Alerts
Page description
This page concentrates operational alert monitoring for SQL Server and Windows. It brings together the Grafana active alert queue, system restart indicators, disk free space, connection volume, transaction log growth, summarized SQL alert status, and the health of Always On or clustered failover environments.
Its goal is to speed up technical triage. It combines historical panels for recent change detection with state tables and indicators so that blocking, service issues, synchronization lag, and situations that can affect availability, capacity, or RPO and RTO targets can be identified quickly.
Variables
| Variable | Description |
|---|---|
Time range | Interval applied to historical trend panels such as uptime, connections, used log size, failover, and synchronization lag. Status tables mainly show the current situation, but they still follow the overall dashboard time context. |
Active alerts
This part opens the dashboard with the list of Grafana alerts that are firing or pending. It works as the entry point to prioritize incidents before reviewing their detail in the rest of the page.
Active alerts
- Description: Lists the active dashboard alerts and shows those in firing, error, or pending state to support quick review of what requires immediate attention.
- Panel type: alertlist
Practical recommendation: Start with the firing alerts and cross-check them against the detail panels on this same page to validate scope, severity, and likely root cause.
Server operational health
This section groups the core system and engine alerts that usually require a first operational check: recent restarts, disk capacity, connection pressure, and transaction log growth.
System Up Time
- Description: Shows the elapsed time since the last system startup for each server. A drop below 24 hours suggests a recent restart, whether caused by planned maintenance, updates, hardware issues, or an unexpected shutdown.
- Panel type: timeseries
- Units: Seconds (s)
Colors by value:
| Value | Color | Description |
|---|---|---|
COYOTE-SQL02 | Series for the second monitored server in this comparison. | |
COYOTE-SQL01 | Series for the first monitored server in this comparison. |
Practical recommendation: If the value drops abruptly, review Windows events 1074, 41, 6008, and 1001 around the restart to distinguish planned maintenance from a power issue, bugcheck, or unplanned reboot.
Disk Free Space
- Description: Shows free-space percentage by disk and server as a horizontal bar gauge. It helps identify the drives with the least remaining capacity so cleanup or expansion can be prioritized before growth affects databases, logs, or backups.
- Panel type: bargauge
- Units: Percent (%)
Thresholds:
| Value | Color | Description |
|---|---|---|
| Remaining values | Free space is critical and below the minimum recommended level. | |
| > 10 | Free space is low and cleanup or expansion should be planned. | |
| > 20 | Free space is in a comfortable operating range. |
Practical recommendation: If any drive drops below 20%, review data, log, tempdb, and backup growth, and keep enough headroom for autogrowth and maintenance operations.
User Connections
- Description: Charts the number of active user connections by server. Sustained growth above the configured threshold can reflect real workload spikes, connection pooling issues, or sessions that are not being recycled correctly.
- Panel type: timeseries
- Units: Connection count
Colors by value:
| Value | Color | Description |
|---|---|---|
COYOTE-SQL01 | Series for the first monitored server. | |
COYOTE-SQL02 | Series for the second monitored server. |
Practical recommendation: If the volume stays very high, review connection pooling, long-lived sleeping sessions, and applications that open connections without releasing them quickly.
Used Log Size
- Description: Measures the total transaction log space in use aggregated by server. When it exceeds the expected operating threshold, it can indicate long-running transactions, missing log backups, or delays in technologies that prevent log truncation.
- Panel type: timeseries
- Units: Kilobytes (KB)
Colors by value:
| Value | Color | Description |
|---|---|---|
COYOTE-SQL02 | Series for the second monitored server. | |
COYOTE-SQL01 | Series for the first monitored server. |
Practical recommendation: If the log grows quickly, validate log-backup cadence, review log_reuse_wait_desc, and look for long-running transactions, because Microsoft recommends frequent log backups to truncate the log and reduce work-loss exposure.
Global SQL alert status
This part summarizes overall SQL alert state by server and complements it with dedicated indicators for capacity, backups, blocking, and services. It is the most useful quick-look view for identifying any KO that requires deeper investigation.
Global SQL alert status
- Description: Summary table by server that indicates whether blocked sessions, long blocks, long queries, and issues with SQL Agent or SQL Mail are present. It helps quickly locate servers with active incidents or degraded services.
- Panel type: table
Colors by value:
| Column | Value | Color | Description |
|---|---|---|---|
| Blocked | OK | No blocked sessions are currently detected. | |
| KO | There are active blocking situations that need review. | ||
| Long_Blocked | OK | No prolonged blocking is detected. | |
| KO | At least one long-lasting block exists. | ||
| Long Querys | OK | No long-running queries are visible under the configured rule. | |
| KO | There are long-running queries worth analyzing. | ||
| SQL Agent Status | OK | The SQL Server Agent service is operational. | |
| KO | SQL Server Agent has an incident or is unavailable. | ||
| SQL Mail Status | OK | Database Mail is responding correctly. | |
| KO | There are send failures or mail-service issues. |
Practical recommendation: Use this table as the initial summary and investigate any KO first, prioritizing long blocks, long queries, and down services because they can directly affect availability and day-to-day operations.
