The Binary Log

The binary log in MySQL has two main declared purpose, replication and PTR (point in time recovery), as declared in the MySQL manual. In the MySQL binary log are stored all that statements that cause a change in the database. In short statements like DDL such as ALTER, CREATE, and so on, and DML like INSERT, UPDATE and DELETE.

As said binary log transform multi thread concurrent data/structure modification in to a serialize steps of activity. To handle this, a lock for the binary log—the LOCK_log mutex—is acquired just before the event is written to the binary log and released just after the event has been written. Because all session threads for the server log statements to the binary log, it is quite common for several session threads to block on this lock.

The other side effect of this is on SLAVE side. There the binary log becomes the relay-log, and will be processed SINGLE threaded.

In the MASTER it looks like:

While on SLAVE:

It is quite clear how the SLAVE cannot work efficiently in the case we will have a high level of  data modification on the MASTER.

And now the story

The other day I was working on a client site, I was performing some simple benchmarking to identify the correct thresholds for monitoring tools and so on.

While doing that I start to stress a little his server a 24CPU 48GB RAM SAN attached storage running Linux kernel 2.6.

The client had decided, before I or (we) can say anything to use MMM for HA, given that to have an up to date replication is a crucial factor for their HA architecture.

The test I was running execute inserts and selects, of 650 Bytes per insert using scalable number of thread, modifiable complexity in the SELECT, can do single insert or batch inserts and finally can do also BLOB/TEXT inserts (in that case the size per insert will jump up).

I was executing my standard low traffic exercises, when I realized that with only 6 threads doing 318 inserts per second (204Kb/s) the slaves were lagging behind the master for an indecent number of seconds.

I was expecting some lag but not so much. I stop any activity on the Master and wait for slave to recover, it takes 25 minutes to the slaves to recover/apply 20 seconds of inserts on the master.

I was shocked. That was too much considering I normally reach the 36 threads running at the same time on 5.1 and 144 on 5.5  before starting to see scaling issue and performance degradations (not in the scope of this article to publish that results).

Anyhow the issue was on replication not on Master. I was in the need to find a solution for the client or at least try before suggesting an architecture review.

Given re-inventing the wheel it is not my “motto” I started to look around for possible solutions, and parallelization seems the most interesting one.

My first though goes to Tungsten from Continuent, this also thanks to the very nice bogs coming from Giuseppe Maxia :

http://datacharmer.blogspot.com/2011/02/advanced-replication-for-masses-part-i.html

http://datacharmer.blogspot.com/2011/02/advanced-replication-for-masses-part-ii.html

http://datacharmer.blogspot.com/2011/03/advanced-replication-for-masses-part.html

Giuseppe was providing a superb introduction on one of the most interesting (actually the only really usable) solution currently available, the “Tungsten Replicator”.

I was reading the articles and get excited more and more.
That was exactly what I was looking for, until I read “A Shard, in this context, is the criteria used to split the transactions across channels. By default, it happens by database”.

Ooops, it will not work for my customer and me, given that his instance is mainly single schema, or at least the major amount of traffic is directed to only one schema.

I pinged Giuseppe, and we had a very interesting talk, he confirms the approach by object/schema, suggesting alternatives for my issue. Unfortunately given the special case no sharding could be done or table distribution, so for the moment and for the specific case this solution will not fit.

Looking around I had seen that almost all use the same approach, which is parallel replication by OBJECT where object is again and unfortunately, mainly the SCHEMA.

Sharding replication by schema is an approach that could work in some cases but not all.

If you have many schemas, with a limited number of well-distributed Questions per second, I agree this solution is a very good one.

If you have only one schema with very high level of Questions per seconds, this will not work anymore, for sure we can distribute the tables in different schemas, but what about transactions doing inserts/modifications in multiple tables?

We should be sure that each action is running on isolate state, so no relation with others. In short it will not work.

Having the possibility to dream, I would like to have parallel replication by MySQL User/Connections.

I know it could looks as a crazy idea, but trying to replicate the behavior of the MASTER on the SLAVES starts from the user(s) behavior, not from the Object inside the server instance.

