But there is no Last_IO_Error reported.
On the master, if a binary log event is larger than max_allowed_packet,
ER_MASTER_FATAL_ERROR_READING_BINLOG and the specific reason of this error is
sent to a slave when it requests a dump from the master, thus leading
the I/O thread to stop.
On a slave, the I/O thread stops when receiving a packet larger than max_allowed_packet.
In both cases, however, there was no Last_IO_Error reported.
This patch adds code to report the Last_IO_Error and exact reason before stopping the
I/O thread and also reports the case the out memory pops up while
handling packets from the master.
If the SQL Thread fails to execute an event due to a temporary error (e.g.
ER_LOCK_DEADLOCK) and the option "--slave_transaction_retries" is set the SQL
Thread should not be aborted and the transaction should be restarted from the
beginning and re-executed.
Unfortunately, a wrong interpretation of the THD::is_fatal_error was preventing
this behavior. In a nutshell, "this variable is set to TRUE if an execution of a
compound statement cannot continue. In particular, it is used to disable access
to the CONTINUE or EXIT handlers of stored routines. So even temporary errors
may have this variable set.
To fix the bug, we have done what follows:
DBUG_ENTER("has_temporary_error");
- if (thd->is_fatal_error)
- DBUG_RETURN(0);
-
DBUG_EXECUTE_IF("all_errors_are_temporary_errors",
if (thd->main_da.is_error())
{
procedures causes crashes!
The problem of that bugreport was mostly fixed by the
patch for bug 38691.
However, attached test case focused on another crash or
valgrind warning problem: SHOW PROCESSLIST query accesses
freed memory of SP instruction that run in a parallel
connection.
Changes of thd->query/thd->query_length in dangerous
places have been guarded with the per-thread
LOCK_thd_data mutex (the THD::LOCK_delete mutex has been
renamed to THD::LOCK_thd_data).
The "get_master_version_and_clock(...)" function in sql/slave.cc ignores
error and passes directly when queries fail, or queries succeed
but the result retrieved is empty.
The "get_master_version_and_clock(...)" function should try to reconnect master
if queries fail because of transient network problems, and fail otherwise.
The I/O thread should print a warning if the some system variables do not
exist on master (very old master)
Bug#45243: crash on win in sql thread clear_tables_to_lock() -> free()
Bug#45242: crash on win in mysql_close() -> free()
Bug#45238: rpl_slave_skip, rpl_change_master failed (lost connection) for STOP SLAVE
Bug#46030: rpl_truncate_3innodb causes server crash on windows
Bug#46014: rpl_stm_reset_slave crashes the server sporadically in pb2
When killing a user session on the server, it's necessary to
interrupt (notify) the thread associated with the session that
the connection is being killed so that the thread is woken up
if waiting for I/O. On a few platforms (Mac, Windows and HP-UX)
where the SIGNAL_WITH_VIO_CLOSE flag is defined, this interruption
procedure is to asynchronously close the underlying socket of
the connection.
In order to enable this schema, each connection serving thread
registers its VIO (I/O interface) so that other threads can
access it and close the connection. But only the owner thread of
the VIO might delete it as to guarantee that other threads won't
see freed memory (the thread unregisters the VIO before deleting
it). A side note: closing the socket introduces a harmless race
that might cause a thread attempt to read from a closed socket,
but this is deemed acceptable.
The problem is that this infrastructure was meant to only be used
by server threads, but the slave I/O thread was registering the
VIO of a mysql handle (a client API structure that represents a
connection to another server instance) as a active connection of
the thread. But under some circumstances such as network failures,
the client API might destroy the VIO associated with a handle at
will, yet the VIO wouldn't be properly unregistered. This could
lead to accesses to freed data if a thread attempted to kill a
slave I/O thread whose connection was already broken.
There was a attempt to work around this by checking whether
the socket was being interrupted, but this hack didn't work as
intended due to the aforementioned race -- attempting to read
from the socket would yield a "bad file descriptor" error.
