Mutex order violation when wsrep bf thread kills a conflicting trx,
the stack is
wsrep_thd_LOCK()
wsrep_kill_victim()
lock_rec_other_has_conflicting()
lock_clust_rec_read_check_and_lock()
row_search_mvcc()
ha_innobase::index_read()
ha_innobase::rnd_pos()
handler::ha_rnd_pos()
handler::rnd_pos_by_record()
handler::ha_rnd_pos_by_record()
Rows_log_event::find_row()
Update_rows_log_event::do_exec_row()
Rows_log_event::do_apply_event()
Log_event::apply_event()
wsrep_apply_events()
and mutexes are taken in the order
lock_sys->mutex -> victim_trx->mutex -> victim_thread->LOCK_thd_data
When a normal KILL statement is executed, the stack is
innobase_kill_query()
kill_handlerton()
plugin_foreach_with_mask()
ha_kill_query()
THD::awake()
kill_one_thread()
and mutexes are
victim_thread->LOCK_thd_data -> lock_sys->mutex -> victim_trx->mutex
This patch is the plan D variant for fixing potetial mutex locking
order exercised by BF aborting and KILL command execution.
In this approach, KILL command is replicated as TOI operation.
This guarantees total isolation for the KILL command execution
in the first node: there is no concurrent replication applying
and no concurrent DDL executing. Therefore there is no risk of
BF aborting to happen in parallel with KILL command execution
either. Potential mutex deadlocks between the different mutex
access paths with KILL command execution and BF aborting cannot
therefore happen.
TOI replication is used, in this approach, purely as means
to provide isolated KILL command execution in the first node.
KILL command should not (and must not) be applied in secondary
nodes. In this patch, we make this sure by skipping KILL
execution in secondary nodes, in applying phase, where we
bail out if applier thread is trying to execute KILL command.
This is effective, but skipping the applying of KILL command
could happen much earlier as well.
This also fixed unprotected calls to wsrep_thd_abort
that will use wsrep_abort_transaction. This is fixed
by holding THD::LOCK_thd_data while we abort transaction.
Reviewed-by: Jan Lindström <jan.lindstrom@mariadb.com>
Problem:
========
This patch addresses two issues.
First, if a CHANGE MASTER command is issued and an error happens
while locating the replica’s relay logs, the logs can be put into an
invalid state where future updates fail and future CHANGE MASTER
calls crash the server. More specifically, right before a replica
purges the relay logs (part of the `CHANGE MASTER TO` logic), the
relay log is temporarily closed with state LOG_TO_BE_OPENED. If the
server errors in-between the temporary log closure and purge, i.e.
during the function find_log_pos, the log should be closed.
MDEV-25284 reveals the log is not properly closed.
Second, upon issuing a RESET SLAVE ALL command, a slave’s GTID
filters are not cleared (DO_DOMAIN_IDS, IGNORE_DOMIAN_IDS,
IGNORE_SERVER_IDS). MySQL had a similar bug report, Bug #18816897,
which fixed this issue to clear IGNORE_SERVER_IDS after issuing
RESET SLAVE ALL in version 5.7.
Solution:
=========
To fix the first problem, the CHANGE MASTER error handling logic was
extended to transition the relay log state to LOG_CLOSED from
LOG_TO_BE_OPENED.
To fix the second problem, the RESET SLAVE ALL logic is extended to
clear the domain_id filter and ignore_server_ids.
Reviewed By:
============
Andrei Elkin <andrei.elkin@mariadb.com>
Problem:
=======
There are two issues that are addressed in this patch:
1) SHOW BINARY LOGS uses caching to store the binary logs that exist
in the log directory; however, if new events are written to the logs,
the caching strategy is unaware. This is okay for users, as it is
okay for SHOW to return slightly old data. The test, however, can
result in inconsistent data. It runs two connections concurrently,
where one shows the logs, and the other adds a new file. The output
of SHOW BINARY LOGS then depends on when the cache is built, with
respect to the time that the second connection rotates the logs.
2) There is a race condition between RESET MASTER and SHOW BINARY
LOGS. More specifically, where they both need the binary log lock to
begin, SHOW BINARY LOGS only needs the lock to build its cache. If
RESET MASTER is issued after SHOW BINARY LOGS has built its cache and
before it has returned the results, the presented data may be
incorrect.
Solution:
========
1) As it is okay for users to see stale data, to make the test
consistent, use DEBUG_SYNC to force the race condition (problem 2) to
make SHOW BINARY LOGS build a cache before RESET MASTER is called.
Then, use additional logic from the next part of the solution to
rebuild the cache.
2) Use an Atomic_counter to keep track of the number of times RESET
MASTER has been called. If the value of the counter changes after
building the cache, the cache should be rebuilt and the analysis
should be restarted.
