecessary.
Sort criteria were suboptimal, included the test name.
Added logic to "reserve" a sequence of tests with same config for one thread
Got rid of sort_criteria hash, put it into the test case itself
Adds little sanity check that expected worker picks up test
Fixed some tests that may fail if starting on running server
Some of these fail only if *same* test is repeated.
Finally, special sorting of tests that do --force-restart
"Grantor" columns' data is lost when replicating mysql.tables_priv.
Slave SQL thread used its default user ''@'' as the grantor of GRANT|REVOKE
statements executing on it.
In this patch, current user is put in query log event for all GRANT and REVOKE
statement, SQL thread uses the user in query log event as grantor.
Rows events were applied wrongly on the temporary table with the same name.
But rows events are generated only for base tables. As temporary
table's data never be binlogged on row mode. Normally, base table of the
same name cannot be updated if a temporary table has the same name.
But there are two cases which can generate rows events on
the base table of same name.
Case1: 'CREATE TABLE ... SELECT' statement.
In mixed format, it will generate rows events if it is unsafe.
Case2: Drop a transactional temporary table in a transaction
(happens only on 5.5+).
BEGIN;
DROP TEMPORARY TABLE t1; # t1 is a InnoDB table
INSERT INTO t1 VALUES(rand()); # t1 is a MyISAM table
COMMIT;
'DROP TEMPORARY TABLE' will be put in the transaction cache and
binlogged after the rows events generated by the 'INSERT' statement.
After this patch, slave opens only base table when applying a rows event.
replication aborts
When recieving a 'SLAVE STOP' command, slave SQL thread will roll back the
transaction and stop immidiately if there is only transactional table updated,
even through 'CREATE|DROP TEMPOARY TABLE' statement are in it. But These
statements can never be rolled back. Because the temporary tables to the user
session mapping remain until 'RESET SLAVE', Therefore it will abort SQL thread
with an error that the table already exists or doesn't exist, when it restarts
and executes the whole transaction again.
After this patch, SQL thread always waits till the transaction ends and then stops,
if 'CREATE|DROP TEMPOARY TABLE' statement are in it.
After ALTER TABLE which changed only table's metadata, row-based
binlog sometimes got corrupted since the tablemap was unexpectedly
set to 0 for subsequent updates to the same table.
ALTER TABLE which changed only table's metadata always reset
table_map_id for the table share to 0. Despite the fact that
0 is a valid value for table_map_id, this step caused problems
as it could have created situation in which we had more than
one table share with table_map_id equal 0. If more than one
table with table_map_id are 0 were updated in the same statement,
updates to these different tables were written into the same
rows event. This caused slave server to crash.
This bug happens only on 5.1. It doesn't affect 5.5+.
This patch solves this problem by ensuring that ALTER TABLE
statements which change metadata only never reset table_map_id
to 0. To do this it changes reopen_table() to correctly use
refreshed table_map_id value instead of using the old one/
resetting it.
When slave executes a transaction bigger than slave's max_binlog_cache_size,
slave will crash. It is caused by the assert that server should only roll back
the statement but not the whole transaction if the error ER_TRANS_CACHE_FULL
happens. But slave sql thread always rollbacks the whole transaction when
an error happens.
Ather this patch, we always clear any error set in sql thread(it is different
from the error in 'SHOW SLAVE STATUS') and it is cleared before rolling back
the transaction.
'CREATE TABLE IF NOT EXISTS ... SELECT' behaviour
BUG#55474, BUG#55499, BUG#55598, BUG#55616 and BUG#55777 are fixed
in this patch too.
This is the 5.1 part.
It implements:
- if the table exists, binlog two events: CREATE TABLE IF NOT EXISTS
and INSERT ... SELECT
- Insert nothing and binlog nothing on master if the existing object
is a view. It only generates a warning that table already exists.
With statement- or mixed-mode logging, "LOAD DATA INFILE" queries
are written to the binlog using special types of log events.
When mysqlbinlog reads such events, it re-creates the file in a
temporary directory with a generated filename and outputs a
"LOAD DATA INFILE" query where the filename is replaced by the
generated file. The temporary file is not deleted by mysqlbinlog
after termination.
To fix the problem, in mixed mode we go to row-based. In SBR, we
document it to remind user the tmpfile is left in a temporary
directory.
