ALM Exception Monitoring

Retrieves exception count metrics from SAP Cloud ALM Exception Monitoring (EXM). Publishes time-series metrics and evaluates alarms 
per configured row.

• API: GET /api/calm-metrics/v1/metrics?provider=exm

Requires a Web Service connector with authentication type CLOUD_ALM.

SAP Cloud ALM Connector

Based on a service key from the SAP Cloud ALM API service instance in the BTP subaccount.

Required OAuth scopes in the authorities list of the instance parameters:

• $XSMASTERAPPNAME.calm-api.exm.read
• $XSMASTERAPPNAME.calm-api.metrics.read

• Publishes six metric types per EXM datapoint: exm.counter exm.counter_available exm.counter_disruption exm.counter_degradation exm.counter_maintenance exm.counter_unknown
• Time window is a 5-minute aligned UTC window with a 1-minute delay: to = floor(now - 1min, 5min), from = to - 5min (format yyyyMMddHHmmss UTC)
• Paginated fetching up to 5000 records per page using x-total-count response header
• All 6 measures are gauge-to-count converted (non-monotonic delta sum) before storage
• Per-row thresholds with standard G2W:80 W2M:90 syntax. Multi Threshold Syntax
• Attributes Filter narrows which datapoints a row evaluates
• Exclusive controls whether a matched datapoint is consumed or passed to later rows
• Glob support for Metric and Service name fields (* = all)
• Optional alarm tag for grouping or routing
• Auto-clear when alarm condition no longer matches
• Load Services button: auto-discovers EXM service IDs and names from the live tenant

Data collection populates Service name and Service ID from the Cloud ALM EXM service registry.

Always run data collection before adding rows. A service name not returned by data collection may not match what the metrics API returns.

Data collection runs one call:

1. POST /api/calm-analytics/v1/analytics/providers/filters with body {“providerName”:“EXM_DATAPROVIDER”,“providerVersion”:“v1”} — returns the EXM service filter list. Finds the entry with key = serviceId and extracts each service UUID and its label (e.g. S4H.100).

Click Load Services to run data collection and populate the surveillance table.

1. Open monitor configuration
2. Click Load Services
3. Table populates with Service name and Service ID from the live tenant
4. Enable rows, set thresholds, save

1. Set Metric = * and/or Service name = * to match all
2. Use only service names returned by data collection

Note: metric names in the Metric field use the snake_case form shown above (e.g. exm.counter_disruption, not exm.counterDisruption).

Narrows which datapoints a row matches. All clauses must match (AND). Matching is case insensitive. Malformed clauses are silently ignored.

Example filter value:

categoryName:ABAP_SHORT_DUMP,serviceType:S4

1. No rows configured: nothing is published. Add at least one active row to collect data.
2. Service name: glob matched against EXM service label. S4H* matches S4H.100. * matches all.
3. Metric: exact match or *. Partial globs do NOT work.
4. Attributes Filter: all clauses must match. Empty = match all datapoints.
5. Exclusive = true: datapoint consumed by first matching row. Later rows skip it.
6. Metric = false: row evaluates alarms but publishes no metric datapoints.

Base key: promonitor.cloud_alm.exm.*

Tags published with each datapoint: service.name sap.service.name service.namespace service.instance.id sap.service.display_name plus datapoint-level tags serviceId serviceName categoryName serviceType useCase

All metrics are stored as non-monotonic delta counts (converted from gauge before storage).

Alarms use a suppression key to deduplicate. Format:

• With category: {monitorId}_{connectorId}_alm_exm_{service}_{metric}_{categoryName}_{rowIdx}
• Without category: {monitorId}_{connectorId}_alm_exm_{service}_{metric}_{rowIdx}

One alarm per unique key. A new value overwrites the previous alarm state for the same key.

The time window is computed at collection time using 5-minute alignment with a 1-minute delay:

adjustedNow = UTC.now() - 1 minute
minuteFloor = floor(adjustedNow.minute / 5) * 5
to   = adjustedNow with minute=minuteFloor, second=0, nanosecond=0
from = to - 5 minutes

Example at 14:37 UTC: from=20260601143000 to=20260601143500

This ensures the query always covers a complete 5-minute window that the EXM backend has already finalized.

Set Service name = * and Metric = *. An empty table sends nothing.

Alarms as soon as 1 exception appears. G2W:1 means any non-zero value triggers warning.

Only datapoints for S4H.100 and category ABAP_SHORT_DUMP are evaluated.

Alarms when any exception occurs during a Disruption period.

Metric = false: alarms fire but no datapoints written to time-series.

Each metric type gets its own alarm tag. Exclusive rows prevent cross-matching.

Row 1 consumes S4H.100 datapoints (Exclusive = true). Row 2 applies a looser threshold to all other services.

Row 1 consumes informational exceptions (Exclusive = true, Alarm = false). Row 2 alarms on all others.

• Metric field does not support partial glob patterns. *disruption does NOT match exm.counter_disruption. Use exact values or *.
• Service ID is auto-populated by data collection. Manual entry is possible if the UUID is known.
• Malformed Attributes Filter clauses are silently ignored. Validate format before saving.
• The time window is always the last complete 5-minute slot. Data more recent than 1 minute is not included.
• Exclusive = true means first matching row wins. Order rows from most specific to most general.
• Thresholds apply to the raw numeric count value of the selected metric.