Home / Repstance User Guide / MONITORING WITH AMAZON CLOUDWATCH / Metrics Reference

Metrics Reference

The function of Install the Metrics Function reads the answer of the status command (see Status of any Capture or Apply Processes) and writes the values below to CloudWatch every minute, in the namespace Repstance.

Dimensions

Metrics of	Dimensions
A Capture Process or an Apply Process	Instance (the parameter InstanceName of the stack), ProcessType (capture or apply) and ProcessId (the id of the process, as text)
The server	Instance

The metrics of a process

Metric	Unit	Meaning	Field of the status answer
Running	None	1 when the last command is RUN and its status is In Process: the process runs. 0 for a process that is stopped, has failed or has only been initialised.	Command, CommandStatus
Failed	None	1 when the status of the command is Failed.	CommandStatus
CaughtUp	None	1 when the process runs and waits for transactions: there is nothing to process at the moment.	CommandStatus, StatusDetails
Loading	None	1 while the process runs the Initial Load.	StatusDetails, Message
LagSeconds	Seconds	The time between the last change in the Source Database and the last change that the process has handled. Only while the process runs.	LastDBLSNTime, CheckpointLSNTime
IdleSeconds	Seconds	The seconds since the activity of the process last changed.	LastActivityTime
TransactionsTotal	Count	The transactions handled since the process was started.	CNTTX
InsertsTotal	Count	The inserts handled since the process was started.	CNTI
UpdatesTotal	Count	The updates.	CNTU
DeletesTotal	Count	The deletes.	CNTD
DDLTotal	Count	The DDL statements.	CNTDDL
LoadedRowsTotal	Count	The rows loaded by the Initial Load since the process was started.	CNTLOBJREC

The counters are totals since the start of the current run of the process, like the numbers of repcli: they start again at 0 when the process is run again. Use the speed instead of the total: the function of CloudWatch RATE gives the change per second. After a restart of the process it gives one negative value.

The metrics of the server

Metric	Unit	Meaning
ApiUp	None	1 when the status command was answered, 0 when it was not.
Processes	Count	The number of the processes that the answer contains.

Every metric is a custom metric of CloudWatch, which is charged per metric and month (see Monitoring with Amazon CloudWatch): a pair of processes has 24, a server 2.

Statistics and periods

The function writes one value per metric and minute, so with a period of 60 seconds every statistic gives the same number. For longer periods use Minimum for Running and ApiUp (the state is bad if it was bad once), Maximum for Failed, LagSeconds and the counters.

An example

Whether the Apply Process 1 has run, for the last 15 minutes in steps of five minutes. The times are UTC; Minimum is 0.0 when the process did not run for the whole step:

aws cloudwatch get-metric-statistics --namespace Repstance --metric-name Running \
--dimensions Name=Instance,Value=prod Name=ProcessType,Value=apply Name=ProcessId,Value=1 \
--start-time 2026-10-22T16:46:00Z --end-time 2026-10-22T17:01:00Z --period 300 --statistics Minimum

{
"Label": "Running",
"Datapoints": [
{
"Timestamp": "2026-10-22T16:56:00+00:00",
"Minimum": 0.0,
"Unit": "None"
},
{
"Timestamp": "2026-10-22T16:51:00+00:00",
"Minimum": 0.0,
"Unit": "None"
},
{
"Timestamp": "2026-10-22T16:46:00+00:00",
"Minimum": 0.0,
"Unit": "None"
}
]
}

A record that the function writes (embedded metric format), here of an Apply Process that is stopped. CloudWatch takes the metrics from the part _aws; the other fields stay in the log and help to find the cause. While a process runs, its record also has LagSeconds, and during an Initial Load a Message that names the table that is loaded, for example Table Name: SOE.ORDERS:

{
"_aws": {
"Timestamp": 1792687684367,
"CloudWatchMetrics": [{
"Namespace": "Repstance",
"Dimensions": [["Instance", "ProcessType", "ProcessId"]],
"Metrics": [
{"Name": "Running", "Unit": "None"}, {"Name": "Failed", "Unit": "None"}, {"Name": "CaughtUp", "Unit": "None"},
{"Name": "Loading", "Unit": "None"}, {"Name": "IdleSeconds", "Unit": "Seconds"}, {"Name": "TransactionsTotal", "Unit": "Count"},
{"Name": "InsertsTotal", "Unit": "Count"}, {"Name": "UpdatesTotal", "Unit": "Count"}, {"Name": "DeletesTotal", "Unit": "Count"},
{"Name": "DDLTotal", "Unit": "Count"}, {"Name": "LoadedRowsTotal", "Unit": "Count"}
]
}]
},
"Instance": "prod",
"ProcessType": "apply",
"ProcessId": "622",
"ProcessName": "tz622",
"Command": "INITIALISED",
"CommandStatus": "Completed",
"StatusDetails": "Waiting for the Next Command",
"Running": 0,
"Failed": 0,
"CaughtUp": 0,
"Loading": 0,
"IdleSeconds": 5767.957,
"TransactionsTotal": 0,
"InsertsTotal": 0,
"UpdatesTotal": 0,
"DeletesTotal": 0,
"DDLTotal": 0,
"LoadedRowsTotal": 0
}

CloudWatch Alarms

TROUBLESHOOTING AND FAQ