SQL Server Health Check
Use Health Check to see if one SQL Server instance is in good condition, and what to do about each problem.
Before you begin
- You need the View SQL Server and View HealthCheck permissions. To open Log Watch, you also need Manage databases.
- Scops runs the check every 30 minutes, and the page reads the newest result every minute. To run the check now, click the refresh button at the top.
- When you show one database, Backups, Storage and Databases show only that database. Server and Always On show the instance.
Read the checklist
The cards at the top open the areas: Overview, Backups, Storage, Databases, Server and, with availability groups, Always On. A card has the color of its worst problem.
Overview shows each check once, in one of three sections:
| Section | The checks |
|---|---|
| Needs attention | Critical problems and warnings, the most urgent first. |
| Could be better | Not urgent: space and speed to win back. |
| Healthy | The checks that passed, by area, and the checks that Scops could not read. |
Click a check to open it.
Read Why it matters and What to do. If a command shows, copy it.
To see all rows of the area, click the link at the bottom, for example Open Backups.
In an area, open a row for the same advice and its detail. Since shows how long the problem has lasted.
Fix an area that Scops cannot read
When the login of Scops cannot read an area, the area shows Not read and why. If a permission is missing, it also shows a script.
Click Copy the script.
Run the script on the instance as a member of
sysadmin.
The next check reads the area. See SQL Server permissions.
Check the backups
Backups shows the newest full, differential and log backup of each database. Open a row to see the last 7 days and the recent backups, with their size and destination.
| Problem | State |
|---|---|
| The newest full or differential backup is older than 24 hours. | Warning |
| It is older than 2 days, and there is no current log backup. | Critical |
| The log backups are current, but the newest full or differential backup is older than 8 days. | Warning |
| The database never had a full backup. | Warning |
| In FULL or BULK_LOGGED recovery, the log was never backed up, or not in the last 24 hours. | Warning |
For a database in an availability group, these are notes only, because another replica can take its backups. The same applies to model and to a database that is not online.
Check the storage
Volumes shows each volume that holds database files. It is amber with less than 10% free, red with less than 5%.
All databases shows the data and log use of each database. Open a row to see its history and its files, with their growth, maximum size and volume.
Fills In shows when a volume or a database becomes full, from the growth of the last 8 days. The bar shows the days left on a one-year scale. It is red within 14 days, amber within 45.
| Problem | State |
|---|---|
| A transaction log is 80% full or more. | Warning. Critical from 90% if the log cannot grow. |
| A filegroup is 90% used, and none of its files can grow. | Warning |
| tempdb has one data file on 4 or more processors, or data files of different sizes. | Warning |
| A file of 1 GB or more grows by a percent. | Could be better |
Check the databases
All databases shows the state, recovery model, compatibility level, page verify and last integrity check of each database. Open a row to see its settings.
| Problem | State |
|---|---|
| The database is SUSPECT, RECOVERY_PENDING or in EMERGENCY. | Critical |
| The last DBCC CHECKDB without errors is older than 8 days, or there is none. | Warning |
| AUTO_CLOSE is on, Query Store records nothing new, or the owner is not a login of the instance. | Warning |
| PAGE_VERIFY is not CHECKSUM, AUTO_SHRINK is on, or the log has 1,000 virtual log files or more. | Could be better |
A database that is OFFLINE, RESTORING or RECOVERING is not a problem. Scops does not check it further.
Check the server
| Card | What it checks |
|---|---|
| Agent jobs | The last run of each job. It warns when SQL Server Agent stopped while jobs are enabled, or when a last run failed. It also warns when a run takes twice its usual time and 30 minutes more. |
| Error log | The last 24 hours of the error log, by kind. See Read the error log. |
| Settings | The settings that matter, with the recommended value. priority boost on is critical, and lightweight pooling on is a warning. Other differences, such as an unlimited max server memory, are Could be better. |
Read the error log
The Error log card counts the lines of each kind in the last 24 hours. Open a kind to read its newest lines. A line under the table names the kinds that Scops did not find. To add them to the table, click show them.
| Kind | State |
|---|---|
| I/O errors 823 and 824, stack dumps, databases marked suspect, and the other critical kinds of Log Watch | Critical |
| Reads that needed a retry (825), I/O taking longer than 15 seconds, out of memory (701, 802), and 100 or more failed logins | Warning |
The card shows what Log Watch found. If Log Watch is off, or cannot read the error log, the card shows Not read and why.
Open Log Watch
On the Error log card, click Open Log Watch. Or, on the Databases page, right-click the instance and select Log Watch.
Select the period, from Now (the last 30 minutes) to Last 7 days. The default is Last 24 hours.
Under Logs, click a file: Current or an archive. The files written before the period are under Older files.
Under Problems, open a problem to read what it means, what to do and its last lines.
To find a word in the problems, type in Search. To read a file, click View Log. The Live Log window shows its newest 100, 500, 1000 or 2000 lines. You can refresh the current file every 5, 15 or 60 seconds.
Settings, at the bottom of the window, shows how Log Watch works:
- It reads each SQL Server instance every 5 minutes by default, from where the last reading ended. The first reading goes back 24 hours.
- Scops keeps the findings for 30 days.
- You turn it on or off in Settings › Jobs › SQL Server Log Watch.
- The login needs EXECUTE on
sys.xp_readerrorlog, with its user inmaster. If the grant is missing, copy the script there.
Check the availability groups
- Availability groups: the primary, the replicas, the state and the listener of each group.
- Replicas: the role, availability mode, failover mode, state, queues and lag of each replica. Open a row to see its databases.
- Lag: how far the replicas were behind, and the send and redo queues, over 1H, 1D, 1W or 1M.
- Failovers: the failovers of the period.
| Problem | State |
|---|---|
| A replica is disconnected, or a database is NOT_HEALTHY. | Critical |
| A replica is not healthy or partly healthy, or the data movement of a database is suspended. | Warning |
| A database on a synchronous-commit replica is not SYNCHRONIZED, or a send or redo queue is 1 GB or more. | Warning |
A secondary replica does not know the state of the other replicas. To see them, open the Health Check of the primary.
Related
Updated November 22, 2026