AceMQ supports MariaDB 24/7 across Americas, EMEA and APAC on Community and Enterprise Server, Galera clusters, MaxScale and replicated estates. Most calls start with one of three problems: a Galera node that cannot rejoin, a cluster that dropped to non-Primary, or queries that slowed after a move to 11.x.
AceMQ provides MariaDB support for production estates: Galera nodes that will not rejoin after an SST, a cluster that dropped to non-Primary during a network partition, replicas that lag through the nightly batch, and query plans that changed after the move to 11.x. Community or Enterprise Server, on premises, in the cloud or on Kubernetes. Every ticket reaches a named senior engineer.
Support quotes answered within 1 hour, by a senior engineer.
Seven failures from real Galera and replication estates, each with the status variable or log line a MariaDB engineer reads first.
Common MariaDB production problems and how AceMQ resolves them
Symptom you're seeing
What we check first
A joiner repeats SST and never finishes
Check SHOW STATUS LIKE 'wsrep_local_state_comment', the donor's wsrep_ log lines and gcache.size against how long a node can be away. A gcache too small for the outage forces a full state transfer instead of an incremental one.
The cluster reports non-Primary after a partition
Check wsrep_cluster_status and wsrep_cluster_size on every node. We decide which component holds the most recent state before bootstrapping, because choosing the wrong node loses transactions.
Flow control throttles all writers
Check wsrep_flow_control_paused and wsrep_local_recv_queue_avg. One slow applier node, often with a different disk or wsrep_slave_threads, throttles the entire cluster.
Large writes stall a Galera cluster
Check wsrep_max_ws_size and the transaction size. Multi-million-row deletes certify as one writeset, so we batch them.
MaxScale routes reads to a stale replica
Check maxctrl list servers for state and replication lag, then max_slave_replication_lag and causal_reads on the read-write split router.
GTID errors after promoting a replica
Check gtid_slave_pos, gtid_binlog_pos and gtid_current_pos on both servers, plus whether log_slave_updates is set. Promotion without matching positions leaves replicas unable to start.
11.x optimizer changes slow a query that was fast on 10.x
Compare EXPLAIN FORMAT=JSON on both versions and check optimizer_switch and the cost model changes before touching indexes.
After each P1 you get a write-up: which node held the latest state, what broke the cluster, and the setting that changes.
Galera is non-Primary and the application is down? Request a quote now and we reply within one hour.
Which MariaDB deployments does AceMQ support, and who covers what?
Deployment
What AceMQ covers
What the vendor covers
Excluded
Community Server
Incidents, replication, Galera, tuning, upgrades
Nothing: MariaDB's support page does not include Community Server
Defects needing a change to MariaDB's code
Enterprise Server
Operations beside the MariaDB ticket
MariaDB subscription support for Enterprise Server and MaxScale
Enterprise-only defects
Galera Cluster
Quorum, state transfer, flow control, network design
MariaDB support applies where the Enterprise subscription covers the nodes; Galera not named on its page
Network faults outside your estate
MaxScale
Routing, failover, monitor settings
Covered by MariaDB's support per its FAQ
Defects needing a vendor patch
Amazon RDS for MariaDB
Parameter groups, queries, replication design
AWS runs the service
AWS service faults
Support options
MariaDB support or independent support?
MariaDB Standard and Premium support
AceMQ independent support
Hours24x7 for urgent issues on both levels
Hours24/7 for all severities by plan
First responseStandard: urgent 30 minutes, high 2 hours, standard 4 hours, low 8 hours. Premium: urgent 15 minutes, high 1 hour, standard 4 hours, low 8 hours
First responseFrom 15 minutes, by plan
ProductsEnterprise Server in the Enterprise Platform, Plus and Ultimate tiers; MaxScale per its FAQ; Community Server not included
ProductsCommunity and Enterprise Server, Galera, MaxScale
ChannelsPhone and live chat on Premium only
ChannelsPhone, email and shared channel
PricingNo public price; request a quote
Pricing$24,000 to $72,000 a year (USD), by SLA, scope and deployment
What's Included
Everything in your MariaDB support contract
One contract spans Community and Enterprise Server, Galera, MaxScale and the replicas behind them, so an incident never stalls on whose component it is.
01
Emergency Incident Response
Cluster without quorum, a node that will not rejoin, replication broken, or a disk filling with undo and binary logs. With response targets from 15 minutes, by plan, a senior engineer joins a live bridge with access to diagnose, not a ticket acknowledgement.
