Skip to content

Est. 2011ยทMicrosoft Partner ID 7033487ยทDelivery under 3 minยทSupport 7 days a week

Your vault is empty.

License Error 14421

Errors 14420 and 14421: Log Shipping Backup and Restore Alerts Firing

10 min read Updated October 5, 2026 SQL Server

Fix it now

These are alerts about age, not failures of the databases. Error 14420 says the primary has not backed up its log within its backup threshold; error 14421 says the secondary is out of sync and has not restored within its restore threshold, and it prints the restored latency. Find the latency before you touch a threshold.

Run on the monitor server, or on each partner if there is no monitor

SELECT * FROM msdb.dbo.log_shipping_monitor_primary;
SELECT * FROM msdb.dbo.log_shipping_monitor_secondary;
  1. Decide which side is complaining. 14420 points at the backup job on the primary; 14421 points at the copy and restore jobs on the secondary.
  2. Open Agent job history on both servers and look at the backup, copy and restore jobs. A failing or disabled job is the commonest answer by a wide margin.
  3. Read the numbers in the alert text itself. Both messages print the threshold in minutes and how long it has actually been, and 14421 also prints the restored latency.
  4. Check the copy destination: the share must be reachable and writable by the account the copy job runs under, with room for the files.
  5. Raise a threshold only once you know the true latency, and only if the business accepts it.

The alert job runs on the monitor server instance. If the monitor is unreachable you get silence from a chain that has stopped, which is the failure mode worth designing against.

If the jobs are green and the monitor numbers are moving again, stop here. If they are green and the alert still fires, the next section explains what the alert is actually measuring.

Why it happens

Log shipping is four scheduled jobs and a set of monitor records. The backup job on the primary backs up the transaction log and writes its history locally and to the monitor. The copy job on each secondary fetches those files. The restore job on each secondary applies them. The alert job, which Microsoft documents as running on the monitor server instance, reads the recorded times and raises an alert when an operation has not happened inside its threshold.

The published text of 14420 is that the log shipping primary database has a backup threshold of so many minutes and has not performed a backup log operation for so many minutes, and to check the agent log and log shipping monitor information. The published text of 14421 is that the secondary has a restore threshold of so many minutes and is out of sync, that no restore was performed for so many minutes, and that the restored latency is so many minutes.

Read those sentences carefully, because they explain the behaviour people find confusing. The alert fires on age, not on failure. A chain that is working perfectly but running slower than its threshold produces exactly the same alert as a chain that stopped three hours ago. Only the job history separates the two.

It also means the alert depends on records being written. The monitor holds when the log was last backed up and when each secondary last copied and restored, so a monitor that cannot be reached, or a server whose clock is wrong, will either raise alerts for a healthy chain or stay quiet about a broken one. That is why Microsoft’s guidance is to put the monitor on a third instance, and why the monitor cannot be changed later without removing log shipping first.

The backup job on the primary is failing or disabled

You have this one if 14420 on its own, with an old last-backup time in the monitor record and a red backup job.

  1. Read the job step history for the real error, usually a full volume or a permission problem on the backup folder.
  2. Confirm the retention setting is not deleting files faster than the copy job collects them.
  3. Re-enable the job if it was disabled during maintenance, then run it once by hand.

If the database was switched to the simple recovery model, log backups cannot be taken at all and log shipping is broken until you set it back and take a fresh full backup.

The copy job cannot reach the share

You have this one if 14421 with backup files piling up on the primary and nothing new arriving in the secondary’s copy folder.

  1. Test the UNC path from the secondary as the account the copy job runs under, not as yourself.
  2. Grant that account read access on the source share and write access on the destination folder.
  3. Check free space on the destination volume and any firewall between the two servers.

The restore job cannot get exclusive access

You have this one if 14421 with copies arriving on time and the restore job failing or skipping, typically on a standby secondary being used for reporting.

  1. Enable the option to disconnect users when restoring, so reporting sessions do not block it.
  2. Move reporting away from the restore window, or restore with no recovery if read access is not needed.
  3. Run the restore job by hand once the blockers are gone and watch it catch up.

The volume of log exceeds the schedule

You have this one if Alerts cluster around index maintenance or bulk loads, and all the jobs are green.

  1. Stagger heavy maintenance so it does not coincide with the copy and restore schedule.
  2. Increase backup frequency so each file is smaller and the chain flows more evenly.
  3. Only then adjust the thresholds, and write down the latency the business has accepted.

The monitor or the clocks are misleading you

You have this one if Alerts for a chain that is demonstrably current, or silence from one that clearly is not.

  1. Check time synchronisation on all three roles with w32tm /query /status and correct any drift.
  2. Confirm the monitor instance is reachable and that its alert job is enabled and running.
  3. Compare the times in the monitor tables against the actual file timestamps in the backup folder.

