// ReadClusterNameByMaster will return the cluster name for a given instance by looking at its master
// and getting it from there.
// It is a non-recursive function and so-called-recursion is performed upon periodic reading of
// instances.
func ReadClusterNameByMaster(instanceKey *InstanceKey, masterKey *InstanceKey) (string, error) {
db, err := db.OpenOrchestrator()
if err != nil {
return "", log.Errore(err)
}
var clusterName string
err = db.QueryRow(`
select
if (
cluster_name != '',
cluster_name,
ifnull(concat(max(hostname), ':', max(port)), '')
) as cluster_name
from database_instance
where hostname=? and port=?`,
masterKey.Hostname, masterKey.Port).Scan(
&clusterName,
)
if err != nil {
return "", log.Errore(err)
}
if clusterName == "" {
return fmt.Sprintf("%s:%d", instanceKey.Hostname, instanceKey.Port), nil
}
return clusterName, err
}
// ReadMaintenanceInstanceKey will return the instanceKey for active maintenance by maintenanceToken
func ReadMaintenanceInstanceKey(maintenanceToken int64) (*InstanceKey, error) {
var res *InstanceKey
query := fmt.Sprintf(`
select
hostname, port
from
database_instance_maintenance
where
database_instance_maintenance_id = %d `,
maintenanceToken)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
instanceKey, merr := NewInstanceKeyFromStrings(m.GetString("hostname"), m.GetString("port"))
if merr != nil {
return merr
}
res = instanceKey
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// ReadClusterInstances reads all instances of a given cluster
func ReadClusterInstances(clusterName string) ([](*Instance), error) {
instances := [](*Instance){}
db, err := db.OpenOrchestrator()
if err != nil {
return instances, log.Errore(err)
}
if strings.Index(clusterName, "'") >= 0 {
return instances, log.Errorf("Invalid cluster name: %s", clusterName)
}
query := fmt.Sprintf(`
select
*,
timestampdiff(second, last_checked, now()) as seconds_since_last_checked,
(last_checked <= last_seen) is true as is_last_check_valid,
timestampdiff(second, last_seen, now()) as seconds_since_last_seen
from
database_instance
where
cluster_name = '%s'
order by
hostname, port`, clusterName)
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
instance := readInstanceRow(m)
instances = append(instances, instance)
return nil
})
return instances, err
}
// ReadOutdatedInstanceKeys reads and returns keys for all instances that are not up to date (i.e.
// pre-configured time has passed since they were last cheked)
func ReadOutdatedInstanceKeys() ([]InstanceKey, error) {
res := []InstanceKey{}
query := fmt.Sprintf(`
select
hostname, port
from
database_instance
where
last_checked < now() - interval %d second`,
config.Config.InstancePollSeconds)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
instanceKey, merr := NewInstanceKeyFromStrings(m.GetString("hostname"), m.GetString("port"))
if merr != nil {
log.Errore(merr)
} else {
res = append(res, *instanceKey)
}
// We don;t return an error because we want to keep filling the outdated instances list.
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
func ReadOutdatedAgentsHosts() ([]string, error) {
res := []string{}
query := fmt.Sprintf(`
select
hostname
from
host_agent
where
IFNULL(last_checked < now() - interval %d minute, true)
`,
config.Config.AgentPollMinutes)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
hostname := m.GetString("hostname")
res = append(res, hostname)
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// ReadClustersInfo reads names of all known clusters and some aggregated info
func ReadClustersInfo() ([]ClusterInfo, error) {
clusters := []ClusterInfo{}
db, err := db.OpenOrchestrator()
if err != nil {
return clusters, log.Errore(err)
}
query := fmt.Sprintf(`
select
cluster_name,
count(*) as count_instances
from
database_instance
group by
cluster_name`)
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
clusterInfo := ClusterInfo{
ClusterName: m.GetString("cluster_name"),
CountInstances: m.GetUint("count_instances"),
}
for pattern, _ := range config.Config.ClusterNameToAlias {
if matched, _ := regexp.MatchString(pattern, clusterInfo.ClusterName); matched {
clusterInfo.ClusterAlias = config.Config.ClusterNameToAlias[pattern]
}
}
clusters = append(clusters, clusterInfo)
return nil
})
return clusters, err
}
// FailStaleSeeds marks as failed seeds where no progress have been seen recently
func FailStaleSeeds() error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
update
agent_seed
set
is_complete=1,
is_successful=0
where
is_complete=0
and (
select
max(state_timestamp) as last_state_timestamp
from
agent_seed_state
where
agent_seed.agent_seed_id = agent_seed_state.agent_seed_id
) < now() - interval ? minute`,
config.Config.StaleSeedFailMinutes,
)
return err
}
// UpdateAgentLastChecked updates the last_check timestamp in the orchestrator backed database
// for a given agent
func UpdateAgentInfo(hostname string, agent Agent) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
update
host_agent
set
last_seen = NOW(),
mysql_port = ?,
count_mysql_snapshots = ?