Trying to move objects from A to B (see from MASTER to SLAVE) is really complicate because it needs to take in to account how people->application behave on a specific OBJECT or set of them. The status of the OBJECTS is subject to constant changes and most of them at the same time, so concurrency and versioning needs to be preserve.

In short what I was thinking was to realize a conceptual shift from REPLICATION to MIMIC approach.

I started my dream looking on what we already have and how it could be used.

We have a flexible binary log format (version4), we have a known set of SCHEMAS, USERS, and connections going from users to schemas. We have (at least for Innodb) the possibility to have good concurrent and versioning mechanism. The MAX_USER_CONNECTIONS can be used to tune the number of threads the specific user can use, and assign also the permissions.

Setting MAX_USER_CONNECTIONS=30 to USER1@ will help us in correctly dimensioning the effort that replication by user will have to do in checking for consistency.

I am aware of the issue given the interaction and relation between user/transactions, and I see the need of having a coordinator that will eventually manage the correct flush of the transaction by epoch.

What I was thinking is that the coordinator on the slave could open as much as MAX_USER_CONNECTIONS on the master, and use them as flushing threads.

As said before, I was thinking to trust the internal versioning, so not supporting transaction engines are excluded.

The epoch and/or Transaction ID would then be used to flush them in correct order, where the flush will be done to the pages in memory and then use internal mechanism for flushing to disk(s), talking of InnoDB.

It will not be real parallelization but also not a serialization given we will trust the internal versioning, so we will respect the EPOCH for the launch order not waiting for the completion.

On MASTER the replicator agent will create an X number of Replication channel for each authorized/configured user per SCHEMA, users and permission must be the same also on SLAVE.

The binary log write could looks like:

On the SLAVE, the logs will be copied and then processed opening 1 to MAX_USER_CONNECTIONS Thread(s). 
Coordination between insert will be taken by replication MUTEX per User based on the EPOCH.

I know quite difficult to realize given the fact that we will have more users writing on different schemas. To have it working correctly, a MUTEX should check the correct process execution at Relay Log Coordinator, ensuring also a correct execution order base on the EPOCH.

Sounds like another serialization, also if we will not wait for the transaction to complete, but only to start.

It seems that a mixed approach will be a better solution, such that replication by SHARD will be used for DB and then per USER inside only by SHARD.  This means archiving action per schema and per user transaction. Replicating the work on the SLAVE ensuring to have the same start order of the transaction for the modifications. This is not so crazy as most can think and not so far from sharding by schema, except the multi-threading during the writes.

Main point will be archiving correctly the information in the binary log:

And execute them in the right order as well, keeping also into account the User/Transaction order:

Relay logs will contain information divide by SCHEMA. Processed using multi thread per User/Connection and organize with reference to the epoch of the starting execution order on the MASTER. Different grades of parallel or serialization execution could be set on the how the single Schema relay coordinator will work.

I know it will not be full parallelization but is not serializing as well, given the possible parallel execution of N transaction per user.

I think that there could be some risk here so in the case we should be able to:

-       Configure the number of executing threads per user

-       Execute only one thread per user

-       Choose between EPOCH execution and on Lock relief execution

Fast/short comment on the new binary log.

From binary log V4 we have the possibility to extend what the binary log store in a flexible way.

The extra headers (from byte 19 to X) are there but not used, while that sounds to me the most interesting place where to store additional information regarding what and how happen on the MASTER.

Not only, an HASH value representing the data on the master after the operation, would be of incredible help in checking if the replicated data on the SLAVE is consistent with the MASTER or not, avoiding to do that with silly methods like we are force to do today.