02
Root Cause Analysis
Every P1 closes with a written RCA: what failed, why, the fix applied, and the topology, schema or configuration change that stops it happening again. Delivered as standard, not on request.
03
Galera Cluster & Replication
Quorum design and arbitrators, gcache sizing for IST, SST method selection, flow control tuning, GTID domains and parallel replication, and MaxScale routing and failover behavior under load.
04
Query & InnoDB Tuning
Slow query log and ANALYZE FORMAT=JSON on your real workload, index design, engine-independent statistics and histograms, buffer pool and redo sizing, and purge lag from long-running transactions.
05
CVE & Patch Advisory
Alerts for CVEs affecting your exact MariaDB build, with tested upgrade paths. For 10.4, 10.5 and 10.6, which are past community end of life, AceMQ's OSSeva platform provides patched builds of the series you already run.
06
Upgrades & Migrations
Rolling upgrades from 10.x to 10.11, 11.4, 11.8 or 12.3 LTS, run node by node on Galera with mariadb-upgrade, plus migrations between MySQL and MariaDB and between self-managed servers and managed services, rehearsed with a rollback plan.
Why AceMQ
What a named MariaDB engineer changes
Named Engineers, Zero Cold Start
The same senior engineers stay on your account. They know your Galera topology, your replication chains and your batch windows, so a P1 call starts with a hypothesis instead of twenty minutes of context transfer.
No Tier-1 Triage Layer
You reach a senior MariaDB engineer directly by phone, email or Slack. Nobody collects information to hand off, and there is no escalation approval between you and the person who can read a wsrep status dump.
MariaDB Is Not MySQL With a New Name
GTID format, Galera, MaxScale, storage engines such as Aria and MyRocks, the JSON type and the 11.x optimizer all differ from MySQL. Advice written for MySQL is often wrong on MariaDB, and we know where the two diverged.
Galera Problems Are Usually Topology Problems
Split brain, endless SSTs and cluster-wide stalls tend to come from node count, gcache size and one slow disk rather than from Galera itself. We fix the incident, then the layout that caused it.
Genuine Follow-the-Sun Coverage
Round-the-clock coverage for customers in 26+ countries. Your 3am quorum loss is someone's mid-afternoon: no overnight skeleton crew, no waiting for a region to come online.
Full-Stack, Not Just mariadbd
We diagnose across the connection pool, MaxScale, the ORM's generated SQL, storage latency, kernel settings and Kubernetes. MariaDB symptoms often have causes well outside the database process.
Which MariaDB setups we cover
Galera nodes in one data center or stretched across three, MaxScale in front, Amazon RDS for MariaDB, and hosts that cannot reach the internet.
MariaDB Community Server 10.4 through 13.x
MariaDB Enterprise Server
MariaDB Galera Cluster
MariaDB MaxScale
Amazon RDS for MariaDB
Asynchronous and semi-synchronous replication
Kubernetes & OpenShift operators
RHEL, Debian & Ubuntu packages
Bare metal & on-premise
Air-gapped / no outbound access
Hybrid cloud
Mixed MySQL and MariaDB estates
Version coverage
MariaDB versions we support
MariaDB versions we support
Release line
What we do on it
13.x rolling releases
24/7 support
12.3
24/7 support, recommended upgrade target
11.8 and 11.4
24/7 support
10.11
24/7 support and upgrade planning
10.6, 10.5 and 10.4
24/7 support and upgrade projects to 11.x or 12.x
Escalation Path
Your first hour of a MariaDB outage
Quorum loss at 3am and a slow report at 3pm are different problems, so each severity gets its own clock.
From 15 min (by plan)first responseEmergencyProduction down: cluster non-Primary, no node will rejoin, replication stopped, or the data disk full of binary logs
1 hourfirst responseCriticalSevere degradation: flow control throttling every writer, endless SSTs, MaxScale sending reads to a stale replica
4 hoursfirst responseHighGTID position mismatches after a promotion, replicas lagging through batch jobs, slow queries after an 11.x upgrade
Next Dayfirst responseStandardQuestions on gcache sizing, wsrep settings, upgrade sequencing and MySQL-to-MariaDB compatibility
Clocks run by plan and severity: a cluster without quorum and a tuning question are never queued together.