where
hostname = ?`,
agent.MySQLPort,
len(agent.LogicalVolumes),
hostname,
)
if err != nil {
return log.Errore(err)
}
return nil
}
// EndMaintenance will terminate an active maintenance via maintenanceToken
func EndMaintenance(maintenanceToken int64) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
res, err := sqlutils.Exec(db, `
update
database_instance_maintenance
set
maintenance_active = NULL,
end_timestamp = NOW()
where
database_instance_maintenance_id = ?
`,
maintenanceToken,
)
if err != nil {
return log.Errore(err)
}
if affected, _ := res.RowsAffected(); affected == 0 {
err = errors.New(fmt.Sprintf("Instance is not in maintenance mode; token = %+v", maintenanceToken))
} else {
// success
instanceKey, _ := ReadMaintenanceInstanceKey(maintenanceToken)
AuditOperation("end-maintenance", instanceKey, fmt.Sprintf("maintenanceToken: %d", maintenanceToken))
}
return err
}
func readAgentBasicInfo(hostname string) (Agent, string, error) {
agent := Agent{}
token := ""
query := fmt.Sprintf(`
select
hostname,
port,
token,
last_submitted,
mysql_port
from
host_agent
where
hostname = '%s'
`, hostname)
db, err := db.OpenOrchestrator()
if err != nil {
return agent, "", err
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
agent.Hostname = m.GetString("hostname")
agent.Port = m.GetInt("port")
agent.LastSubmitted = m.GetString("last_submitted")
agent.MySQLPort = m.GetInt64("mysql_port")
token = m.GetString("token")
return nil
})
if token == "" {
return agent, "", log.Errorf("Cannot get agent/token: %s", hostname)
}
return agent, token, nil
}
// ReadProblemInstances reads all instances with problems
func ReadProblemInstances() ([](*Instance), error) {
instances := [](*Instance){}
db, err := db.OpenOrchestrator()
if err != nil {
return instances, log.Errore(err)
}
query := fmt.Sprintf(`
select
*,
timestampdiff(second, last_checked, now()) as seconds_since_last_checked,
(last_checked <= last_seen) is true as is_last_check_valid,
timestampdiff(second, last_seen, now()) as seconds_since_last_seen
from
database_instance
where
(last_seen < last_checked)
or (not ifnull(timestampdiff(second, last_checked, now()) <= %d, false))
or (not slave_sql_running)
or (not slave_io_running)
or (seconds_behind_master > 10)
order by
hostname, port`, config.Config.InstancePollSeconds)
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
instance := readInstanceRow(m)
instances = append(instances, instance)
return nil
})
return instances, err
}
// ReadAgents returns a list of all known agents
func ReadCountMySQLSnapshots(hostnames []string) (map[string]int, error) {
res := make(map[string]int)
query := fmt.Sprintf(`
select
hostname,
count_mysql_snapshots
from
host_agent
where
hostname in (%s)
order by
hostname
`, sqlutils.InClauseStringValues(hostnames))
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
res[m.GetString("hostname")] = m.GetInt("count_mysql_snapshots")
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// AuditOperation creates and writes a new audit entry by given params
func AuditOperation(auditType string, instanceKey *InstanceKey, message string) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
if instanceKey == nil {
instanceKey = &InstanceKey{}
}
_, err = sqlutils.Exec(db, `
insert
into audit (
audit_timestamp, audit_type, hostname, port, message
) VALUES (
NOW(), ?, ?, ?, ?
