Server version, edition, patch level (CU), collation, clustering/HADR and licensing. Call first in any health check.

Server time and timezone: local time, UTC time and UTC offset. Use for time-related job, backup or replication problems.

Wait statistics: the top wait types (what the server waits on: CPU, I/O, locks, memory), each as % of all waits, with average wait and signal (CPU) %. Idle waits are excluded. sampleSeconds=0 (default): since the last restart. sampleSeconds=5: only waits during a 5-second sample, the "what is slow right now" view.

Top queries by CPU, logical reads, memory grant, spills, duration or executions, from the plan cache (sys.dm_exec_query_stats), with statement text and the Plan Handle for mssql_execution_plan. topN default 25, max 100.

Execution plan (showplan XML) of a cached query, by plan handle. Reads the plan cache only; never runs the query. Get the Plan Handle from mssql_top_queries. Large plans are cut at 35,000 characters.

Running requests now: active user sessions with status, command, CPU, reads, elapsed time, wait type, blocking session, open transactions and the current statement. includeSleeping=true also lists idle sessions that hold an open transaction. For blocked sessions only, use mssql_blocking.

Blocking and locking now: blocked sessions with the blocking session, wait type, wait time and query text. An empty result means no blocking.

Memory pressure: Page Life Expectancy (PLE) from the Buffer Manager. PLE below 300 signals memory pressure; below 1000 on modern servers is worth a look.

Buffer cache hit ratio (%). Below 95% suggests too little memory for the workload.

Disk I/O latency per database file: reads, writes, average read and write stall (ms) and a rating from Exceptional to Critical. Use to find slow storage.

Disk space: drives with less than 10% free space. Any row is a warning. Include in every health check. Not available on Azure SQL Database.

TempDB health: file layout (count, equal sizes, autogrowth), space use (user objects, internal objects, version store, free space) and the top 10 sessions using tempdb. Flags too few data files, unequal sizes and percent autogrowth.

Server configuration (sp_configure, sys.configurations) with a Flag for risky values: max server memory not capped, cost threshold for parallelism at 5, MAXDOP 0, priority boost, optimize for ad hoc workloads off, xp_cmdshell on, and more. onlyFlagged=true (default) returns only flagged settings.

Error log: read the SQL Server ERRORLOG (or SQL Server Agent log) via sp_readerrorlog. Optionally filter to lines containing one or two search strings (e.g. "Backup", "3013") and select an archived log file. Returns newest entries first. Needs VIEW ANY ERROR LOG (SQL Server 2022+ / Managed Instance) or securityadmin. Not available on Azure SQL Database.

All databases on the instance: state, recovery model, compatibility level, data and log size, read-only, auto close, auto shrink, page verify, create date.

Offline databases: databases that are not ONLINE (OFFLINE, RESTORING, RECOVERING, SUSPECT, EMERGENCY). Any row needs attention.

Database settings health for every online user database: compatibility level, recovery model, page verify, auto close, auto shrink, auto statistics, target recovery time and Query Store, with Flags for risky values. onlyFlagged=true (default) returns only flagged databases.

Database owners: user databases not owned by sa (SID 0x01, so a renamed sa still counts), including orphaned owners whose login no longer exists. Not available on Azure SQL Database.

Backups: the most recent Full, Differential and Log backup per database in the last 30 days. Databases with no backup show NULL dates. For one database's history use mssql_db_backup_history. Not available on Azure SQL Database.

CHECKDB and corruption: the last known good DBCC CHECKDB date per online database (flags never, or more than 7 days ago) and the rows in msdb.dbo.suspect_pages (corruption already found). Never runs CHECKDB. Databases the login cannot read show UNKNOWN. Not available on Azure SQL Database.

VLF count: databases whose transaction log has more virtual log files than minVlfCount (default 1000), worst first. Many VLFs slow recovery, restores and log backups. Needs SQL Server 2016 SP2 or later.