Below the binary log V4 format description:

v4 format description event (size >= 91 bytes; the size is 76 + the number of event types):

+=====================================+
| event  | timestamp         0 : 4    |
| header +----------------------------+
|        | type_code         4 : 1    | = FORMAT_DESCRIPTION_EVENT = 15
|        +----------------------------+
|        | server_id         5 : 4    |
|        +----------------------------+
|        | event_length      9 : 4    | >= 91
|        +----------------------------+
|        | next_position    13 : 4    |
|        +----------------------------+
|        | flags            17 : 2    |
+=====================================+
| event  | binlog_version   19 : 2    | = 4
| data   +----------------------------+
|        | server_version   21 : 50   |
|        +----------------------------+
|        | create_timestamp 71 : 4    |
|        +----------------------------+
|        | header_length    75 : 1    |
|        +----------------------------+
|        | post-header      76 : n    | = array of n bytes, one byte per event
|        | lengths for all            |   type that the server knows about
|        | event types                |
+=====================================+

v4 event header:
+============================+
| timestamp         0 : 4    |
+----------------------------+
| type_code         4 : 1    |
+----------------------------+
| server_id         5 : 4    |
+----------------------------+
| event_length      9 : 4    |
+----------------------------+
| next_position    13 : 4    |
+----------------------------+
| flags            17 : 2    |
+----------------------------+
| extra_headers    19 : x-19 |
+============================+

Conclusion

My favorite mottos are:
 “only the ones that do nothing do not make mistakes”

and

“don’t shoot on the pianist”.

As I say from the beginning mine is a dream, and I am just proposing an idea, specially because I see that almost all were following the same mind path focused on OBJECTS, probably for a good reason, but never the less I am following my dream.

It could happen that someone would like to see/try to work on that or that will use part of the idea revisiting modifying and making it possible to implement.

For the moment it is too complex and taking too much time, for me to do it alone, but it could be a good idea for a plug-in in my next life.

References:

In addition to Giuseppe blog, which for now it seems to me the only one proposing something that could be implement, I did some research, but at the end the actors are always the same:

Percona

http://www.percona.com/docs/wiki/percona-server:blueprints:parallel-replication

http://www.mysqlperformanceblog.com/2007/01/30/making-mysql-replication-parallel/

MySQL forge

http://forge.mysql.com/worklog/task.php?id=5569

Other

http://feedblog.org/2007/01/30/design-proposal-for-multithreaded-replication-in-mysql/

http://scale-out-blog.blogspot.com/2010/10/parallel-replication-on-mysql-report.html

Note

I already received a comment fromGiuseppe about how to handle the Locks and prevent dead lock... working on that (I have some ideas and details will provide soon).

{joscommentenable}

The Binary Log

The binary log in MySQL has two main declared purpose, replication and PTR (point in time recovery), as declared in the MySQL manual. In the MySQL binary log are stored all that statements that cause a change in the database. In short statements like DDL such as ALTER, CREATE, and so on, and DML like INSERT, UPDATE and DELETE.

As said binary log transform multi thread concurrent data/structure modification in to a serialize steps of activity. To handle this, a lock for the binary log—the LOCK_log mutex—is acquired just before the event is written to the binary log and released just after the event has been written. Because all session threads for the server log statements to the binary log, it is quite common

for several session threads to block on this lock.

It is not in the scope of this discussion to provide a detailed description of the binary log, which could be found at http://dev.mysql.com/doc/refman/5.1/en/binary-log.html.

What it is relevant is that MySQL write "a" log and read "a" log, serializing what is happening in the MASTER in a multi thread scenario. Also relevant is that MySQL has to deal with different behavior in relation to TRANSACTIONAL and NON-TRANSACTIONAL storage engine, trying to behave consistently.

In order to do that MySQL had changes his behavior while evolving, so it changes the way he was and is using the TRANSACTION CACHE also creating not few confusion.

Transaction Cache

What is the Transaction Cache or TC? It is a "per thread" cache used by MySQL to store the statements that will be needed to be written in the binary log.

This cache is active ONLY when the Binary log is active, otherwise it will not be used, and, as it declare, it is a "Transaction" cache so used to store statements executed in a transaction, and that will be consistently flushed in the binary log serializing the multi thread execution.

But what if we also have non transactional storage engine involved in our operations? Well here is where the fun begins, and also where the difference in how MySQL acts makes our life quite complicate.