Disk Free Space alert
- Description: Auxiliary indicator intended to reflect whether the low disk free-space alert condition is active within the dashboard.
- Panel type: timeseries
Practical recommendation: If this indicator becomes active, immediately review the detailed free-space panel and confirm there is enough headroom for data, logs, and backups before write failures appear.
Backup State
- Description: Auxiliary indicator intended to show backup-related alert state and warn about possible deviations in the backup strategy.
- Panel type: timeseries
Practical recommendation: Verify the date of the latest Full, Differential, and Log backup and adjust the frequency to align it with the committed RPO.
Blocked Process > 5
- Description: Alert indicator aimed at detecting blocking situations that exceed the short threshold configured in the dashboard, useful for spotting concurrency issues before they escalate.
- Panel type: timeseries
Practical recommendation: If activity appears here, identify the blocking chain as early as possible and review queries, open transactions, and repeated lock-related waits.
SQL Mail Error
- Description: Auxiliary indicator that alerts on SQL Mail or Database Mail issues, which matter because they can prevent notifications and automated outputs from being sent.
- Panel type: timeseries
Practical recommendation: Check Database Mail status, failed-message queue, and SMTP connectivity so critical notifications are not lost.
Blocked Process > 30m
- Description: Alert indicator for long-duration blocking. It highlights cases where a blocking session has been affecting other operations for too long.
- Panel type: timeseries
Practical recommendation: Treat any sustained activation as a priority incident, because long-running blocking typically ends in timeouts, accumulating waits, and visible application degradation.
SQL Agent Status
- Description: Auxiliary indicator that helps confirm whether SQL Server Agent remains available to run maintenance, backup, and automation jobs.
- Panel type: timeseries
Practical recommendation: If it becomes unavailable, review the service, its account, and recent job history, because losing Agent commonly affects backups and scheduled tasks.
High availability
This section focuses on Always On and failover health. It helps confirm that replicas keep the expected role and that synchronization lag is not compromising recovery times or read consistency on secondary replicas.
Always On replica roles
- Description: Table that shows, for each availability group, the server hosting every replica and the current role it is performing. It is useful for checking the expected configuration and quickly identifying which node is acting as
PRIMARYand which remainSECONDARY. - Panel type: table
Practical recommendation: After any maintenance or failover, confirm that the expected node is back in the PRIMARY role and that the visible secondary replicas match the intended design.
Failover
- Description: Time series of the role reported by each availability group or clustered instance, with states such as
PRIMARY,SECONDARY, orRESOLVING. It helps detect role changes, unplanned transitions, and periods of instability in the high-availability layer. - Panel type: timeseries
Practical recommendation: If you see unexpected role changes, correlate them with maintenance windows, cluster events, and replica health to confirm that an undesired failover did not occur.
Data Synchronization Lag
- Description: Shows the amount of data still waiting to be redone on secondary replicas. Sustained growth means more delay in applying changes, potentially stale read-only data on secondaries, and more time needed to complete failover.
- Panel type: timeseries
- Units: Kilobytes (KB)
Colors by value:
| Value | Color | Description |
|---|---|---|
COYOTE-AG:test | Series for the first highlighted availability group in the panel. | |
COYOTE-AG:WideWorldImporters | Series for the second highlighted availability group in the panel. |
Practical recommendation: If the queue grows persistently, compare it with your redo_rate baseline, review network or I/O saturation, and confirm that secondary read workloads are not blocking redo.
Data Synchronization Lag Time(Beta)
- Description: Estimates synchronization delay between replicas in seconds. It complements the previous panel by translating queue size into a more direct measure of lag that can impact RPO and the freshness of secondary reads.
- Panel type: timeseries
- Units: Seconds (s)
Colors by value:
| Value | Color | Description |
|---|---|---|
COYOTE-AG:test | Series for the first highlighted availability group in the panel. | |
COYOTE-AG:WideWorldImporters | Series for the second highlighted availability group in the panel. |
Practical recommendation: Use this panel as an early warning for stale secondaries and review group health if time lag increases steadily during normal workload.
Extra Alerts
This last part reserves additional alerts focused on query-specific incidents. It complements the global SQL summary to warn about errors and unusually long executions.
Error Query Alert
- Description: Additional indicator intended to warn about queries that end in error or meet the alert condition defined for execution failures.
- Panel type: timeseries
Practical recommendation: If it activates, correlate the error timestamp with the SQL Server Error Log, recent deployments, and possible connectivity or permission failures.
Long Query Alert
- Description: Additional indicator for long-running queries that deserve review because of their possible impact on CPU, I/O, blocking, and user experience.
- Panel type: timeseries
Practical recommendation: Review the most expensive queries in the period, their execution plans, and dominant waits before they turn into blocking or sustained degradation.