The solution is to add a hook to the client API that is called
from the client code before the VIO of a handle is deleted.
This hook allows the slave I/O thread to detach the active vio
so it does not point to freed memory.
timeout
In STMT and MIXED modes, a statement that changes both non-transactional and
transactional tables must be written to the binary log whenever there are
changes to non-transactional tables. This means that the statement gets into the
binary log even when the changes to the transactional tables fail. In particular
, in the presence of a failure such statement is annotated with the error number
and wrapped in a begin/rollback. On the slave, while applying the statement, it
is expected the same failure and the rollback prevents the transactional changes
to be persisted.
Unfortunately, statements that fail due to concurrency issues (e.g. deadlocks,
timeouts) are logged in the same way causing the slave to stop as the statements
are applied sequentially by the SQL Thread. To fix this bug, we automatically
ignore concurrency failures on the slave. Specifically, the following failures
are ignored: ER_LOCK_WAIT_TIMEOUT, ER_LOCK_DEADLOCK and ER_XA_RBDEADLOCK.
The reason for the crash was rotate_relay_log (mi=0x0) did not verify
the passed value of active_mi. There are more cases where active_mi
is supposed to be non-zero e.g change_master(), stop_slave(), and it's
reasonable to protect from a similar crash all of them with common
fixes.
Fixed with spliting end_slave() in slave threads release and slave
data clean-up parts (a new close_active_mi()). The new function is
invoked at the very end of close_connections() so that all users of
active_mi are proven to have left.
with gcc 4.3.2
Compiling MySQL with gcc 4.3.2 and later produces a number of
warnings, many of which are new with the recent compiler
versions.
This bug will be resolved in more than one patch to limit the
size of changesets. This is the first patch, fixing a number
of the warnings, predominantly "suggest using parentheses
around && in ||", and empty for and while bodies.
with gcc 4.3.2
Compiling MySQL with gcc 4.3.2 and later produces a number of
warnings, many of which are new with the recent compiler
versions.
This bug will be resolved in more than one patch to limit the
size of changesets. This is the first patch, fixing a number
of the warnings, predominantly "suggest using parentheses
around && in ||", and empty for and while bodies.
The server was not cleaning the last IO error and error number when
resetting slave.
This patch addresses this issue by backporting into 5.1 part of the
patch in BUG 34654. A fix for this issue had already been pushed into
6.0 as part of the aforementioned bug, however the patch also included
some refactoring. The fix for 5.1 does not take into account the
refactoring part.
stop/start slave
When stopping and restarting the slave while it is replicating
temporary tables, the server would crash or raise an assertion
failure. This was due to the fact that although temporary tables are
saved between slave threads restart, the reference to the thread in
use (table->in_use) was not being properly updated when the restart
happened (it would still reference the old/invalid thread instead of
the new one).
This patch addresses this issue by resetting the reference to the new
slave thread on slave thread restart.
165 changesets with 23 conflicts:
Text conflict in mysql-test/r/lock_multi.result
Text conflict in mysql-test/t/lock_multi.test
Text conflict in mysql-test/t/mysqldump.test
Text conflict in sql/item_strfunc.cc
Text conflict in sql/log.cc
Text conflict in sql/log_event.cc
Text conflict in sql/parse_file.cc
Text conflict in sql/slave.cc
Text conflict in sql/sp.cc
Text conflict in sql/sp_head.cc
Text conflict in sql/sql_acl.cc
Text conflict in sql/sql_base.cc
Text conflict in sql/sql_class.cc
Text conflict in sql/sql_crypt.cc
Text conflict in sql/sql_db.cc
Text conflict in sql/sql_lex.cc
Text conflict in sql/sql_parse.cc
Text conflict in sql/sql_select.cc
Text conflict in sql/sql_table.cc
Text conflict in sql/sql_view.cc
Text conflict in storage/innobase/handler/ha_innodb.cc
Text conflict in storage/myisam/mi_packrec.c
Text conflict in tests/mysql_client_test.c
Updates to Innobase, taken from main 5.1:
bzr: ERROR: Some change isn't sane:
File mysql-test/r/innodb-semi-consistent.result is owned by Innobase and should not be updated.