Actors involved (who control what)

The TC (transaction Cache) is currently controlled by few parameters, and it is important to have them in mind:

      log-bin

o   this variable is to enable the binary log, if not present all the structure related to it is not active;

      binlog_cache_size

o   The size of the cache to hold changes to the binary log during a transaction

      binlog_stmt_cache_size

o   this variable determines the size of the cache for the binary log to hold nontransactional statements issued during a transaction.

      binlog_direct_non_transactional_updates

o   This variable when enabled, causes updates to non-transactional tables to be written directly to the binary log, rather than to the transaction cache.

      Binlog_cache_use vs Binlog_cache_disk_use

o   Status Variables to use for checking the efficiency of the cache.

      Binlog_stmt_cache_use vs Binlog_stmt_cache_disk_use

o   Status Variables to use for checking the efficiency of the cache.

MySQL Transaction cache and MySQL versions

Ok let start trying to define some milestones, like what is what, in which version, and what works and what doesn't.

binlog_cache_size is present from the start but behave different, in:

5.1 controls the whole transaction cache

5.5.3 up to 5.5.8 controls both size of the statement and none cache

5.5.9 and newer controls only the transaction cache demanded to transactions

binlog_stmt_cache_size

5.1 not present

5.5.9 it controls the dimension of the transaction cache demanded to store the NON transactional statements present in a transaction.

binlog_direct_non_transactional_updates

5.1.44 Introduced. it is DISABLE by default. If enable it allows all statements executed, also AFTER a Transaction (we will see later what this means), to be directly flushed to the binary log without going to the TC. It has effect only if the binlog_format is set to STATEMENT or MIXED.

5.5.2 Introduced.

5.5.5 It was modify to be active ONLY if binlog_format is STATEMENT in any other case it is ignored.

This feature, regardless the MySQL version, create more problems then solutions, and honestly I think it should be removed, please see the bug (http://bugs.mysql.com/bug.php?id=51291) to understand what I mean. In any case it has now a very limited scope of action, and better to consider it as not present.

How the whole works

The first point to take in mind, is that the TC is used only during a transaction, also the new TCStatement, is suppose to store only statement that are issued during a transaction, and it will be flush at the moment of the COMMIT, actually it go the queue for acquiring the LOG_LOCK mutex.

All other statements will directly go to the LOG_LOCK mutex queue.

Said that let us see how the TC works and what it does and how.

In general terms when you mix the Transactional and NON-transactional statements these rules apply, unless the infamous

binlog_direct_non_transactional_updates is enable (only 5.1):

1. If the statement is marked as transactional, it is written to the transaction cache.

2. If the statement is not marked as transactional and there are no statements in the

transaction cache, the statement is written directly to the binary log.

3. If the statement is not marked as transactional, but there are statements in the

transaction cache, the statement is written to the transaction cache.

Interesting point here is what and how is define as NON Transactional or as TRANSACTIONAL, from reading the manual and the many reference we have that:

After MySQL 5.1.31

a statement is considered non-transactional if it changes only non-transactional tables;

a statement is transactional if it changes only transactional tables.

Finally a statement that changes both non-transactional and transactional tables is considered “mixed”. Mixed statements, like transactional statements, are cached and logged when the transaction commits.

Before MySQL 5.1.31

a statement was considered non-transactional if the first changes it makes change non-transactional tables;

a statement was considered transactional if the first changes it makes change transactional tables.

In term of code:

Before MySQL 5.1.31

After MySQL 5.1.31

A graph could help as well to understand what happens between threads and flush of the TC in the binary log.

In the graphs we have three treads running in parallel, remember that inside mysql the main entity is the THD "The THD class defines a thread descriptor. It contains the information pertinent to the thread that is handling the given request. Each client connection is handled by a thread. Each thread has a descriptor object."

The first thread will do all the operations using TRANSACTION storage engine (InnoDB);

the second thread will use NON-TRANSACTIONAL (MyISAM) in the first insert, then TRANSACTIONAL (InnoDB);

the third statement will do first a TRANSACTIONAL insert and then will use it for a NON-TRANSACTIONAL.

In the first one I have assume the 5.1.31 and over behavior.