)
`,
auditType,
instanceKey.Hostname,
instanceKey.Port,
message,
)
if err != nil {
return log.Errore(err)
}
return err
}
// BeginMaintenance will make new maintenance entry for given instanceKey.
func BeginMaintenance(instanceKey *InstanceKey, owner string, reason string) (int64, error) {
db, err := db.OpenOrchestrator()
var maintenanceToken int64 = 0
if err != nil {
return maintenanceToken, log.Errore(err)
}
res, err := sqlutils.Exec(db, `
insert ignore
into database_instance_maintenance (
hostname, port, maintenance_active, begin_timestamp, end_timestamp, owner, reason
) VALUES (
?, ?, 1, NOW(), NULL, ?, ?
)
`,
instanceKey.Hostname,
instanceKey.Port,
owner,
reason,
)
if err != nil {
return maintenanceToken, log.Errore(err)
}
if affected, _ := res.RowsAffected(); affected == 0 {
err = errors.New(fmt.Sprintf("Cannot begin maintenance for instance: %+v", instanceKey))
} else {
// success
maintenanceToken, _ = res.LastInsertId()
AuditOperation("begin-maintenance", instanceKey, fmt.Sprintf("maintenanceToken: %d, owner: %s, reason: %s", maintenanceToken, owner, reason))
}
return maintenanceToken, err
}
// submitSeedStateEntry submits a seed state: a single step in the overall seed process
func submitSeedStateEntry(seedId int64, action string, errorMessage string) (int64, error) {
db, err := db.OpenOrchestrator()
if err != nil {
return 0, log.Errore(err)
}
res, err := sqlutils.Exec(db, `
insert
into agent_seed_state (
agent_seed_id, state_timestamp, state_action, error_message
) VALUES (
?, NOW(), ?, ?
)
`,
seedId,
action,
errorMessage,
)
if err != nil {
return 0, log.Errore(err)
}
id, err := res.LastInsertId()
return id, err
}
// EndMaintenanceByInstanceKey will terminate an active maintenance using given instanceKey as hint
func EndMaintenanceByInstanceKey(instanceKey *InstanceKey) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
res, err := sqlutils.Exec(db, `
update
database_instance_maintenance
set
maintenance_active = NULL,
end_timestamp = NOW()
where
hostname = ?
and port = ?
and maintenance_active = 1
`,
instanceKey.Hostname,
instanceKey.Port,
)
if err != nil {
return log.Errore(err)
}
if affected, _ := res.RowsAffected(); affected == 0 {
err = errors.New(fmt.Sprintf("Instance is not in maintenance mode: %+v", instanceKey))
} else {
// success
AuditOperation("end-maintenance", instanceKey, "")
}
return err
}
// ForgetLongUnseenInstances will remove entries of all instacnes that have long since been last seen.
func ForgetLongUnseenAgents() error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
delete
from host_agent
where
last_submitted < NOW() - interval ? hour`,
config.Config.UnseenAgentForgetHours,
)
return err
}
// ForgetLongUnseenInstances will remove entries of all instacnes that have long since been last seen.
