Load Balancer on FromDual GmbH

MaxScale configuration synchronisation

oli.sennhauser@fromdual.com (Oli Sennhauser) — Thu, 04 Apr 2024 09:53:46 +0000

Overview

A feature that I recently discovered while browsing is the MaxScale configuration synchronisation functionality.

Sharding with MariaDB MaxScale

oli.sennhauser@fromdual.com (Oli Sennhauser) — Tue, 19 Mar 2024 17:02:58 +0000

Overview

This feature should more or less work with MariaDB MaxScale 6.x.y, 22.08.x, 23.02.x, 23.08.x and 24.02.x. We have tested it with the latest MaxScale version 23.08.05, as we encountered problems with an older version (MXS-5026).

MariaDB MaxScale Load Balancer with Master/Master Replication

oli.sennhauser@fromdual.com (Oli Sennhauser) — Fri, 13 May 2022 16:39:27 +0000

For this how-to we were working with a MariaDB 10.6.5 database server and the newest MariaDB MaxScale 6.3.0 (aka 2.6) load balancer/proxy.

Traffic mirroring with MariaDB MaxScale

oli.sennhauser@fromdual.com (Oli Sennhauser) — Thu, 24 Dec 2020 16:27:33 +0000

Recently we had the case that a customer claimed that MariaDB 10.3 Binary Log is using 150% more space on disk than MySQL 5.7 Binary Log. Because I never observed something similar, but to be honest, I did not look to intensively for this situation, we had to do some clarifications.

