ScyllaDB
To test the ScyllaDB Health Check endpoint, use the following curl command:
curl http://localhost:PORT/health
Health Function
The Health
function checks the health of the ScyllaDB Cluster by pinging
the Coordinator Node. It
returns a simple map containing a health message.
Functionality
Ping ScyllaDB Server: The function pings the ScyllaDB through server to check its availability.
- If the ping fails, it logs the error and terminates the program.
- If the ping succeeds, it returns a health message indicating that the server with some .
Sample Output
The Health
returns a JSON-like map structure with a single key indicating the health status:
{
"message": "It's healthy",
"status": "up",
"scylla_cluster_nodes_up": "3",
"scylla_cluster_nodes_down": "0",
"scylla_cluster_size": "1",
"scylla_current_datacenter": "datacenter1",
"scylla_current_time": "2024-11-04 22:59:21.69 +0000 UTC",
"scylla_health_check_duration": "16.896976ms",
"scylla_keyspaces": "6"
}
ScyllaDB Setup
Before starting the cluster, ensure the fs.aio-max-nr value is
sufficient (e.g. 1048576
or 2097152
or more).
If you prefer to configure it manually, run one of the following commands to check the current value:
sysctl --all | grep --word-regexp -- 'aio-max-nr'
sysctl fs.aio-max-nr
cat /proc/sys/fs/aio-max-nr
If the value is lower than required, you can use one of these commands:
# Update config non-persistent
sysctl --write fs.aio-max-nr=1048576
Here's some links for more relevant information and automation:
- Repository: gvieira/ws-scylla - Simple ScyllaDB Cluster management with Makefiles
- ScyllaDB University: 101 Essentials Track - Learn the base concepts of ScyllaDB
Code implementation
Here you can check how the Health Check is done under the hood:
func (s *service) Health() map[string]string {
ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
defer cancel()
stats := make(map[string]string)
// Check ScyllaDB health and populate the stats map
startedAt := time.Now()
// Execute a simple query to check connectivity
query := "SELECT now() FROM system.local"
iter := s.Session.Query(query).WithContext(ctx).Iter()
var currentTime time.Time
if !iter.Scan(¤tTime) {
if err := iter.Close(); err != nil {
stats["status"] = "down"
stats["message"] = fmt.Sprintf("Failed to execute query: %v", err)
return stats
}
}
if err := iter.Close(); err != nil {
stats["status"] = "down"
stats["message"] = fmt.Sprintf("Error during query execution: %v", err)
return stats
}
// ScyllaDB is up
stats["status"] = "up"
stats["message"] = "It's healthy"
stats["scylla_current_time"] = currentTime.String()
// Retrieve cluster information
// Get keyspace information
getKeyspacesQuery := "SELECT keyspace_name FROM system_schema.keyspaces"
keyspacesIterator := s.Session.Query(getKeyspacesQuery).Iter()
stats["scylla_keyspaces"] = strconv.Itoa(keyspacesIterator.NumRows())
if err := keyspacesIterator.Close(); err != nil {
log.Fatalf("Failed to close keyspaces iterator: %v", err)
}
// Get cluster information
var currentDatacenter string
var currentHostStatus bool
var clusterNodesUp uint
var clusterNodesDown uint
var clusterSize uint
clusterNodesIterator := s.Session.Query("SELECT dc, up FROM system.cluster_status").Iter()
for clusterNodesIterator.Scan(¤tDatacenter, ¤tHostStatus) {
clusterSize++
if currentHostStatus {
clusterNodesUp++
} else {
clusterNodesDown++
}
}
if err := clusterNodesIterator.Close(); err != nil {
log.Fatalf("Failed to close cluster nodes iterator: %v", err)
}
stats["scylla_cluster_size"] = strconv.Itoa(int(clusterSize))
stats["scylla_cluster_nodes_up"] = strconv.Itoa(int(clusterNodesUp))
stats["scylla_cluster_nodes_down"] = strconv.Itoa(int(clusterNodesDown))
stats["scylla_current_datacenter"] = currentDatacenter
// Calculate the time taken to perform the health check
stats["scylla_health_check_duration"] = time.Since(startedAt).String()
return stats
}
Note
Scylladb does not support advanced health check functions like SQL databases or Redis.
The current implementation is based on queries at system
related keyspaces.