func ForgetLongUnseenInstances() error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
delete
from database_instance
where
last_seen < NOW() - interval ? hour`,
config.Config.UnseenInstanceForgetHours,
)
AuditOperation("forget-unseen", nil, "")
return err
}
// ReadActiveMaintenance returns the list of currently active maintenance entries
func ReadActiveMaintenance() ([]Maintenance, error) {
res := []Maintenance{}
query := fmt.Sprintf(`
select
database_instance_maintenance_id,
hostname,
port,
begin_timestamp,
timestampdiff(second, begin_timestamp, now()) as seconds_elapsed,
maintenance_active,
owner,
reason
from
database_instance_maintenance
where
maintenance_active = 1
order by
database_instance_maintenance_id
`)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
maintenance := Maintenance{}
maintenance.MaintenanceId = m.GetUint("database_instance_maintenance_id")
maintenance.Key.Hostname = m.GetString("hostname")
maintenance.Key.Port = m.GetInt("port")
maintenance.BeginTimestamp = m.GetString("begin_timestamp")
maintenance.SecondsElapsed = m.GetUint("seconds_elapsed")
maintenance.IsActive = m.GetBool("maintenance_active")
maintenance.Owner = m.GetString("owner")
maintenance.Reason = m.GetString("reason")
res = append(res, maintenance)
return err
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// ForgetInstance removes an instance entry from the orchestrator backed database.
// It may be auto-rediscovered through topology or requested for discovery by multiple means.
func ForgetInstance(instanceKey *InstanceKey) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
delete
from database_instance
where
hostname = ? and port = ?`,
instanceKey.Hostname,
instanceKey.Port,
)
AuditOperation("forget", instanceKey, "")
return err
}
// SearchInstances reads all instances qualifying for some searchString
func SearchInstances(searchString string) ([](*Instance), error) {
instances := [](*Instance){}
db, err := db.OpenOrchestrator()
if err != nil {
return instances, log.Errore(err)
}
if strings.Index(searchString, "'") >= 0 {
return instances, log.Errorf("Invalid searchString: %s", searchString)
}
query := fmt.Sprintf(`
select
*,
timestampdiff(second, last_checked, now()) as seconds_since_last_checked,
(last_checked <= last_seen) is true as is_last_check_valid,
timestampdiff(second, last_seen, now()) as seconds_since_last_seen
from
database_instance
where
hostname like '%%%s%%'
or cluster_name like '%%%s%%'
or server_id = '%s'
or version like '%%%s%%'
or port = '%s'
or concat(hostname, ':', port) like '%%%s%%'
order by
cluster_name,
hostname, port`, searchString, searchString, searchString, searchString, searchString, searchString)
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
instance := readInstanceRow(m)
instances = append(instances, instance)
return nil
})
if err != nil {
return instances, log.Errore(err)
}
err = PopulateInstancesAgents(instances)
if err != nil {
return instances, log.Errore(err)
}
return instances, err
}
// readSeeds
func readSeeds(whereCondition string, limit string) ([]SeedOperation, error) {
res := []SeedOperation{}
query := fmt.Sprintf(`
select
agent_seed_id,
target_hostname,
source_hostname,
start_timestamp,
end_timestamp,
is_complete,
is_successful
from
agent_seed
%s
order by
agent_seed_id desc
%s
`, whereCondition, limit)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
seedOperation := SeedOperation{}
seedOperation.SeedId = m.GetInt64("agent_seed_id")
seedOperation.TargetHostname = m.GetString("target_hostname")
seedOperation.SourceHostname = m.GetString("source_hostname")
seedOperation.StartTimestamp = m.GetString("start_timestamp")
seedOperation.EndTimestamp = m.GetString("end_timestamp")
seedOperation.IsComplete = m.GetBool("is_complete")
seedOperation.IsSuccessful = m.GetBool("is_successful")
res = append(res, seedOperation)
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// ReadRecentAudit returns a list of audit entries order chronologically descending, using page number.