MariaDB MaxScale Configuration Variables

oli.sennhauser@fromdual.com (Oli Sennhauser) — Wed, 04 Nov 2020 22:49:39 +0000

Global Options

Variable	Versions					Values	Default	Unit	Comment
threads	1.4		2.3	2.4	2.5	{ <n> \| auto }	1	thread
thread_stack_size		2.2				<n>			Ignored and deprecated in 2.3
rebalance_period					2.5	<n>	0	second
rebalance_threshold					2.5	<n>	20	delta load
rebalance_window					2.5	<n>	10	second
auth_connect_timeout	1.4	2.2	2.3	2.4		<n>	3	second
auth_read_timeout	1.4	2.2	2.3	2.4		<n>	1	second	Deprecated in 2.5
auth_write_timeout	1.4	2.2	2.3	2.4		<n>	2	second	Deprecated in 2.5
query_retries			2.3	2.4	2.5	<n>	1	time	Added in 2.1.10, disabled by default until 2.3.0
query_retry_timeout			2.3	2.4	2.5		5
passive			2.3	2.4	2.5		false
ms_timestamp	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0	-
skip_permission_checks			2.3	2.4	2.5	{ 0 \| 1 }	0
syslog	1.4		2.3	2.4	2.5	{ 0 \| 1 }	1	-
maxlog	1.4		2.3	2.4	2.5	{ 0 \| 1 }	1	-
log_to_shm	1.4					{ 0 \| 1 }	0	-	Deprecated and ignored in 2.3.0 and newer
log_warning	1.4		2.3	2.4	2.5	{ 0 \| 1 }	1	-
log_notice	1.4		2.3	2.4	2.5	{ 0 \| 1 }	1	-
log_info	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0	-
log_debug	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0	-
log_warn_super_user					2.5		false
log_messages									Deprecated, use log_notice instead
log_trace									Deprecated, use log_info instead
log_augmentation	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0	-
log_throttling			2.3	2.4	2.5	{ <n>, <n>, <n> }	10, 1000, 10000	time, millisecond, millisecond
logdir	1.4		2.3	2.4	2.5	<dirpath>	/tmp/	-
datadir	1.4		2.3	2.4	2.5	<dirpath>	/home/user/maxscale_data/	-
libdir	1.4		2.3	2.4	2.5	<dirpath>	/home/user/lib64/	-
cachedir	1.4		2.3	2.4	2.5	<dirpath>	/tmp/maxscale_cache/	-
piddir	1.4		2.3	2.4	2.5	<dirpath>	/tmp/maxscale_cache/	-
execdir	1.4		2.3	2.4	2.5	<dirpath>	/usr/local/bin/	-
connector_plugindir			2.3	2.4	2.5	<dirpath>	/usr/lib/plugin/
persistdir			2.3	2.4	2.5	<dirpath>	/var/lib/maxscale/maxscale.cnf.d/
module_configdir			2.3	2.4	2.5	<dirpath>	/var/lib/maxscale/
language	1.4		2.3	2.4	2.5	<dirpath>	/home/user/lang/	-
query_classifier			2.3	2.4	2.5	<classifier>	qc_sqlite
query_classifier_cache_size			2.3	2.4	2.5	<n>	15% of system Memory	byte
query_classifier_args			2.3	2.4	2.5
substitute_variables			2.3	2.4	2.5	{ true \| false }	false
sql_mode			2.3	2.4	2.5	{ default \| oracle }	default
local_address			2.3	2.4	2.5	<IP address>	*
users_refresh_time			2.3	2.4	2.5	<n>	30	second
user_refresh_interval				2.4	2.5	<n>	0	second
retain_last_statements			2.3	2.4	2.5	<n>	0	statement
dump_last_statements			2.3	2.4	2.5	{ on_error \| on_close \| never }	never	-
session_trace			2.3	2.4	2.5	<n>	0	statement
writeq_high_water			2.3	2.4	2.5
writeq_low_water			2.3	2.4	2.5
load_persisted_configs			2.3	2.4	2.5	{ 0 \| 1 }	1
max_auth_errors_until_block				2.4	2.5	<n>	10
debug					2.5
admin_host			2.3	2.4	2.5	<IP address>	127.0.0.1
admin_port			2.3	2.4	2.5	<n>	8989
admin_auth			2.3	2.4	2.5		1
admin_ssl_key			2.3	2.4	2.5
admin_ssl_cert			2.3	2.4	2.5
admin_ssl_ca_cert			2.3	2.4	2.5
admin_enabled			2.3	2.4	2.5	{ 0 \| 1 }
admin_gui					2.5		1
admin_secure_gui					2.5
admin_log_auth_failures				2.4	2.5	{ 0 \| 1 }	1
admin_pam_readwrite_service				2.4	2.5
admin_pam_readonly_service				2.4	2.5

Service Options

Variable	Versions					Values	Default	Unit	Comment
type	1.4		2.3	2.4	2.5	service		-
router	1.4		2.3	2.4	2.5	{ readconnroute \| readwritesplit \| schemarouter \| binlogrouter }		-
router_option	1.4		2.3	2.4	2.5	{ slave \| master,slave }		-
filters	1.4		2.3	2.4	2.5	{ counter \| QLA }		-
targets					2.5	{ server \| service, ... }
servers	1.4		2.3	2.4	2.5	{ <server1>, <server2>, ... }		-
cluster				2.4	2.5	<name>
user	1.4		2.3	2.4	2.5	{ <username> }		-
passwd	1.4		2.3	2.4	2.5	{ <password> }		-
enable_root_user	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0	-
localhost_match_wildcard_host	1.4		2.3	2.4		{ 0 \| 1 }	0		Deprecated and ignored.
version_string	1.4		2.3	2.4	2.5	<string>	5.5.5-10.0.0 MaxScale <MaxScale version>
weightby	1.4		2.3	2.4		<server_weight>			Deprecated in 2.3 and removed in 2.5, use rank instead
auth_all_servers	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0
strip_db_esc	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0
optimize_wildcard	1.4			2.4		{ 0 \| 1 }	0
retry_on_failure	1.4		2.3			{ 0 \| 1 }	1		Ignored by 2.4 and removed in the future.
log_auth_warnings	1.4		2.3	2.4	2.5	{ 0 \| 1 }	0
connection_timeout	1.4		2.3	2.4	2.5	<n>		second
max_connections			2.3	2.4	2.5	<n>		connection
max_retry_interval		2.2	2.3	2.4		<n>	3600	second
session_track_trx_state			2.3	2.4	2.5		false		Only supported by MariaDB 10.3 and newer
retain_last_statements			2.3	2.4	2.5
connection_keepalive					2.5	<n>	300
net_write_timeout				2.4	2.5		0	second

Routing Modules

ReadConnRoute

For (weighted) load balancing.