Reviewed By:
============
Andrei Elkin: <andrei.elkin@mariadb.com>
Problem:
========
180511 11:07:58 [ERROR] Slave I/O: Unexpected master's heartbeat data:
heartbeat is not compatible with local info;the event's data: log_file_name
mysql-bin.000009 log_pos 1054262041, Error_code: 1623
Analysis:
=========
In replication setup when master server doesn't have any events to send to
slave server it sends an 'Heartbeat_log_event'. This event carries the
current binary log filename and offset details. The offset values is stored
within 4 bytes of event header. When the size of binary log is higher than
UINT32_MAX the log_pos values will not fit in 4 bytes memory. It overflows
and hence slave stops with an error.
Fix:
===
Since we cannot extend the common_header of Log_event class, a greater than
4GB value of Log_event::log_pos is made to be transported with a HeartBeat
event's sub-header. Log_event::log_pos in such case is set to zero to
indicate that the 8 byte sub-header is allocated in the event.
In case of cross version replication following behaviour is expected
OLD - Server without fix
NEW - Server with fix
OLD<->NEW : works bidirectionally as long as the binlog offset is
(normally) within 4GB.
When log_pos > UINT32_MAX
OLD->NEW : The 'log_pos' is bound to overflow and NEW slave may report
an invalid event/incompatible heart beat event error.
NEW->OLD : Since patched server sets log_pos=0 on overflow, OLD slave will
report invalid event error.
rocksdb_checkpoint_request() should call FlushWAL(sync=true) (which does
write-out and sync), not just SyncWAL() (which just syncs without writing
out)
Followup: the test requires debug sync facility
(This is a backport to 10.5)
Problem:
========
Auto purge of relaylogs stops when relay-log-file is
'slave-relay-log.999999' and slave_parallel_threads is enabled.
Analysis:
=========
The problem is that in Relay_log_info::inc_group_relay_log_pos() function,
when two log names are compared via strcmp() function, it gives correct
result, when log name sequence numbers are of same digits(6 digits), But
when the number goes to 7 digits, a 999999 compares greater than
1000000, which is wrong, hence the bug.
Fix:
====
Extract the numeric extension part of the file name, convert it into
unsigned long and compare.
Thanks to David Zhao for the contribution.
Problem:
=======
SHOW BINLOG EVENTS FROM <"random"-pos> caused a variety of failures as
reported in MDEV-18046. They are fixed but that approach is not future-proof
as well as is not optimal to create extra check for being constructed event
parameters.
Analysis:
=========
"show binlog events from <pos>" code considers the user given position as a
valid event start position. The code starts reading data from this event start
position onwards and tries to map it to a set of known events. Each event has
a specific event structure and asserts have been added to ensure that, read
event data, satisfies the event specific requirements. When a random position
is supplied to "show binlog events command" the event structure specific
checks will fail and they result in assert.
For example: https://jira.mariadb.org/browse/MDEV-18046
In the bug description user executes CREATE TABLE/INSERT and ALTER SQL
commands.
When a crazy offset like "SHOW BINLOG EVENTS FROM 365" is provided code
assumes offset 365 as valid event begin and proceeds to EVENT_LEN_OFFSET reads
some random length and comes up with a crazy event which didn't exits in the
binary log. In this quoted example scenario, event read at offset 365 is
considered as "Update_rows_log_event", which is not present in binary log.
Since this is a random event its validation fails and code results in
assert/segmentation fault, as shown below.
mysqld: /data/src/10.4/sql/log_event.cc:10863: Rows_log_event::Rows_log_event(
const char*, uint, const Format_description_log_event*):
Assertion `var_header_len >= 2' failed.
181220 15:27:02 [ERROR] mysqld got signal 6 ;
#7 0x00007fa0d96abee2 in __assert_fail () from /lib/x86_64-linux-gnu/libc.so.6
#8 0x000055e744ef82de in Rows_log_event::Rows_log_event (this=0x7fa05800d390,
buf=0x7fa05800d080 "", event_len=254, description_event=0x7fa058006d60) at
/data/src/10.4/sql/log_event.cc:10863
#9 0x000055e744f00cf8 in Update_rows_log_event::Update_rows_log_event
Since we are reading random data repeating the same command SHOW BINLOG EVENTS
FROM 365 produces different types of crashes with different events. MDEV-18046
reported 10 such crashes.
In order to avoid such scenarios user provided starting offset needs to be
validated for its correctness. Best way of doing this is to make use of
checksums if they are available. MDEV-18046 fix introduced the checksum based
validation.
The issue still remains in cases where binlog checksums are disabled. Please
find the following bug reports.
MDEV-22473: binlog.binlog_show_binlog_event_random_pos failed in buildbot,
server crashed in read_log_event
MDEV-22455: Server crashes in Table_map_log_event,
binlog.binlog_invalid_read_in_rotate failed in buildbot
Fix:
====
When binlog checksum is disabled, perform scan(via reading event by event), to
validate the requested FROM <pos> offset. Starting from offset 4 read the
event_length of next_event in the binary log. Using the next_event length
advance current offset to point to next event. Repeat this process till the
current offset is less than or equal to crazy offset. If current offset is
higher than crazy offset provide appropriate invalid input offset error.