func ReadRecentAudit(page int) ([]Audit, error) {
res := []Audit{}
query := fmt.Sprintf(`
select
audit_id,
audit_timestamp,
audit_type,
hostname,
port,
message
from
audit
order by
audit_timestamp desc
limit %d
offset %d
`, config.Config.AuditPageSize, page*config.Config.AuditPageSize)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
audit := Audit{}
audit.AuditId = m.GetInt64("audit_id")
audit.AuditTimestamp = m.GetString("audit_timestamp")
audit.AuditType = m.GetString("audit_type")
audit.AuditInstanceKey.Hostname = m.GetString("hostname")
audit.AuditInstanceKey.Port = m.GetInt("port")
audit.Message = m.GetString("message")
res = append(res, audit)
return err
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// UpdateAgentLastChecked updates the last_check timestamp in the orchestrator backed database
// for a given agent
func UpdateAgentLastChecked(hostname string) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
update
host_agent
set
last_checked = NOW()
where
hostname = ?`,
hostname,
)
if err != nil {
return log.Errore(err)
}
return nil
}
// SeedOperationState reads states for a given seed operation
func ReadSeedStates(seedId int64) ([]SeedOperationState, error) {
res := []SeedOperationState{}
query := fmt.Sprintf(`
select
agent_seed_state_id,
agent_seed_id,
state_timestamp,
state_action,
error_message
from
agent_seed_state
where
agent_seed_id = %d
order by
agent_seed_state_id desc
`, seedId)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
seedState := SeedOperationState{}
seedState.SeedStateId = m.GetInt64("agent_seed_state_id")
seedState.SeedId = m.GetInt64("agent_seed_id")
seedState.StateTimestamp = m.GetString("state_timestamp")
seedState.Action = m.GetString("state_action")
seedState.ErrorMessage = m.GetString("error_message")
res = append(res, seedState)
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// updateSeedStateEntry updates seed step state
func updateSeedStateEntry(seedStateId int64, reason error) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
update
agent_seed_state
set error_message = ?
where
agent_seed_state_id = ?
`,
reason.Error(),
seedStateId,
)
if err != nil {
return log.Errore(err)
}
return reason
}
// ReadAgents returns a list of all known agents
func ReadAgents() ([]Agent, error) {
res := []Agent{}
query := `
select
hostname,
port,
token,
last_submitted,
mysql_port
from
host_agent
order by
hostname
`
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
agent := Agent{}
agent.Hostname = m.GetString("hostname")
agent.Port = m.GetInt("port")
agent.MySQLPort = m.GetInt64("mysql_port")
agent.Token = ""
agent.LastSubmitted = m.GetString("last_submitted")
res = append(res, agent)
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
func getHostAttributesByClause(whereClause string) ([]HostAttributes, error) {
res := []HostAttributes{}
query := fmt.Sprintf(`
select
hostname,
attribute_name,
attribute_value,
submit_timestamp ,
ifnull(expire_timestamp, '') as expire_timestamp
from
host_attributes
%s
order by
hostname, attribute_name
`, whereClause)
db, err := db.OpenOrchestrator()
if err != nil {
goto Cleanup
}
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
hostAttributes := HostAttributes{}
hostAttributes.Hostname = m.GetString("hostname")
hostAttributes.AttributeName = m.GetString("attribute_name")
hostAttributes.AttributeValue = m.GetString("attribute_value")
hostAttributes.SubmitTimestamp = m.GetString("submit_timestamp")
hostAttributes.ExpireTimestamp = m.GetString("expire_timestamp")
res = append(res, hostAttributes)
return nil
})
Cleanup:
if err != nil {
log.Errore(err)
}
return res, err
}
// ReadClusters reads names of all known clusters
func ReadClusters() ([]string, error) {
clusterNames := []string{}
db, err := db.OpenOrchestrator()
if err != nil {
return clusterNames, log.Errore(err)
}
query := fmt.Sprintf(`
select
cluster_name
from
database_instance
group by
cluster_name`)
err = sqlutils.QueryRowsMap(db, query, func(m sqlutils.RowMap) error {
clusterNames = append(clusterNames, m.GetString("cluster_name"))
return nil
})
return clusterNames, err
}
// UpdateInstanceLastChecked updates the last_check timestamp in the orchestrator backed database
// for a given instance
func UpdateInstanceLastChecked(instanceKey *InstanceKey) error {
db, err := db.OpenOrchestrator()
if err != nil {
return log.Errore(err)
}
_, err = sqlutils.Exec(db, `
update
database_instance
set
last_checked = NOW()
where
hostname = ?
and port = ?`,
instanceKey.Hostname,
instanceKey.Port,
)
if err != nil {
return log.Errore(err)
}
return nil
}
