Oct
04
2018
--

The Importance of mysqlbinlog –version

Importance of MySQL binlog version

Importance of MySQL binlog versionWhen deciding on your backup strategy, one of the key components for Point In Time Recovery (PITR) will be the binary logs. Thankfully, the mysqlbinlog command allows you to easily take binary log backups, including those that would otherwise be encrypted on disk using encrypt_binlog=ON.

When

mysqlbinlog

  is used with

--raw --read-from-remote-server --stop-never --verify-binlog-checksum

  then it will retrieve binary logs from whichever master it is pointed to, and store them locally on disk in the same format as they were written on the master. Here is an example with the extra arguments that would normally be used:

/usr/bin/mysqlbinlog --raw --read-from-remote-server \
 --stop-never --connection-server-id=1234 \
 --verify-binlog-checksum \
 --host=localhost --port=3306 mysql-bin.000024

This would retrieve the localhost binary logs (starting from mysql-bin.000024) reporting as server_id 1234, verify the checksum and then write each of them to disk.

Changes to the mysqlbinlog source code are relatively infrequent, except for when developing for a new major version, so you may be fooled into thinking that the specific version that you use is not so important—a little like the client version. This is something that is more likely to vary when you are taking remote backups.

Here is the result from the 5.7 branch of mysql-server to show the history of commits by year:

$ git blame --line-porcelain client/mysqlbinlog.cc | egrep "^(committer-time|committer-tz)" | cut -f2 -d' ' | while read ct; do read ctz; date --date "Jan 1, 1970 00:00:00 ${ctz} + ${ct} seconds" --utc +%Y; done | sort -n | uniq -c
   105 2000
    52 2001
    52 2002
   103 2003
   390 2004
   113 2005
    58 2006
   129 2007
   595 2008
    53 2009
   349 2010
   151 2011
   382 2012
   191 2013
   308 2014
   404 2015
    27 2016
    42 2017
    15 2018

Since the first GA release of 5.7 (October 2015), there haven’t been too many bugs and so if you aren’t using new features then you may think that it is OK to keep using the same version as before:

$ git log --regexp-ignore-case --grep bug --since="2015-10-19" --oneline client/mysqlbinlog.cc
1ffd7965a5e Bug#27558169 BACKPORT TO 5.7 BUG #26826272: REMOVE GCC 8 WARNINGS [noclose]
17c92835bb3 Bug #24674276 MYSQLBINLOG -R --HEXDUMP CRASHES FOR INTVAR,                   USER_VAR, OR RAND EVENTS
052dbd7b079 BUG#26878022 MYSQLBINLOG: ASSERTION `(OLD_MH->M_KEY == KEY) ||              (OLD_MH->M_KEY == 0)' FAILED
543129a577c BUG#26878022 MYSQLBINLOG: ASSERTION `(OLD_MH->M_KEY == KEY) || (OLD_MH->M_KEY == 0)' FAILED
ba1a99c5cd7 Bug#26825211 BACKPORT FIX FOR #25643811 TO 5.7
1f0f4476b28 Bug#26117735: MYSQLBINLOG READ-FROM-REMOTE-SERVER NOT HONORING REWRITE_DB FILTERING
12508f21b63 Bug #24323288: MAIN.MYSQLBINLOG_DEBUG FAILS WITH A LEAKSANITIZER ERROR
e8e5ddbb707 Bug#24609402 MYSQLBINLOG --RAW DOES NOT GET ALL LATEST EVENTS
22eec68941f Bug#23540182:MYSQLBINLOG DOES NOT FREE THE EXISTING CONNECTION BEFORE OPENING NEW REMOTE ONE
567bb732bc0 Bug#22932576 MYSQL5.6 DOES NOT BUILD ON SOLARIS12
efc42d99469 Bug#22932576 MYSQL5.6 DOES NOT BUILD ON SOLARIS12
6772eb52d66 Bug#21697461 MEMORY LEAK IN MYSQLBINLOG

However, this is not always the case and some issues are more obvious than others! To help show this, here are a couple of the issues that you might happen to notice.

Warning: option ‘stop-never-slave-server-id’: unsigned value <xxxxxxxx> adjusted to <yyyyy>

The server_id that is used by a server in a replication topology should always be unique within the topology. One of the easy ways to ensure this is to use a conversion of the external IPv4 address to an integer, such as INET_ATON , which provides you with an unsigned integer.

The introduction of

--connection-server-id

 (which deprecates

--stop-never-slave-server-id

 ) changes the behaviour here (for the better). Prior to this you may experience warnings where your server_id was cast to the equivalent of an UNSIGNED SMALLINT. This didn’t seem to be a reported bug, just fixed as a by-product of the change.