--------------------------------------------------------------

The three statements start almost at the same time, and none of them has data in any cache or buffer. Said that it is probably true that statement 1 in Thread2 will be executed (fully) faster, given it will take less as overhead, and giving that is the first statement in the THD2 transaction, it will be treated as NON-TRANSACTIONAL and directly sent to the binary-log's queue.

THD1 is fully TRANSACTIONAL, it will do his internal action then flush to binary-log.

Then THD2 had done also the second part of the operation TRANSACTIONAL and it is flushing it as well.

Last the THD3 which had first insert a record from TRANSACTIONAL table and then use it to populate the NON-TRANSACTIONAL.

Note

It looks more or less fine, isen't it? Well but assume that all the three THD share the InnoDB table and that as per default the ISOLATION LEVEL is Repeatable Read... Do you get it??

Actions and value on the MASTER will do and set values, that are potentially different from the ones in the SLAVE given the different order in the execution.

Immagine to have updates, instead insert and/or WHERE condition by values. Results could be different given the position, and remember, we are still talking about the STATEMENT replication given it is still the default ALSO in 5.5.

Finally remember that the issue it is NOT that THD1/2/3 could potentially change each other value, but the fact that they do something on the MASTER which could potentially different in the SLAVE.

That's it, the changes introduced in MySQL were done to reduce this wrong behavior, or to in some way try to keep consistency.

Let see what and how in the others graphs.

What about using binlog_direct_non_transactional_updates, will it helps?

I admit that I have exaggerated a little bit, in this example but my intention was to show how dangerous this option could be.

THD2 statement 1, as usual will be the first one, then THD1 consistently apply the whole set, but then it could happens that given the option set binlog_direct_non_transactional_updates THD3 will be faster in flushing the NON-TRANSACTIONAL statement, writing to the binary log before THD2, adding fragmentation and breaking the order in the binary log.

As said I have exaggerate a little, but theoretically this is very likely to happen in a standard context.

So in addition to the already found bug(s) (see the reference section), this option for me it is not really an additional help, but an issue generator.

Transaction Cache in MySQL 5.5

From MySQL 5.5.4 the TC is now divide in two parts, one control/store the TRANSACTIONAL statements, another store ONLY the NON_TRANSACTIONAL statements.

What does it means? In theory it means that any statement which was suppose to be part of a TRANSACTION but related to a NON-TRANSACTIONAL table (like statement at the start of a transaction), instead going directly to the binary log should be moved to the TC-statement, and then flushed when the TC is flushed.

The involvement of the binlog_direct_non_transactional_updates complicate  a little all the scenario and in addition to that, it also create some major issue in the correct flush, so I am ignoring it here.

Let us see what happens using the new TC mechanism:

THD1 is as usual consistent in the way it flush to binary log. Arbitrarily I have decide that this time THD2 is taking a little bit more time like a millisecond, and that is enough for THD1 to acquire the LOG_LOCK before it.

The scope as you have probably guess was to make more ORDERED the graph, given that this time having one or the other flushing before or after will not really make any difference this time.

This because the TCStatement will be flushed (at least this is what I have seen and understood) when the TC is flushed.

Result is that the flush will take place using same the order on master and on the slave.

Finally THD3 also if will have the possibility to use the TCstatements it will not giving the fact that the statement using a NON-TRANSACTIONAL storage engine is NOT the first statement, so the mixed transaction will be placed in the TC as expected.

Conclusion

There was not so much to say from the beginning, to be honest, the TC and TCstatements are useful only for that case we do mix-transactions. The only significant changes are between after 5.1.31 and before and the introduction of the TCstatement. All the other tentative to add flexibility to this area, had in my vision increased bad behaviors.

It is easy for Master and Slave to store or modify data in a unexpected way, issue that any DBA face probably too often in MySQL.

I am not saying that all that cases should bring back to this specific issue, but for sure it is one of the less known and difficult to track, because many things do influence the how MySQL will flush to binary log.

The golden rule is, do not mix NON-TRANSACTIONAL and TRANSACTIONAL, but if you really have to do that, be sure of doing it in using 5.5.9 and above, which at least gives you more insurance of flushing to binary log the commands consistently.

