The issue is caused by 59a0236da4 commit.
The initial intention of the commit was to speed up
"mariabackup --prepare".
The call stack of binlog position reading is the following:
▾ trx_rseg_mem_restore
▾ trx_rseg_array_init
▾ trx_lists_init_at_db_start
▸ srv_start
Both trx_lists_init_at_db_start() and trx_rseg_mem_restore() contain
special cases for srv_operation == SRV_OPERATION_RESTORE condition, and
on this condition only rseg headers are read to parse binlog position.
Performance impact is not so big.
The solution is to revert 59a0236da4.
mariabackup does not load dictionary during backup, but it loads
tablespaces, that is why fil_system.max_assigned_id is not set with
dict_check_tablespaces_and_store_max_id(). There is no sense to issue the
warning during backup.
When the log is stored in persistent memory, log_sys.buf[] is
a ring buffer that directly maps to the circular ib_logfile0 file.
There were several errors that could occur in the special case
when a log record ends exactly at the end of the log file and the
next record would start at log_sys.buf[log_sys.START_OFFSET].
mariabackup.huge_lsn,strict_full_crc32: Write the first record
at the very end of the circular file, to reproduce the failure
scenarios.
recv_sys_t::parse(): On PMEM, wrap the end offset of the record
from log_sys.file_size to log_sys.START_OFFSET if needed.
Otherwise, both InnoDB recovery and mariadb-backup would try
to parse the next record from an invalid address.
filename_to_spacename(): Remove an assumption about the format
of file names. While the server currently writes file names like
./databasename/tablename.ibd we might want to stop writing the
redundant ./ prefix in the future. The test mariabackup.huge_lsn
is generating such file names.
xtrabackup_copy_logfile(): Correctly copy a record that ends at
the very end of the log_sys.buf[].
The errors in mariadb-backup were reproduced with the test
mariabackup.huge_lsn,strict_full_crc32 and an additional patch
to use the start checkpoint of the test:
diff --git a/storage/innobase/log/log0recv.cc b/storage/innobase/log/log0recv.cc
index 27dce5fa17d..e17a1692d6f 100644
--- a/storage/innobase/log/log0recv.cc
+++ b/storage/innobase/log/log0recv.cc
@@ -1796,7 +1796,8 @@ dberr_t recv_sys_t::find_checkpoint()
continue;
}
- if (checkpoint_lsn >= log_sys.next_checkpoint_lsn)
+ if (checkpoint_lsn >= log_sys.next_checkpoint_lsn &&
+ checkpoint_lsn != 0x1000fffffe10)
{
log_sys.next_checkpoint_lsn= checkpoint_lsn;
log_sys.next_checkpoint_no= field == log_t::CHECKPOINT_1;
A prominent bottleneck in mtr_t::commit() is log_sys.mutex between
log_sys.append_prepare() and log_close().
User-visible change: The minimum innodb_log_file_size will be
increased from 1MiB to 4MiB so that some conditions can be
trivially satisfied.
log_sys.latch (log_latch): Replaces log_sys.mutex and
log_sys.flush_order_mutex. Copying mtr_t::m_log to
log_sys.buf is protected by a shared log_sys.latch.
Writes from log_sys.buf to the file system will be protected
by an exclusive log_sys.latch.
log_sys.lsn_lock: Protects the allocation of log buffer
in log_sys.append_prepare().
sspin_lock: A simple spin lock, for log_sys.lsn_lock.
Thanks to Vladislav Vaintroub for suggesting this idea, and for
reviewing these changes.
mariadb-backup: Replace some use of log_sys.mutex with recv_sys.mutex.
buf_pool_t::insert_into_flush_list(): Implement sorting of flush_list
because ordering is otherwise no longer guaranteed. Ordering by LSN
is needed for the proper operation of redo log checkpoints.
log_sys.append_prepare(): Advance log_sys.lsn and log_sys.buf_free by
the length, and return the old values. Also increment write_to_buf,
which was previously done in log_close().
mtr_t::finish_write(): Obtain the buffer pointer from
log_sys.append_prepare().
log_sys.buf_free: Make the field Atomic_relaxed,
to simplify log_flush_margin(). Use only loads and stores
to avoid costly read-modify-write atomic operations.
buf_pool.flush_list_requests: Replaces
export_vars.innodb_buffer_pool_write_requests
and srv_stats.buf_pool_write_requests.