ERROR: Could not find server version: Master reported unrecognized MySQL version ‘xxx’

When running mysqlbinlog, the version of MySQL is checked so that the event format is set correctly. Here is the code from MySQL 5.7:

switch (*version) {
 case '3':
   glob_description_event= new Format_description_log_event(1);
   break;
 case '4':
   glob_description_event= new Format_description_log_event(3);
   break;
 case '5':
   /*
     The server is soon going to send us its Format_description log
     event, unless it is a 5.0 server with 3.23 or 4.0 binlogs.
     So we first assume that this is 4.0 (which is enough to read the
     Format_desc event if one comes).
   */
   glob_description_event= new Format_description_log_event(3);
   break;
 default:
   glob_description_event= NULL;
   error("Could not find server version: "
         "Master reported unrecognized MySQL version '%s'.", version);
   goto err;
 }

This section of the code last changed in 2008, but of course there is another vendor that no longer uses a 5-prefixed-version number: MariaDB. With MariaDB, it is impossible to take a backup without using a MariaDB version of the program, as you are told that the version is unrecognised. The MariaDB source code contains a change to this section to resolve the issue when the version was bumped to 10:

83c02f32375b client/mysqlbinlog.cc (Michael Widenius    2012-05-31 22:39:11 +0300 1900) case 5:
83c02f32375b client/mysqlbinlog.cc (Michael Widenius    2012-05-31 22:39:11 +0300 1901) case 10:

Interestingly, MySQL 8.0 gets a little closer to not producing an error (although it still does), but finally sees off those rather old ancestral relatives:

 switch (*version) {
   case '5':
   case '8':
   case '9':
     /*
       The server is soon going to send us its Format_description log
       event.
     */
     glob_description_event = new Format_description_log_event;
     break;
   default:
     glob_description_event = NULL;
     error(
         "Could not find server version: "
         "Master reported unrecognized MySQL version '%s'.",
         version);
     goto err;
 }

These are somewhat trivial examples. In fact, you are more likely to suffer from more serious differences, perhaps ones that do not become immediately apparent, if you are not matching the mysqlbinlog version to the one provided by the version for the server producing the binary logs.

Sadly, it is not so easy to check the versions as the reported version was seemingly left unloved for quite a while (Ver 3.4), so you should check the binary package versions (e.g. using Percona-Server-client-57-5.7.23 with Percona-Server-server-57-5.7.23). Thankfully, the good news is that MySQL 8.0 fixes it!

So reduce the risk and match your package versions!

Jul
31
2014
--

MariaDB: Selective binary logs events

In the first post in a series on MariaDB features we find interesting, we begin with selectively skipping replication of binlog events. This feature is available on MariaDB 5.5 and 10.

By default when using MySQL’s standard replication, all events are logged in the binary log and those binary log events are replicated to all slaves (it’s possible to filter out some schema). But with this feature, it’s also possible to bypass some events to be replicated on the slave(s) even if they are written in the binary log. Having those event in the binary logs is always useful for point-in-time recovery.

Indeed, usually when we need to not replicate an event, we set sql_log_bin = 0 and the event is bypassed: neither written into the binlog, neither replicated to slave(s).

So with this new feature, it’s possible to just set a session variable to tag events that will be written into the binary log and bypassed on demand on some slaves.

And it’s really easy to use, on the master you do:

set skip_replication=1;

and on the slave(s) having replicate_events_marked_for_skip='FILTER_ON_MASTER' or 'FILTER_ON_SLAVE' the events skipped on the master won’t be replicated.

The valid values for replicate_events_marked_for_skip are:

  • REPLICATE (default) : skipped events are replicated on the slave
  • FILTER_ON_SLAVE : events so marked will be skipped on the slave and not replicated
  • FILTER_ON_MASTER : the filtering will be done on the master so the slave won’t even receive it and then save network bandwidth

That’s a cool feature but when this can be very useful?

Use case:

For archiving this can be very interesting. Indeed most of the time when people is archiving data, they use something like pt-archiver that deletes the data and copy the removed data on an archive server.

Thanks to this feature, instead of having an archiving server where we copy the deleted data, it’s possible to have a slave where we won’t delete the data. This will be much faster (smarter?) and allows to have an archiving server always up to date. Of course in this case sql_log_bin = 0 would have worked (if we ignore the point-in-time recovery).
But with a Galera Cluster? Yes that’s where this feature is really cool, if we would have used sql_log_bin = 0 on a Galera Cluster node, all other nodes would have ignored the delete and the result would be inconsistency between the nodes.

So if you use an asynchronous slave as an archiving server of a Galera Cluster, this feature is really mandatory.

As illustrated below, you can have a MariaDB Galera Cluster node joining a Percona XtraDB Cluster that will be used to delete historical data using pt-archiver:

archiving

pt-archiver is started with --set-vars "skip_replication=1"

The post MariaDB: Selective binary logs events appeared first on MySQL Performance Blog.

Powered by WordPress | Theme: Aeros 2.0 by TheBuckmaker.com