Last notes

Remember that any statement NON-TRANSACTIONAL in the TC cache will be flushed on COMMIT or ROLLBACK, and that in the latest case a WARNING message should be written in the MySQL error log.

Note that I will post later (when I will have the time) some code and related raw data for documentation purpose.

References:

--------------

http://dev.mysql.com/doc/refman/5.1/en/replication-features-transactions.html

http://dev.mysql.com/doc/refman/5.1/en/replication-options-binary-log.html#sysvar_binlog_cache_size

http://dev.mysql.com/doc/refman/5.1/en/replication-options-binary-log.html#sysvar_binlog_direct_non_transactional_updates

http://dev.mysql.com/doc/refman/5.5/en/replication-options-binary-log.html#sysvar_binlog_stmt_cache_size

http://bugs.mysql.com/bug.php?id=51291

http://www.chriscalender.com/?tag=large-transaction

http://mysqlmusings.blogspot.com/2009/02/mixing-engines-in-transactions.html

##############
#REPLICATION
#############
log_bin            =  /var/mysql/log/binlog51/mysql-bin
relay-log        =  /var/mysql/log/relaylog/relay

expire_logs_days      = 30
max_binlog_size       = 500M
log-slave-updates      = 1
max_binlog_cache_size = 4294967295
sync_binlog              = 1    

#skip-slave-start
replicate-ignore-db=mysql

mysql> show binary logs;

...
100929 11:21:47 [Note] Slave SQL thread initialized, starting replication 
in log 'mysql-bin.002671' at position 1073741674, relay log './xxxxxxxxxxxx-relay-bin.000001' position: 4
100929 11:21:47 [Note] Slave I/O thread: connected to master This email address is being protected from spambots. You need JavaScript enabled to view it.:3306',
  replication started in log 'mysql-bin.002671' at position 1073741674
100929 11:21:47 [ERROR] Error reading packet from server: Could not find first log 
file name in binary log index file ( server_errno=1236)
100929 11:21:47 [ERROR] Got fatal error 1236: 'Could not find first log file name 
in binary log index file' from master when reading data from binary log
100929 11:21:47 [Note] Slave I/O thread exiting, read up to log 'mysql-bin.002671', position 1073741674
100929 11:22:08 [Note] Error reading relay log event: slave SQL thread was killed

 if (log_type == LOG_BIN) 
 { 
 if (!no_auto_events) 
 { 
 /* 
 We log the whole file name for log file as the user may decide 
 to change base names at some point. 
 */ 
 THD *thd = current_thd; /* may be 0 if we are reacting to SIGHUP */ 
 Rotate_log_event r(thd,new_name+dirname_length(new_name), 
 0, LOG_EVENT_OFFSET, 0); 
 r.write(&log_file); 
 bytes_written += r.data_written; 
 } 
 /* 
 Update needs to be signalled even if there is no rotate event 
 log rotation should give the waiting thread a signal to 
 discover EOF and move on to the next log. 
 */ 
 signal_update(); 
 } 
 old_name=name; 
 save_log_type=log_type; 
 name=0;    // Don't free name 
 close(LOG_CLOSE_TO_BE_OPENED | LOG_CLOSE_INDEX);

The function will close the current BINLOG and also the the <binlog>.index associated, then will ROTATE
the binarylog, and create new one with a new name, finally reopen a new  <binlog>.index.

{joscommentenable}

I have a problem, It's been months since I used MySQL and (I believe) I had set it up with a root password.  Now I can't log on to MySQL as root MySQL user and create a new user or manage an existing user (I can log onto server)...

Familiar situation. :)

Do so:

 service mysql stop

wait until MySQL shuts down. Then run

mysqld_safe --skip-grant-tables &

then you will be able to login as root with no password.

mysql -uroot mysql 

In MySQL command line prompt issue the following command:

UPDATE user SET password=PASSWORD("abcd") WHERE user="root"; 

FLUSH PRIVILEGES;

At this time your root password is reset to "abcd" and MySQL will now
know the privileges and you'll be able to login with your new password:
 

mysql -uroot -pabcd mysql