Protected by buf_pool.flush_list_mutex.
buf_pool_t::insert_into_flush_list(): Do not invoke page_cleaner_wakeup().
Let the caller do that after a batch of calls.
recv_recover_page(): Invoke a minimal part of
buf_pool.insert_into_flush_list().
ReleaseBlocks::modified: A number of pages added to buf_pool.flush_list.
ReleaseBlocks::operator(): Merge buf_flush_note_modification() here.
log_t::set_capacity(): Renamed from log_set_capacity().
- Store the deferred tablespace name while loading the tablespace
for backup process.
- Mariabackup stores the list of space ids which has page0 INIT_PAGE
records. backup_first_page_op() and first_page_init() was introduced
to track the page0 INIT_PAGE records.
- backup_file_op() and log_file_op() was changed to handle
FILE_MODIFY redo log records. It is used to identify the
deferred tablespace space id.
- Whenever file operation redo log was processed by backup,
backup_file_op() should check whether the space name exist
in deferred tablespace. If it is then it needs to store the
space id, name when FILE_MODIFY, FILE_RENAME redo log processed
and it should delete the tablespace name from defer list in other
cases.
- backup_fix_ddl() should check whether deferred tablespace has
any page0 init records. If it is then consider the tablespace
as newly created tablespace. If not then backup should try
to reload the tablespace with SRV_BACKUP_NO_DEFER mode to
avoid the deferring of tablespace.
The InnoDB redo log used to be formatted in blocks of 512 bytes.
The log blocks were encrypted and the checksum was calculated while
holding log_sys.mutex, creating a serious scalability bottleneck.
We remove the fixed-size redo log block structure altogether and
essentially turn every mini-transaction into a log block of its own.
This allows encryption and checksum calculations to be performed
on local mtr_t::m_log buffers, before acquiring log_sys.mutex.
The mutex only protects a memcpy() of the data to the shared
log_sys.buf, as well as the padding of the log, in case the
to-be-written part of the log would not end in a block boundary of
the underlying storage. For now, the "padding" consists of writing
a single NUL byte, to allow recovery and mariadb-backup to detect
the end of the circular log faster.
Like the previous implementation, we will overwrite the last log block
over and over again, until it has been completely filled. It would be
possible to write only up to the last completed block (if no more
recent write was requested), or to write dummy FILE_CHECKPOINT records
to fill the incomplete block, by invoking the currently disabled
function log_pad(). This would require adjustments to some logic around
log checkpoints, page flushing, and shutdown.
An upgrade after a crash of any previous version is not supported.
Logically empty log files from a previous version will be upgraded.
An attempt to start up InnoDB without a valid ib_logfile0 will be
refused. Previously, the redo log used to be created automatically
if it was missing. Only with with innodb_force_recovery=6, it is
possible to start InnoDB in read-only mode even if the log file
does not exist. This allows the contents of a possibly corrupted
database to be dumped.
Because a prepared backup from an earlier version of mariadb-backup
will create a 0-sized log file, we will allow an upgrade from such
log files, provided that the FIL_PAGE_FILE_FLUSH_LSN in the system
tablespace looks valid.
The 512-byte log checkpoint blocks at 0x200 and 0x600 will be replaced
with 64-byte log checkpoint blocks at 0x1000 and 0x2000.
The start of log records will move from 0x800 to 0x3000. This allows us
to use 4096-byte aligned blocks for all I/O in a future revision.
We extend the MDEV-12353 redo log record format as follows.
(1) Empty mini-transactions or extra NUL bytes will not be allowed.
(2) The end-of-minitransaction marker (a NUL byte) will be replaced
with a 1-bit sequence number, which will be toggled each time when the
circular log file wraps back to the beginning.
(3) After the sequence bit, a CRC-32C checksum of all data
(excluding the sequence bit) will written.
(4) If the log is encrypted, 8 bytes will be written before
the checksum and included in it. This is part of the
initialization vector (IV) of encrypted log data.
