MySQL Observability Made Simple
Coroot captures every query on the wire with eBPF and reads MySQL's own diagnostics to deliver DBA-grade coverage: query performance, locks, InnoDB internals, replication, Galera, and backups. All with a quick setup.
Complete MySQL Visibility
Coroot gives you end-to-end insight into MySQL performance. eBPF captures every query on the wire, while the server's own diagnostics explain exactly why it behaves the way it does.
Full Query Transparency
See every query shape, its latency, and success rate on the wire, with no instrumentation, no driver changes, and no overhead. Long-running queries are attributed while they are still running, not only once they finish.
Root-Cause Clarity
Internal metrics pinpoint why performance drops: a blocking transaction, an undersized buffer pool, redo log stalls, Galera flow control, or a lagging replica.
Works With Encrypted Traffic
Coroot monitors TLS connections safely by observing traffic with eBPF before encryption and after decryption.
Per-Application Insights
Understand how each service talks to MySQL: connection counts, query latency, and error rates, broken down by client application.
Correlation That Matters
Connect wire-level latency with InnoDB internals, locks, and replication for the complete performance story.
Runs Anywhere
MySQL, Percona Server, MariaDB, Percona XtraDB Cluster, Group Replication, and managed cloud databases, all with a quick, read-only setup.
Built-in Health Checks, Zero Configuration
Beyond query performance, Coroot runs the checks a seasoned DBA would, and tells you the root cause, not just that something is wrong.
Availability
Detects an unreachable instance, including a Galera node that lost quorum and stopped accepting connections entirely.
Query latency
Flags latency regressions seen on the wire and ranks the likely causes, from the heaviest query to buffer pool pressure and lock waits.
Connections
Warns before you exhaust max_connections, and catches the point where MySQL starts refusing new connections outright.
Connection errors
Detects clients failing to connect during the handshake (bad credentials, TLS errors, or timeouts) while the server still has capacity.
Replication status
Catches a stopped IO or SQL thread and reports the replica's own error text, plus network problems between replication peers.
Replication lag
Detects replicas falling behind the source, so you know before a read replica starts serving stale data.
Galera replication
Surfaces flow control pausing writes cluster-wide, certification conflicts from multi-node writes, and nodes stuck joining or without quorum.
Group Replication
Detects members that are not ONLINE, growing certification and applier queues, and a group that has lost redundancy.
Backups
Catches overdue, failed, or stale backups from the Percona XtraDB Cluster Operator, with schedule and point-in-time recovery state.
Deep insights into database performance + AI
See how applications talk to MySQL and why queries slow down. Every connection, every query, every bottleneck, visible with minimal setup.
AI Root Cause Analysis
When latency or errors spike, Coroot investigates for you: it traces the impact across your services, correlates it with what's happening inside MySQL, and explains the root cause in plain language. Here it follows timeouts from the API gateway down to InnoDB row locks, and names the exact transaction holding them.
Query Performance Analysis
Coroot collects per-query-shape statistics from performance_schema (executions, total time, and rows examined versus returned) to surface your most expensive queries and spot full scans and missing indexes. Long-running queries are attributed while they're still running, not only once they finish.
Query Tracing & Performance
eBPF captures every MySQL query on the wire with detailed tracing: a latency and errors heatmap, error rates, and individual queries per client application, all without any driver instrumentation, even over TLS. Select any region of the heatmap to pull up the exact queries behind it.
Health Checks & Instance Status
Every check runs out of the box with a sensible threshold you can tune per project or per application: availability, latency, replication status and lag, connections, connection errors, Galera, and Group Replication. Below them, each instance reports its cluster state, throughput, latency, size, and version at a glance.
Locks, Deadlocks & Long Transactions
When queries are blocked, Coroot names the transaction holding the locks, not just that contention exists. Alongside it: deadlocks and lock-wait timeouts that force your application to retry, table lock waits, and the InnoDB history list length that reveals purge lagging behind a long-running transaction.
Galera & Group Replication
Cluster replication is monitored on its own terms. Coroot detects Galera and Group Replication automatically and alerts on what actually stops writes: flow control throttling the whole cluster, a node falling behind on apply, or certification conflicts from writing the same rows on more than one node. Every alert names the affected node and what to do about it.
Inside the Engine
The InnoDB internals that quietly cause latency, and the disk growth that quietly causes outages. Coroot collects them from MySQL's own counters and turns them into answers.
InnoDB buffer pool
Size in bytes, used versus free, dirty pages, and hit rate. When the working set stops fitting, Coroot names it as the reason queries got slower, instead of leaving you to infer it from a cache-hit percentage.
Redo log & write stalls
Redo throughput and the waits that mean writes are stalling on the log buffer or the log disk. That is the difference between "MySQL is slow" and "your storage is the bottleneck".
Binary logs & undo growth
Binary logs and undo tablespaces grow independently of your table data and routinely fill a data volume. Coroot tracks both, reports them as disk growth sources, and calls out retention that is disabled or purge that is lagging behind a long transaction.
Row operations & throughput
Reads, inserts, updates, deletes, commits and rollbacks, plus sorts and temporary tables spilling to disk. These are the signals that explain a change in load.
From Database to Application
MySQL doesn't live in isolation. Coroot shows how query performance impacts the services that depend on it.
Application-Aware Monitoring
See which services issue the slowest queries and how their behavior changes over time.
Deployment Correlation
Instantly connect application releases and configuration changes with shifts in query patterns and database load.
End-to-End Performance
Trace latency from the application layer down to MySQL to understand the full user impact.
Cluster & Operator Aware
Coroot understands how your database is run. It detects Galera and Group Replication automatically, reads backup state from the Percona Operator, and handles flavour differences for you.
Percona XtraDB Cluster
Backups and point-in-time recovery state, plus cluster status, read straight from the operator's custom resources.
Galera & Group Replication
The cluster flavour is detected automatically, and each node's replication state is shown where you expect the instance status.
MySQL, Percona & MariaDB
Flavour differences are handled for you, so the same checks and charts work across MySQL, Percona Server, and MariaDB.
Works With All MySQL Deployments
From Kubernetes clusters to managed cloud databases, monitor MySQL anywhere with consistent visibility
Kubernetes
Percona XtraDB Cluster and self-managed MySQL, with automatic pod and service discovery
Cloud
Managed and self-managed MySQL on any cloud, monitored through a read-only user
Self-Hosted
On-premise instances and clusters with full observability across queries, locks, InnoDB, replication, and backups
Why MySQL Monitoring Is Usually a Pain
Traditional monitoring tools only see one side of the story: either application metrics or database internals, but never how they connect.
How Coroot Solves This
Coroot's dual-layer approach combines wire protocol capture with MySQL's own diagnostics for complete visibility.
Ready to Monitor Your MySQL Databases?
Start monitoring MySQL with zero configuration. Works with any deployment, from Kubernetes clusters to cloud managed databases.
✓ Zero configuration ✓ TLS monitoring ✓ Open source agent ✓ Prometheus compatible
Explore more of Coroot
One platform for metrics, logs, traces, and profiling. Here is what else it does.