Autogrowth issues: data and log files with percent growth, tiny growth (1 MB or less) or autogrowth off, largest files first. Not available on Azure SQL Database.

SQL Agent jobs: every job with its schedule, frequency, interval and last run status. Not available on Azure SQL Database.

SQL Agent failed jobs: jobs that failed in the last 7 days, with date, time and error message. Not available on Azure SQL Database.

SQL Agent alerts: checks alerts for corruption errors 823/824/825 and severities 19-25 and whether an enabled operator gets them, and lists all alerts and operators. Missing corruption alerts become CRITICAL when msdb.dbo.suspect_pages has rows. Not available on Azure SQL Database.

Security: members of the sysadmin and securityadmin server roles, each marked Expected (in expectedAccounts), Built-in (sa, NT SERVICE*, NT AUTHORITY\SYSTEM, ##...##) or REVIEW. Needs VIEW ANY DEFINITION to see all members. Not available on Azure SQL Database.

Database info for one database: state, recovery model, compatibility level, data and log size, read-only, auto close, auto shrink, page verify, create date.

Database collation: the default collation of one database. Use with mssql_db_collation_mismatches for collation conflict errors.

Collation mismatches: columns (dbo schema) whose collation differs from the database default. A common cause of collation conflict errors in joins with temp tables.

Tables overview for one database: every table with row count, size, primary key, index counts (unused, fragmented, missing), index health, read/write workload and access pattern, largest first. Reads index physical stats (LIMITED), which can be slow on very large databases; use mssql_table_info for a few tables.

Index health for one database: every non-PK index with key and included columns, usage (never used, write-only), fragmentation, fill factor advice, size and a suggested DROP, REBUILD or REORGANIZE script to review. Reads index physical stats (LIMITED), which can be slow on very large databases; use mssql_table_index_health for a few tables.

Missing indexes for one database, from the missing-index DMVs, ranked by improvement score: urgency, improvement %, uses, equality, inequality and included columns. Only tables with 1000+ rows and real query volume.

Backup history of one database in the last 7 days: type, start and finish, duration, size, compressed size and file location. Not available on Azure SQL Database.

Table keys: every user table with row count, heap or clustered index, primary key and other unique indexes. Flags heaps, tables without a primary key, and tables where only the primary key is unique. onlyFlagged=true (default) returns only flagged tables.

Foreign keys review for one database: each foreign key with trusted, disabled, indexed (unindexed foreign keys slow joins and deletes) and column type or collation mismatches, plus the WITH CHECK CHECK CONSTRAINT fix; and likely missing foreign keys found by column name (for example Votes.PostId -> Posts.Id). Name matching is a guess, so confirm each one.

Check constraints that are untrusted or disabled, with the ALTER TABLE ... WITH CHECK CHECK CONSTRAINT fix. The optimizer ignores untrusted constraints, and they may hide invalid rows. For foreign keys use mssql_db_foreign_keys.

Identity headroom: identity columns with the % of their type range already used (tinyint to bigint, decimal), most used first. WARN at 50%, RISK at 80%: when an identity runs out, every insert fails.

Column design issues: deprecated types (text, ntext, image), risky types (datetime, float, real, sql_variant) with a suggested type, and tables whose maximum row width is over 6000 bytes (near the 8060-byte row limit).

Table details for the given tables: row count, size, primary key type, index counts, read/write workload, access pattern and volatility.

Index health for the given tables: fragmentation, usage, fill factor advice and a suggested maintenance script.

Missing indexes for the given tables, ranked by improvement score.

Run one read-only SELECT (or WITH ... SELECT) query: ad-hoc analysis of system views and DMVs when no other tool fits. One statement only; INSERT, UPDATE, EXEC, SET, USE, SELECT INTO and anything else that writes or changes the session are rejected. Use databaseName instead of USE to run in another database. Returns at most 500 rows; add TOP, WHERE or GROUP BY to narrow.