(5) File names, page numbers, and checkpoint information will not be
encrypted. Only the payload bytes of page-level log will be encrypted.
The tablespace ID and page number will form part of the IV.
(6) For padding, arbitrary-length FILE_CHECKPOINT records may be written,
with all-zero payload, and with the normal end marker and checksum.
The minimum size is 7 bytes, or 7+8 with innodb_encrypt_log=ON.
In mariadb-backup and in Galera snapshot transfer (SST) scripts, we will
no longer remove ib_logfile0 or create an empty ib_logfile0. Server startup
will require a valid log file. When resizing the log, we will create
a logically empty ib_logfile101 at the current LSN and use an atomic rename
to replace ib_logfile0 with it. See the test innodb.log_file_size.
Because there is no mandatory padding in the log file, we are able
to create a dummy log file as of an arbitrary log sequence number.
See the test mariabackup.huge_lsn.
The parameter innodb_log_write_ahead_size and the
INFORMATION_SCHEMA.INNODB_METRICS counter log_padded will be removed.
The minimum value of innodb_log_buffer_size will be increased to 2MiB
(because log_sys.buf will replace recv_sys.buf) and the increment
adjusted to 4096 bytes (the maximum log block size).
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be removed:
os_log_fsyncs
os_log_pending_fsyncs
log_pending_log_flushes
log_pending_checkpoint_writes
The following status variables will be removed:
Innodb_os_log_fsyncs (this is included in Innodb_data_fsyncs)
Innodb_os_log_pending_fsyncs (this was limited to at most 1 by design)
log_sys.get_block_size(): Return the physical block size of the log file.
This is only implemented on Linux and Microsoft Windows for now, and for
the power-of-2 block sizes between 64 and 4096 bytes (the minimum and
maximum size of a checkpoint block). If the block size is anything else,
the traditional 512-byte size will be used via normal file system
buffering.
If the file system buffers can be bypassed, a message like the following
will be issued:
InnoDB: File system buffers for log disabled (block size=512 bytes)
InnoDB: File system buffers for log disabled (block size=4096 bytes)
This has been tested on Linux and Microsoft Windows with both sizes.
On Linux, only enable O_DIRECT on the log for innodb_flush_method=O_DSYNC.
Tests in 3 different environments where the log is stored in a device
with a physical block size of 512 bytes are yielding better throughput
without O_DIRECT. This could be due to the fact that in the event the
last log block is being overwritten (if multiple transactions would
become durable at the same time, and each of will write a small
number of bytes to the last log block), it should be faster to re-copy
data from log_sys.buf or log_sys.flush_buf to the kernel buffer,
to be finally written at fdatasync() time.
The parameter innodb_flush_method=O_DSYNC will imply O_DIRECT for
data files. This option will enable O_DIRECT on the log file on Linux.
It may be unsafe to use when the storage device does not support
FUA (Force Unit Access) mode.
When the server is compiled WITH_PMEM=ON, we will use memory-mapped
I/O for the log file if the log resides on a "mount -o dax" device.
We will identify PMEM in a start-up message:
InnoDB: log sequence number 0 (memory-mapped); transaction id 3
On Linux, we will also invoke mmap() on any ib_logfile0 that resides
in /dev/shm, effectively treating the log file as persistent memory.
This should speed up "./mtr --mem" and increase the test coverage of
PMEM on non-PMEM hardware. It also allows users to estimate how much
the performance would be improved by installing persistent memory.
On other tmpfs file systems such as /run, we will not use mmap().
mariadb-backup: Eliminated several variables. We will refer
directly to recv_sys and log_sys.
backup_wait_for_lsn(): Detect non-progress of
xtrabackup_copy_logfile(). In this new log format with
arbitrary-sized blocks, we can only detect log file overrun
indirectly, by observing that the scanned log sequence number
is not advancing.
xtrabackup_copy_logfile(): On PMEM, do not modify the sequence bit,
because we are not allowed to modify the server's log file, and our
memory mapping is read-only.
trx_flush_log_if_needed_low(): Do not use the callback on pmem.