Full reference

Which job runs where

Job Runs on Does
Backup Primary server instance Backs up the log, writes history locally and to the monitor, removes old files and history
Copy Each secondary server instance Copies backup files from the primary and writes history
Restore Each secondary server instance Restores the copied files, writes history, removes old files and history
Alert Monitor server instance Raises an alert when a backup or restore has not happened inside its threshold

Reading the alert as a sentence

Alert What it actually asserts
14420 with a failing backup job No log backup is being produced on the primary
14420 with a healthy backup job Backups run but the monitor records are not being updated, or a clock is out
14421 with a failing copy job Files are produced but never reach the secondary
14421 with a failing restore job Files arrive but cannot be applied, often a locking or state problem
14421 with everything green The chain is simply slower than the threshold you set. Read the restored latency in the message

Where the numbers come from

  • msdb.dbo.log_shipping_monitor_primary holds the primary’s last backup time as the monitor sees it.
  • msdb.dbo.log_shipping_monitor_secondary holds the last copy and restore times for each secondary.
  • The thresholds are per database and are set when you configure log shipping, not globally.
  • The monitor server records history and status for backup and restore operations and raises the alerts; it cannot be changed after configuration without removing log shipping first.

Before you raise a threshold

  1. Measure the actual latency across a normal day and a maintenance window, using the monitor tables rather than impressions.
  2. Decide what recovery point the business is agreeing to, because that is what a restore threshold encodes.
  3. Change the threshold and record the decision somewhere that outlives the person who made it.
  4. Re-check a week later that the alert now fires on genuine problems rather than on routine ones.

Raising a threshold over a chain that has actually stopped means you find out at the worst possible time that your standby is hours old. Establish which of the two you have before you touch the number.

When log shipping is the wrong shape for the requirement

Log shipping is deliberately simple: it moves files on a schedule and someone decides when to fail over. It cannot fail over on its own, and a secondary in standby mode has to be disconnected to apply the next restore, so it is a poor fit for a reporting workload that must always be available. If your alerts are really telling you that the business has outgrown a manual standby, the answer is a different feature rather than a shorter schedule.

When a licence is the actual fix

Everything above is free to fix, and log shipping runs on Enterprise and Standard alike; only Express is excluded. The edition question only arises if you decide to replace log shipping rather than repair it. In SQL Server 2025 the full availability group feature, with up to eight secondary replicas and secondaries you can read and back up, is listed as Enterprise. Standard offers basic availability groups instead, documented as two replicas with a single database, no read access on the secondary and no backups on the secondary, which is a narrower thing than most people picture. Plenty of sites should stay on log shipping and spend nothing. If yours should not, Arco can supply SQL Server 2025 Enterprise (2-core pack) licences and check the core count and edition your intended topology actually needs before you commit.

Every code this article covers

Code What it points at Source
14421 The log shipping secondary has a restore threshold in minutes, is out of sync, has not restored for that many minutes, and has the restored latency shown Microsoft Learn
14420 The log shipping primary has a backup threshold in minutes and has not performed a backup log operation within it Microsoft Learn

Confirm the fix worked

  1. All the log shipping jobs report success on their most recent run.
  2. The monitor tables show last backup, copy and restore times that are recent and moving.
  3. The latest file in the backup folder matches the latest file restored on the secondary.
  4. A full cycle runs without intervention and raises no alert.
  5. The thresholds in the configuration match the latency you actually measured.

Questions people ask about this

Can I just raise the thresholds to stop the alerts?

Sometimes that is right, but do it after measuring the real latency. Both messages print the threshold and the elapsed time, so the evidence you need is in the alert itself.

Does log shipping need a particular edition?

It is available on Enterprise and Standard and not on Express. Nothing in this article is fixed by buying anything; the edition question only arises if you replace log shipping with availability groups.

The alert fires but the secondary is up to date. Why?

Almost always a clock difference or a monitor record that is not being updated. The alert compares times recorded by different servers, so drift is enough to produce a phantom alert.

Where should the monitor live?

On a third instance where possible, so it still reports when either partner is down, which is exactly when you want to be told. Microsoft also notes that the monitor cannot be changed once configured without removing log shipping first, so decide before you build it.

What does “restored latency” in the 14421 message mean?

It is the number of minutes the secondary is behind, printed in the message alongside the threshold and the time since the last restore. It is the most useful figure in the alert and the one to quote when deciding whether the chain is broken or merely slow.

Related error codes

Was this article helpful?

Your feedback helps us improve our documentation.

Related articles

Free Fix Error 3154: The Backup Set Holds a Backup of a Different Database Free Fix SQL Server Error 18456: Read the State Number to Find the Real Cause Free Fix Setup Error 0x84B10001: SQL Server Installation Fails Before It Starts Free Fix Setup Error 0x851A0019: Could Not Find the Database Engine Startup Handle
โ† Back to Knowledge Base