File mysql-test/t/innodb-semi-consistent.test is owned by Innobase and should not be updated.
File storage/innobase/handler/ha_innodb.cc is owned by Innobase and should not be updated.
File storage/innobase/ibuf/ibuf0ibuf.c is owned by Innobase and should not be updated.
File storage/innobase/include/row0mysql.h is owned by Innobase and should not be updated.
File storage/innobase/include/srv0srv.h is owned by Innobase and should not be updated.
File storage/innobase/include/trx0trx.h is owned by Innobase and should not be updated.
File storage/innobase/include/trx0trx.ic is owned by Innobase and should not be updated.
File storage/innobase/lock/lock0lock.c is owned by Innobase and should not be updated.
File storage/innobase/page/page0cur.c is owned by Innobase and should not be updated.
File storage/innobase/row/row0mysql.c is owned by Innobase and should not be updated.
File storage/innobase/row/row0sel.c is owned by Innobase and should not be updated.
File storage/innobase/srv/srv0srv.c is owned by Innobase and should not be updated.
File storage/innobase/trx/trx0trx.c is owned by Innobase and should not be updated.
(Set env var 'ALLOW_UPDATE_INNOBASE_OWNED' to override.)
The issue of the current bug is unguarded access to mi->slave_running
by the shutdown thread calling end_slave() that is bug#29968
(alas happened not to be cross-linked with the current bug)
Fixed:
with removing the unguarded read of the running status
and perform reading it in terminate_slave_thread()
at time run_lock is taken (mostly bug#29968 backporting, still with some
improvements over that patch - see the error reporting from
terminate_slave_thread()).
Issue of bug#38716 is fixed here for 5.0 branch as well.
Note:
There has been a separate artifact identified -
a race condition between init_slave() and end_slave() -
reported as Bug#44467.
In order to define the --slave-load-tmpdir, the init_relay_log_file()
was calling fn_format(MY_PACK_FILENAME) which internally was indirectly
calling strmov_overlapp() (through pack_dirname) and the following
warning message was being printed out while running in Valgrind:
"source and destination overlap in strcpy".
We fixed the issue by removing the flag MY_PACK_FILENAME as it was not
necessary. In a nutshell, with this flag the function fn_format() tried
to replace a directory by either "~", "." or "..". However, we wanted
exactly to remove such strings.
In this patch, we also refactored the functions init_relay_log_file()
and check_temp_dir(). The former was refactored to call the fn_format()
with the flag MY_SAFE_PATH along with the MY_RETURN_REAL_PATH, in order
to avoid issues with long directories and return an absolute path,
respectively. The flag MY_SAFE_UNPACK_FILENAME was removed too as it was
responsible for removing "~", "." or ".." only from the file parameter
and we wanted to remove such strings from the directory parameter in
the fn_format(). This result is stored in an rli variable, which is then
processed by the other function in order to verify if the directory exists
and if we are able to create files in it.
We didn't expect "error: no error", although this is
in fact a legitimate state (if something is erroneous
on the master, but not on the slave, e.g. INSERT fails
on master due to UNIQUE constraint which does not exist
on slave).
We now list the ignore for "0: no error" the same way
as any other ignore; moreover, if no or an empty
--slave-skip-errors is passed at start-up, we show
"OFF" instead of empty list, as intended. (The code
for that was there, but was only run for the empty-
argument case, even if it subsequently tested for
both conditions.)
RBR was not considering the option --slave-skip-errors.
To fix the problem, we are reporting the ignored ERROR(s) as warnings thus avoiding
stopping the SQL Thread. Besides, it fixes the output of "SHOW VARIABLES LIKE
'slave_skip_errors'" which was showing nothing when the value "all" was assigned
to --slave-skip-errors.
@sql/log_event.cc
skipped rbr errors when the option skip-slave-errors is set.