Using neither flush_lock nor write_lock around PMEM writes seems
to yield the best performance. The pmem_persist() calls may
still be somewhat slower than the pwrite() and fdatasync() based
interface (PMEM mounted without -o dax).
recv_sys_t::buf: Remove. We will use log_sys.buf for parsing.
recv_sys_t::MTR_SIZE_MAX: Replaces RECV_SCAN_SIZE.
recv_sys_t::file_checkpoint: Renamed from mlog_checkpoint_lsn.
recv_sys_t, log_sys_t: Removed many data members.
recv_sys.lsn: Renamed from recv_sys.recovered_lsn.
recv_sys.offset: Renamed from recv_sys.recovered_offset.
log_sys.buf_size: Replaces srv_log_buffer_size.
recv_buf: A smart pointer that wraps log_sys.buf[recv_sys.offset]
when the buffer is being allocated from the memory heap.
recv_ring: A smart pointer that wraps a circular log_sys.buf[] that is
backed by ib_logfile0. The pointer will wrap from recv_sys.len
(log_sys.file_size) to log_sys.START_OFFSET. For the record that
wraps around, we may copy file name or record payload data to
the auxiliary buffer decrypt_buf in order to have a contiguous
block of memory. The maximum size of a record is less than
innodb_page_size bytes.
recv_sys_t::parse(): Take the smart pointer as a template parameter.
Do not temporarily add a trailing NUL byte to FILE_ records, because
we are not supposed to modify the memory-mapped log file. (It is
attached in read-write mode already during recovery.)
recv_sys_t::parse_mtr(): Wrapper for recv_sys_t::parse().
recv_sys_t::parse_pmem(): Like parse_mtr(), but if PREMATURE_EOF would be
returned on PMEM, use recv_ring to wrap around the buffer to the start.
mtr_t::finish_write(), log_close(): Do not enforce log_sys.max_buf_free
on PMEM, because it has no meaning on the mmap-based log.
log_sys.write_to_buf: Count writes to log_sys.buf. Replaces
srv_stats.log_write_requests and export_vars.innodb_log_write_requests.
Protected by log_sys.mutex. Updated consistently in log_close().
Previously, mtr_t::commit() conditionally updated the count,
which was inconsistent.
log_sys.write_to_log: Count swaps of log_sys.buf and log_sys.flush_buf,
for writing to log_sys.log (the ib_logfile0). Replaces
srv_stats.log_writes and export_vars.innodb_log_writes.
Protected by log_sys.mutex.
log_sys.waits: Count waits in append_prepare(). Replaces
srv_stats.log_waits and export_vars.innodb_log_waits.
recv_recover_page(): Do not unnecessarily acquire
log_sys.flush_order_mutex. We are inserting the blocks in arbitary
order anyway, to be adjusted in recv_sys.apply(true).
We will change the definition of flush_lock and write_lock to
avoid potential false sharing. Depending on sizeof(log_sys) and
CPU_LEVEL1_DCACHE_LINESIZE, the flush_lock and write_lock could
share a cache line with each other or with the last data members
of log_sys.
Thanks to Matthias Leich for providing https://rr-project.org traces
for various failures during the development, and to
Thirunarayanan Balathandayuthapani for his help in debugging
some of the recovery code. And thanks to the developers of the
rr debugger for a tool without which extensive changes to InnoDB
would be very challenging to get right.
Thanks to Vladislav Vaintroub for useful feedback and
to him, Axel Schwenke and Krunal Bauskar for testing the performance.
The InnoDB DATA DIRECTORY attribute is not implemented via
symbolic links but something similar, *.isl files that contain
the names of data files.
InnoDB failed to ignore the DATA DIRECTORY attribute even though
the server was started with --skip-symbolic-links.
Native ALTER TABLE in InnoDB will retain the DATA DIRECTORY attribute
of the table, no matter if the table will be rebuilt or not.
Generic ALTER TABLE (with ALGORITHM=COPY) as well as TRUNCATE TABLE
will discard the DATA DIRECTORY attribute.
