What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
If an agent’s SQLite state database keeps reporting corruption, first stop every process that can write to it and preserve the database together with any -wal, -shm, or -journal files. Diagnose a copy, not the original. A crash during an ordinary transaction is normally handled by SQLite’s automatic recovery; recurring corruption warrants checking how the database is copied, restored, synchronized, and accessed, as well as the SQLite runtime version.
What an SQLite corruption report does—and does not—tell you
SQLite returns SQLITE_CORRUPT when it detects damage to the database’s structure, format, or control elements. That identifies a problem SQLite found in the file; it does not identify what caused it.
As an Amazon Associate I earn from qualifying purchases.
An application crash or power failure during a transaction does not, by itself, mean the database should remain corrupt. SQLite uses transaction journals to roll back incomplete work when the database is next opened. Persistent errors therefore call for investigation rather than an assumption that SQLite failed to handle a routine crash.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchSQLite’s own corruption guidance emphasizes that it must be able to see journal files to recover after a crash or power failure. In WAL mode, committed transactions may still reside in the write-ahead log. The main database and its sidecars must be treated as related state, not as interchangeable files that can safely be separated or deleted.
#1 Best Overall
What can make corruption recur?
Copying or restoring only the main database file
A filesystem copy of the main database made while writes are active may capture an inconsistent mix of old and new content. If a write failed, omitting its rollback journal or WAL can also discard recovery information. Moving, deleting, or pairing a database with the wrong sidecar can disrupt recovery. These are common ways backup and restore workflows can turn a recoverable interruption into lost work or a damaged database.
Writes that bypass SQLite, or unsafe concurrency
SQLite protects its transactions through its own locking and coordination. Another process or thread that writes bytes directly to the database file bypasses those protections. Record which processes and threads open the file, whether they share the same path, and whether any backup, sync, restore, or cleanup task touches it while the agent is running.
Rank #2
For WAL databases, every connection must use WAL correctly. SQLite documents a rare WAL-reset bug that can cause corruption when WAL mode is in use, two or more connections access the same file, and writes or checkpoints overlap. SQLite reports that the bug was likely present in versions 3.7.0 through 3.51.2, was fixed in 3.51.3 and later, and was backported to 3.44.6 and 3.50.7. The bug was reported on March 3, 2026; check the runtime actually used by the agent, not just a separately installed command-line program. The documented trigger is unusual and requires the specified concurrency conditions.
Free tools Windows power users keep installed
One-click scans. No signup required.
Storage, operating-system behavior, or disabled synchronization
SQLite’s corruption guidance also identifies unreliable storage or operating-system behavior and disabled sync operations as risks. In particular, PRAGMA synchronous=OFF omits sync operations and permits write reordering, which can lead to corruption after a power loss or hard reset. Do not use it as a supposedly safe speed setting for persistent agent state. SQLite documents FULL as its default synchronous setting for maximum reliability.
Rank #3
Application or runtime changes
Compare the first known failure with agent upgrades, changes to the SQLite library or runtime build, edits to journal-mode or synchronous settings, and changes to backup or restore code. These facts do not prove a cause, but they help distinguish a new application behavior from a storage or file-handling problem. Record the journal mode and SQLite runtime version from the process that opens the database.
How to preserve and diagnose the database safely
- Stop writers. Stop the agent and any other process that may access or modify the database. Do not begin by deleting a WAL, journal, or shared-memory file.
- Preserve the complete file set. Make a byte-for-byte working copy of the database and all accompanying SQLite sidecars, and keep the original untouched. If the agent is using WAL mode, the
-walfile may contain committed transactions; separating it from the database can lose those transactions or cause corruption. - Record the circumstances. Note the SQLite runtime version, journal mode, filesystem and storage context, recent agent upgrades, backup or restore actions, and which processes or threads use the same file. These details narrow the investigation; none alone establishes the cause.
- Check a copy. Open the copied database with the SQLite command-line shell and run
PRAGMA integrity_check;. For a quicker but less thorough check, runPRAGMA quick_check;. For example:sqlite3 copied.db "PRAGMA integrity_check;". Save the output and any opening errors. - Keep the result in perspective. An integrity check reports detected problems; it does not explain how they happened. Do not run recovery commands against the only original or treat a clean check as proof that the agent’s application-level state is correct.
How to recover data when the database is damaged
Restore a known-good backup when one exists
SQLite’s FAQ recommends recovering from a backup if one is available. Restore it as a separate candidate, then check that it opens and that the agent’s expected schema and critical state are present before switching production to it. A backup is useful only if it is a consistent copy and its contents are acceptable for the recovery point it represents.
Use .recover for best-effort salvage
If no suitable backup exists, or normal extraction fails, SQLite’s command-line shell provides .recover. It scans salvageable content and emits SQL that can be used to construct a separate database:
sqlite3 corrupt.db .recover >data.sql
sqlite3 recovered.db <data.sqlRun this against a preserved copy. SQLite describes recovery as a salvage undertaking, not a guarantee of restoration. Rows may be missing, altered, moved, or resurrected from pages that appear free; recovered tables can also violate constraints. Unassociated content may be placed in a lost_and_found table. The optional --ignore-freelist argument avoids scanning pages that appear free, reducing the chance of bringing previously deleted content back into the output.
Best Value
Validate the rebuilt database before using it
- Run an integrity check on the rebuilt file and retain its output.
- Compare the recovered schema and row counts with known-good records, backups, or application expectations.
- Inspect critical agent state and check whether constraints or relationships are violated.
- Test the agent against a copy of the recovered database before replacing production state.
Recovery output can be defective and may need review or rework. If exact state matters, a salvage result without a verified backup is not proof that every committed record has been restored correctly.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to make consistent backups while the agent is running
SQLite identifies three ways to create a consistent live copy. Choose based on how the application is built and how the copy needs to be transferred.
| Method | How it works | Use when |
|---|---|---|
| Online Backup API | SQLite API creates a snapshot as of the start of the copy operation; copying can be incremental while other users continue working. | The application can call SQLite’s backup interface and needs a consistent copy during ongoing use. |
VACUUM INTO |
SQLite creates a separate, vacuumed database copy. | A SQL command can be run through a connection to the database and a separate copy is needed. |
sqlite3_rsync |
SQLite’s utility copies a database over SSH using a bandwidth-efficient protocol. SQLite lists it as available beginning with version 3.47.0, released October 21, 2024. | The backup workflow needs remote transfer and the utility is available in the deployed SQLite version. |
A raw filesystem copy is appropriate only when the database is quiescent—no writes are in progress. When copying after a failed write, preserve the associated rollback journal or WAL with the database. Keep backups separate from live state and periodically verify that they open and can be restored. An external drive can serve as a backup destination, but the destination does not make an inconsistent live copy safe.
Quick Recap
What to check before closing the incident
- Identify the exact SQLite library version used by the agent and whether it uses WAL mode.
- Confirm that all connections use SQLite normally and that no process manipulates the database or its sidecars independently.
- Review backup, synchronization, and restore procedures for live main-file-only copies or sidecar deletion.
- Check synchronous settings and the reliability of the storage and operating-system path.
- Document whether recovery came from a verified backup or from best-effort salvage, and validate the resulting state before returning the agent to service.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