Problem:
=======
The "Start binlog_dump" message hasn't been updated to include the slave's
requested GTID position:
20:05:05 139836760311552 [Note] Start binlog_dump to slave_server(2), pos(, 4)
For diagnostic purposes, it would be helpful if the GTID position were
included.
Fix:
===
Imporve "Start binlog_dump" print message to include "using_gtid" and
"GTID position" requested by slave.
Ex:
[Note] Start binlog_dump to slave_server(2), pos(, 4), using_gtid(1),
gtid('1-1-201,2-2-100')
[Note] Start binlog_dump to slave_server(3), pos('mariadb-bin.004142',
507988273), using_gtid(0), gtid('')
Lifted long standing limitation to the XA of rolling it back at the
transaction's
connection close even if the XA is prepared.
Prepared XA-transaction is made to sustain connection close or server
restart.
The patch consists of
- binary logging extension to write prepared XA part of
transaction signified with
its XID in a new XA_prepare_log_event. The concusion part -
with Commit or Rollback decision - is logged separately as
Query_log_event.
That is in the binlog the XA consists of two separate group of
events.
That makes the whole XA possibly interweaving in binlog with
other XA:s or regular transaction but with no harm to
replication and data consistency.
Gtid_log_event receives two more flags to identify which of the
two XA phases of the transaction it represents. With either flag
set also XID info is added to the event.
When binlog is ON on the server XID::formatID is
constrained to 4 bytes.
- engines are made aware of the server policy to keep up user
prepared XA:s so they (Innodb, rocksdb) don't roll them back
anymore at their disconnect methods.
- slave applier is refined to cope with two phase logged XA:s
including parallel modes of execution.
This patch does not address crash-safe logging of the new events which
is being addressed by MDEV-21469.
CORNER CASES: read-only, pure myisam, binlog-*, @@skip_log_bin, etc
Are addressed along the following policies.
1. The read-only at reconnect marks XID to fail for future
completion with ER_XA_RBROLLBACK.
2. binlog-* filtered XA when it changes engine data is regarded as
loggable even when nothing got cached for binlog. An empty
XA-prepare group is recorded. Consequent Commit-or-Rollback
succeeds in the Engine(s) as well as recorded into binlog.
3. The same applies to the non-transactional engine XA.
4. @@skip_log_bin=OFF does not record anything at XA-prepare
(obviously), but the completion event is recorded into binlog to
admit inconsistency with slave.
The following actions are taken by the patch.
At XA-prepare:
when empty binlog cache - don't do anything to binlog if RO,
otherwise write empty XA_prepare (assert(binlog-filter case)).
At Disconnect:
when Prepared && RO (=> no binlogging was done)
set Xid_cache_element::error := ER_XA_RBROLLBACK
*keep* XID in the cache, and rollback the transaction.
At XA-"complete":
Discover the error, if any don't binlog the "complete",
return the error to the user.
Kudos
-----
Alexey Botchkov took to drive this work initially.
Sergei Golubchik, Sergei Petrunja, Marko Mäkelä provided a number of
good recommendations.
Sergei Voitovich made a magnificent review and improvements to the code.
They all deserve a bunch of thanks for making this work done!
The merge a8ed0f77a3734be15f95a67b5880ed96919e3236
was accidentally a null-merge.
MDEV-18046: Assortment of crashes, assertion failures and ASAN errors in mysql_show_binlog_events
MDEV-18046: Assortment of crashes, assertion failures and ASAN errors in mysql_show_binlog_events
Problem:
========
SHOW BINLOG EVENTS FROM <pos> reports following assert when ASAN is enabled.
uint32 binlog_get_uncompress_len(const char*):
Assertion `(buf[0] & 0xe0) == 0x80' failed
Fix:
===
**Part11: Converted debug assert to error handler code**
Problem:
========
SHOW BINLOG EVENTS FROM <pos> causes a variety of failures, some of which are
listed below. It is not a race condition issue, but there is some
non-determinism in it.
Analysis:
========
"show binlog events from <pos>" code considers the user given position as a
valid event start position. The code starts reading data from this event start
position onwards and tries to map it to a set of known events. Each event has
a specific event structure and asserts have been added to ensure that read
event data satisfies the event specific requirements. When a random position
is supplied to "show binlog events command" the event structure specific
checks will fail and they result in assert.
Fix:
====
The fix is split into different parts. Each part addresses either an ASAN
issue or an assert/crash.
**Part1: Checksum based position validation when checksum is enabled**
Using checksum validate the very first event read at the user specified
position. If there is a checksum mismatch report an appropriate error for the
invalid event.