All tests have been run with and without the ./mtr option
--mysqld=--skip-symbolic-links
and some tests that use the InnoDB DATA DIRECTORY attribute
have been adjusted for this.
bzip2/lz4/lzma/lzo/snappy compression is now provided via *services*
they're almost like normal services, but in include/providers/
and they're supposed to provide exactly the same interface
as original compression libraries (but not everything,
only enough of if for the code to compile).
the services are implemented via dummy functions that return
corresponding error values (LZMA_PROG_ERROR, LZO_E_INTERNAL_ERROR, etc).
the actual compression libraries are linked into corresponding
provider plugins. Providers are daemon plugins that when loaded
replace service pointers to point to actual compression functions.
That is, run-time dependency on compression libraries is now on plugins,
and the server doesn't need any compression libraries to run, but
will automatically support the compression when a plugin is loaded.
InnoDB and Mroonga use compression plugins now. RocksDB doesn't,
because it comes with standalone utility binaries that cannot
load plugins.
MDEV-26193 exposed a race condition in the test
mariabackup.log_page_corruption by no longer waking up purge tasks
on every transaction commit. (Note that there never was a guarantee
that the purge tasks would actually run as quickly as the test expected
it to happen; that would depend on the CPU load.)
If an ALTER TABLE that affected a table with FULLTEXT INDEX was
in progress while a backup was made, InnoDB would crash or hang
during the first startup after restoring the backup,
while trying to drop the #sql-alter- table for the DDL operation.
drop_garbage_tables_after_restore(): Invoke dict_sys.unlock()
before invoking the FTS functions. Also, invoke purge_sys.stop_FTS()
in debug builds to silence debug assertions. (Purge is not yet
running at this point.)
The issue was that we sent two different signals to different threads
after each other. The DEBUG_SYNC functionality cannot handle this (as
the signal is stored in a global variable) and the first one can get
lost.
Fixed by using the same signal for both threads.
In commit 1c5ae99194 (MDEV-25666)
we had changed Mariabackup so that it would no longer skip files
whose names start with #sql. This turned out to be wrong.
Because operations on such named files are not protected by any
locks in the server, it is not safe to copy them.
Not copying the files may make the InnoDB data dictionary
inconsistent with the file system. So, we must do something
in InnoDB to adjust for that.
If InnoDB is being started up without the redo log (ib_logfile0)
or with a zero-length log file, we will assume that the server
was restored from a backup, and adjust things as follows:
dict_check_sys_tables(), fil_ibd_open(): Do not complain about
missing #sql files if they would be dropped a little later.
dict_stats_update_if_needed(): Never add #sql tables to
the recomputing queue. This avoids a potential race condition when
dropping the garbage tables.
drop_garbage_tables_after_restore(): Try to drop any garbage tables.
innodb_ddl_recovery_done(): Invoke drop_garbage_tables_after_restore()
if srv_start_after_restore (a new flag) was set and we are not in
read-only mode (innodb_read_only=ON or innodb_force_recovery>3).
The tests and dbug_mariabackup_event() instrumentation
were developed by Vladislav Vaintroub, who also reviewed this.
Problem:
=======
When the semisync master is crashed and restarted as slave it could
recover transactions that former slaves may never have seen.
A known method existed to clear out all prepared transactions
with --tc-heuristic-recover=rollback does not care to adjust
binlog accordingly.
Fix:
===
The binlog-based recovery is made to concern of the slave semisync role of
post-crash restarted server.
No changes in behavior is done to the "normal" binloggging server
and the semisync master.
When the restarted server is configured with
--rpl-semi-sync-slave-enabled=1
the refined recovery attempts to roll back prepared transactions
and truncate binlog accordingly.
In case of a partially committed (that is committed at least
in one of the engine participants) such transaction gets committed.
It's guaranteed no (partially as well) committed transactions
exist beyond the truncate position.
In case there exists a non-transactional replication event
(being in a way a committed transaction) past the
computed truncate position the recovery ends with an error.
As after master crash and failover to slave, the demoted-to-slave
ex-master must be ready to face and accept its own (generated by)
events, without generally necessary --replicate-same-server-id.
So the acceptance conditions are relaxed for the semisync slave
to accept own events without that option.
While gtid_strict_mode ON ensures no duplicate transaction can be
(re-)executed the master_use_gtid=none slave has to be
configured with --replicate-same-server-id.
*NOTE* for reviewers.
This patch does not handle the user XA which is done
in next git commit.