A CREATE...SELECT that fails is written to the binary log if a non-transactional
statement is updated. If the logging format is ROW, the CREATE statement and the
changes are written to the binary log as distinct events and by consequence the
created table is not rolled back in the slave.
In this patch, we opted to let the slave goes out of sync by not writting to the
binary log the CREATE statement. We do this by simply reseting the binary log's
cache.
/*![:version:] Query Code */, where [:version:] is a sequence of 5
digits representing the mysql server version(e.g /*!50200 ... */),
is a special comment that the query in it can be executed on those
servers whose versions are larger than the version appearing in the
comment. It leads to a security issue when slave's version is larger
than master's. A malicious user can improve his privileges on slaves.
Because slave SQL thread is running with SUPER privileges, so it can
execute queries that he/she does not have privileges on master.
This bug is fixed with the logic below:
- To replace '!' with ' ' in the magic comments which are not applied on
master. So they become common comments and will not be applied on slave.
- Example:
'INSERT INTO t1 VALUES (1) /*!10000, (2)*/ /*!99999 ,(3)*/
will be binlogged as
'INSERT INTO t1 VALUES (1) /*!10000, (2)*/ /* 99999 ,(3)*/
Problem: when SHOW BINLOG EVENTS was issued, it increased the value of
@@session.max_allowed_packet. This allowed a non-root user to increase
the amount of memory used by her thread arbitrarily. Thus, it removes
the bound on the amount of system resources used by a client, so it
presents a security risk (DoS attack).
Fix: it is correct to increase the value of @@session.max_allowed_packet
while executing SHOW BINLOG EVENTS (see BUG 30435). However, the
increase should only be temporary. Thus, the fix is to restore the value
when SHOW BINLOG EVENTS ends.
The value of @@session.max_allowed_packet is also increased in
mysql_binlog_send (i.e., the binlog dump thread). It is not clear if this
can cause any trouble, since normally the client that issues
COM_BINLOG_DUMP will not issue any other commands that would be affected
by the increased value of @@session.max_allowed_packet. However, we
restore the value just in case.
DROP USER
RENAME USER CURRENT_USER() ...
GRANT ... TO CURRENT_USER()
REVOKE ... FROM CURRENT_USER()
ALTER DEFINER = CURRENT_USER() EVENTbut, When these statements are binlogged, CURRENT_USER() just is binlogged
as 'CURRENT_USER()', it is not expanded to the real user name. When slave
executes the log event, 'CURRENT_USER()' is expand to the user of slave
SQL thread, but SQL thread's user name always NULL. This breaks the replication.
After this patch, session's user will be written into query log events
if these statements call CURREN_USER() or 'ALTER EVENT' does not assign a definer.
DROP USER
RENAME USER CURRENT_USER() ...
GRANT ... TO CURRENT_USER()
REVOKE ... FROM CURRENT_USER()
ALTER DEFINER = CURRENT_USER() EVENTbut, When these statements are binlogged, CURRENT_USER() just is binlogged
as 'CURRENT_USER()', it is not expanded to the real user name. When slave
executes the log event, 'CURRENT_USER()' is expand to the user of slave
SQL thread, but SQL thread's user name always NULL. This breaks the replication.
After this patch, session's user will be written into query log events
if these statements call CURREN_USER() or 'ALTER EVENT' does not assign a definer.
When using Unique Keys with nullable parts in RBR, the slave can
choose the wrong row to update. This happens because a table with
an unique key containing nullable parts cannot strictly guarantee
uniqueness. As stated in the manual, for all engines, a UNIQUE
index allows multiple NULL values for columns that can contain
NULL.
We fix this at the slave by extending the checks before assuming
that the row found through an unique index is is the correct
one. This means that when a record (R) is fetched from the storage
engine and a key that is not primary (K) is used, the server does
the following:
- If K is unique and has no nullable parts, it returns R;
- Otherwise, if any field in the before image that is part of K
is null do an index scan;
- If there is no NULL field in the BI part of K, then return R.
A side change: renamed the existing test case file and added a
test case covering the changes in this patch.
Add code to waiting for a set of errors.
Add code to waiting for an error instead of waiting for io thread to stop, as
after 'START SLAVE', the status of io thread is still not running.
But it doesn't mean slave io thread encounters an error.