Galera Load Balancer the underestimated wallflower

oli.sennhauser@fromdual.com (Oli Sennhauser) — Thu, 21 Sep 2017 15:25:18 +0000

There are some pretty sophisticated Load Balancers for Galera Clusters setups out in the market (ProxySQL, MaxScale, HAproxy, …). They have many different exotic features. You can nearly do everything with them. But this comes at the cost of complexity. Non of them is simple any more.

Making HAProxy High Available for MySQL Galera Cluster

oli.sennhauser@fromdual.com (Oli Sennhauser) — Sun, 14 Dec 2014 18:37:50 +0000

After properly installing and testing a Galera Cluster we see that the set-up is not finished yet. It needs something in front of the Galera Cluster that balances the load over all nodes.
So we install a load balancer in front of the Galera Cluster. Typically nowadays HAProxy is chosen for this purpose. But then we find, that the whole Galera Cluster is still not high available in case the load balancer fails or dies. So we need a second load balancer for high availability.
But how should we properly failover when the HAProxy load balancer dies? For this purpose we put a Virtual IP (VIP) in front of the HAProxy load balancer pair. The Virtual IP is controlled and failovered with Keepalived.

Galera Load Balancer Documentation

oli.sennhauser@fromdual.com (Oli Sennhauser) — Sat, 17 Aug 2013 22:08:40 +0000

About

glbd is a simple TCP connection balancer made with scalability and performance in mind. It was inspired by pen, but unlike pen its functionality is limited only to balancing generic TCP connections.

Unbreakable MySQL Cluster with Galera and Linux Virtual Server (LVS)

oli.sennhauser@fromdual.com (Oli Sennhauser) — Thu, 13 Jun 2013 17:13:47 +0000

Recently we had to set-up a 3-node Galera Cluster with a Load Balancer in front of it. Because Galera Cluster nodes (mysqld) still reply to TCP requests on port 3306 when they are expelled from the Cluster it is not sufficient to just leave it to the Load Balancer to check the port if a Galera node is properly running or not.

MySQL and Galera Load Balancer (GLB)

oli.sennhauser@fromdual.com (Oli Sennhauser) — Sat, 07 Apr 2012 10:10:21 +0000

When you install a Galera Cluster for MySQL for High Availability (HA) it is not enough to install the Database Cluster to achieve this goal. You also have to make the application aware of this HA functionality. This is typically done with some kind of load balancing mechanism between the database and the application.

Linux Container with Incus for focmm unit testing

oli.sennhauser@fromdual.com (Oli Sennhauser) — Thu, 13 Oct 2022 19:11:40 +0200

Currently we are in the testing phase of the next release of our Ops Center (focmm). This testing is quite complex because it includes testing the interaction of various different components like a MariaDB Galera Cluster, a Galera Load Balancer, a virtual IP, etc.

Load Balancer on FromDual GmbH

MaxScale configuration synchronisation

Table of contents

Overview

Sharding with MariaDB MaxScale

Table of contents

Overview

MariaDB MaxScale Load Balancer with Master/Master Replication

Traffic mirroring with MariaDB MaxScale

MariaDB MaxScale Configuration Variables

Table of Contents

Global Options

Service Options

Routing Modules

ReadConnRoute

Galera Load Balancer the underestimated wallflower

Making HAProxy High Available for MySQL Galera Cluster

Galera Load Balancer Documentation

About

Unbreakable MySQL Cluster with Galera and Linux Virtual Server (LVS)

MySQL and Galera Load Balancer (GLB)

Table of Contents

Linux Container with Incus for focmm unit testing