Prefer a named MariaDB engineer who already knows your topology? Ask for a quote or write to us.
MariaDB support usually costs $24,000–72,000 a year. Galera node count, MaxScale use, how many replicas you run and the response plan you pick move a quote within that range.
All prices are in US dollars (USD). Response targets start at 15 minutes by plan, 24/7, and quote requests get a reply within 1 hour. Covering several technologies under one contract earns a multi-technology discount.
Support In Practice
Galera and upgrade incidents we have closed
MariaDB support work from AceMQ. Each card links to the full write-up: the Galera or replication estate, what AceMQ diagnosed or changed, and the result.
Support
MariaDB Galera Quorum Loss and Split-Brain Support
European Telecommunications Operator
Recovering a four-node Galera cluster after a site-link failure left both halves non-Primary, and redesigning it so one link fault cannot do that again.
AceMQ offers MariaDB support in four tiers, each handing more of the day-to-day work to AceMQ: 24/7 support and incident management for Galera clusters and replicas your team runs, proactive monitoring that finds gcache, quorum and lag problems first, managed MariaDB on your own infrastructure, and fully managed MariaDB on AceMQ infrastructure. MariaDB 10.4 to 13.x is supported, and 10.4 to 10.6 can be kept patched through OSSeva.
Your team runs itAceMQ runs it
Tier 1
MariaDB 24/7 Support & Incident Management
You run the cluster, we take the hard calls
Best for: In-house DBAs who want a Galera specialist for incidents they cannot settle alone.
Tier 2
MariaDB 24/7 Support with Proactive Monitoring
Scheduled reviews of topology, gcache and replicas
Best for: Teams who would rather find a small gcache or a lagging replica in a review than during an outage.
Tier 3
Managed MariaDB on Your Infrastructure
We operate the nodes inside your network
Best for: Organizations whose MariaDB servers must stay on their own hardware, in their own data center or cloud account.
Tier 4
Fully Managed MariaDB on AceMQ Infrastructure
We run your MariaDB nodes and backups for you
Best for: Teams that want the database working and the on-call rota handled elsewhere.
One partner for MariaDB
Everything else AceMQ does for MariaDB.
Next steps for MariaDB teams: replica and Galera health reviews, MySQL migration paths and the wider database practice.
MariaDB, under an Enterprise subscription with Standard and Premium support levels, and independent providers like AceMQ. Community Server is outside MariaDB's support offer, so independent support is the common choice.
How is supporting MariaDB different from supporting MySQL?
Galera, MaxScale, different GTID semantics and a diverging optimizer. A MySQL runbook will mislead on a MariaDB cluster.
Our Galera node will not rejoin. What now?
Check state, donor choice and gcache first, then decide between incremental and full transfer. We avoid bootstrapping from the wrong node.
What breaks in a 10.x to 11.x upgrade?
Optimizer and cost model behaviour, client authentication and TLS defaults, and replication assumptions. We rehearse on a replica with real traffic.
Can you move us between MySQL and MariaDB?
Yes, in either direction, after a compatibility test of SQL, replication format and client libraries.
Does MaxScale need separate support?
MariaDB covers MaxScale under its subscription. We support it either way, including failover and split-read tuning.
How fast do you respond to a MariaDB outage?
Response starts within 15 minutes on the top plans, 24/7. On a Sev 1 such as a Galera cluster stuck in non-Primary, an engineer joins a bridge, compares wsrep_cluster_status and the latest committed state per node, and picks the one to bootstrap. P2 is 1 hour, P3 4 hours, P4 next business day.
Can we use AceMQ for a single MariaDB project?
Yes. A one-off piece of work such as a Galera recovery, a 10.x to 11.x upgrade or a MySQL migration is quoted separately from the annual agreement. Teams that outgrow a single project usually move to the yearly plan, which adds 24/7 cover and named engineers.
What drives the cost of MariaDB support?
Expect $24,000 to $72,000 a year (USD). The quote moves with the number of Galera nodes and MaxScale instances, whether you run Community or Enterprise Server, how many replicas and sites you have, and the response plan you choose. Send your topology and we reply with a quote within 1 hour.
Let's talk
Your Galera Cluster Shouldn't Depend on One Person's Memory
Describe your cluster layout and version. We reply with a quote within 1 hour.
A senior MariaDB engineer, not a sales queueQuote requests answered within 1 hourAd hoc engagements or an annual partnership