@sql/slave.cc
fixed the output of for SHOW VARIABLES LIKE 'slave_skip_errors'"
@test-cases
fixed the output of rpl.rpl_idempotency
updated the test case rpl_skip_error
Bug#319 if while a non-transactional slave is replicating a transaction possible problem
It is impossible to roll back a mixed engines transaction when one of the engine is
non-transaction. In replication that fact is crucial because the slave can not safely
re-apply a transction that was interrupted with STOP SLAVE.
Fixed with making STOP SLAVE not be effective immediately in the case the current
group of replication events has modified a non-transaction table. In order for slave to leave
either the group needs finishing or the user issues KILL QUERY|CONNECTION slave_thread_id.
Compiling with debug and assigning an invalid directory to --slave-load-tmpdir
was crashing the slave due to the following assertion DBUG_ASSERT(! is_set() ||
can_overwrite_status). This assertion assumes that a thread can change its
state once (i.e. ok,error, etc) before aborting, cleaning/resuming or completing
its execution unless the overwrite flag (i.e. can_overwrite_status) is true.
The Append_block_log_event::do_apply_event which is responsible for creating
temporary file(s) was not cleaning the thread state. Thus a failure while
trying to create a file in an invalid temporary directory was causing the crash.
To fix the problem we check if the temporary directory is valid before starting
the SQL Thread and reset the thread state before creating a file in
Append_block_log_event::do_apply_event.
Some errors that cause the slave SQL thread to stop are not shown in the
Slave_SQL_Error column of "SHOW SLAVE STATUS". Instead, the error is only
in the server's error log.
That makes it difficult to analyze the error for the user. One example of an error
that stops the slave but is not shown by "SHOW SLAVE STATUS" is when @@global.init_slave
is set incorrectly (e.g., it contains something that is not valid SQL).
Three failures were not correctly reported:
1 - Failures during slave thread initialization
2 - Failures while initializing the relay log position right after
starting the slave thread.
3 - Failures while processing queries passed through the init_slave
option.
This patch fixes the issues by reporting the errors through relay-info->report.
- Remove bothersome warning messages. This change focuses on the warnings
that are covered by the ignore file: support-files/compiler_warnings.supp.
- Strings are guaranteed to be max uint in length
Several system variables did not behave like system variables should do.
When trying to SET them or use them in SELECT, they were reported as
"unknown system variable". But they appeared in SHOW VARIABLES.
This has been fixed by removing the "fixed_vars" array of variables
and integrating the variables into the normal system variables chain.
All of these variables do now behave as read-only global-only
variables. Trying to SET them tells they are read-only, trying to
SELECT the session value tells they are global only. Selecting the
global value works. It delivers the same value as SHOW VARIABLES.
In order to handle CHAR() fields, 8 bits were reserved for
the size of the CHAR field. However, instead of denoting the
number of characters in the field, field_length was used which
denotes the number of bytes in the field.
Since UTF-8 fields can have three bytes per character (and
has been extended to have four bytes per character in 6.0),
an extra two bits have been encoded in the field metadata
work for fields of type Field_string (i.e., CHAR fields).
Since the metadata word is filled, the extra bits have been
encoded in the upper 4 bits of the real type (the most
significant byte of the metadata word) by computing the
bitwise xor of the extra two bits. Since the upper 4 bits
of the real type always is 1111 for Field_string, this
means that for fields of length <256, the encoding is
identical to the encoding used in pre-5.1.26 servers, but
for lengths of 256 or more, an unrecognized type is formed,
causing an old slave (that does not handle lengths of 256
or more) to stop.
The crash appeared to be a result of allocating an instance of Discrete_interval
automatically that that was referred in out-of-declaration scope.
Fixed with correcting backing up and restoring scheme of
auto_inc_intervals_forced, introduced by bug#33029, by means of shallow copying;
added simulation code that forces executing those fixes of the former bug that
targeted at master-and-slave having incompatible bug#33029-prone versions.
