Skip to content

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

Your vault is empty.

Free Fix Event ID 231

Event ID 231: a passive database copy failed and now needs a reseed

9 min read Updated October 5, 2026 Exchange Server

Fix it now

A database copy hit an error and stopped rather than replaying past something it could not trust. The database it was protecting is now less redundant than you think. Find the error first, then reseed: reseeding onto the cause produces a second failed copy and a wasted night.

Run these in the Exchange Management Shell once the cause is dealt with

Get-MailboxDatabaseCopyStatus DB01 | Format-List Name,Status,CopyQueueLength,ReplayQueueLength,ContentIndexState,ErrorMessage
Suspend-MailboxDatabaseCopy -Identity DB01\SERVER2
Update-MailboxDatabaseCopy -Identity DB01\SERVER2 -DeleteExistingFiles -SourceServer SERVER3
Resume-MailboxDatabaseCopy -Identity DB01\SERVER2
  1. Read the event on your own server and note the database, the copy and the error it recorded. Microsoft publishes no description for these entries, so those fields are the diagnosis.
  2. Check the System log on that member for storage errors, check free space on its database and log volumes, and check what the antivirus agent is doing there.
  3. Look in the MailboxDatabaseFailureItems channel, which Microsoft documents as holding events associated with failures affecting a replicated mailbox database.
  4. If only the content index failed, seed the catalogue alone with Update-MailboxDatabaseCopy -CatalogOnly instead of rebuilding the whole copy.
  5. Do not activate a copy that has just failed. Let it become healthy and let the queues drain first.

Microsoft requires a copy to be suspended before it can be updated, and -DeleteExistingFiles removes only the log files it checks for: it fails if other files are present at the target, so database files there must be removed by hand.

If the copy is healthy with queues near zero, you are done. If it fails again, the next section covers the four things that usually stop a copy.

Why it happens

Continuous replication is a two-stage pipeline. Log generations are shipped from the server hosting the active copy, then replayed into the passive database. If replay hits an error, the safest thing the service can do is stop, because replaying past a fault would build a copy that is quietly different from the one it is meant to mirror. A suspended copy is therefore a working safety mechanism, not a bug, and the useful question is what it stopped on.

A reseed copies the entire database across the network, so it is worth choosing the source deliberately. Microsoft documents -SourceServer as naming a Mailbox server with a passive copy to use as the source, which keeps the load off the server your users are connected to. -BeginSeed starts the operation asynchronously and exits, which is useful in a script, and -ManualResume leaves you to resume replication yourself when you are ready.

The search catalogue is worth treating as a separate question. -CatalogOnly seeds the content index catalogue for the copy on its own, which is much cheaper than rebuilding the database copy. Microsoft notes that catalogue seeding always occurs over the MAPI network, even if you name a DAG network with -Network, so plan the bandwidth on that basis rather than on the replication network’s.

The passive server hit a storage or IO error

You have this one if The System log on that member carries disk or storage port errors close to the failure.

  1. Deal with the storage first. Reseeding onto a disk that is failing produces a second failed copy.
  2. Check for storage timeout and reset entries in the System log, and run the vendor’s diagnostics against the disks holding that copy.
  3. Once the storage is sound, suspend, reseed with -DeleteExistingFiles, and resume.
  4. Watch the copy and replay queues until they settle near zero.

The copy fell far enough behind that the logs it needed were truncated

You have this one if The copy was suspended or disconnected for a long period and cannot resume, because the source no longer holds the generations it wants.

  1. Compare the copy queue length with the generations still present on the source before deciding.
  2. Reseed rather than trying to resume. There is nothing left to catch up to.
  3. Fix whatever left the copy suspended for that long, which is usually a maintenance window nobody closed.

A suspended copy blocks log truncation on the active server while it is suspended, and becomes unrecoverable once truncation finally happens. It is not a harmless state to leave running.

Only the content index failed

You have this one if The database copy status is healthy while the content index state is not.

  1. Seed the catalogue alone: Update-MailboxDatabaseCopy -Identity DB01\SERVER2 -CatalogOnly.
  2. Expect it to use the MAPI network regardless of any DAG network you name, because Microsoft documents catalogue seeding as always doing so.
  3. Confirm the search services are running on the member and that the volume holding the index has free space.

The volume filled, or something interfered with the files

You have this one if The failure coincides with a full disk, an antivirus event, or a backup window on that member.

  1. Check free space on both the database and log volumes for that copy.
  2. Apply Microsoft’s published Exchange exclusions on the passive member as well as the active one. Passive copies are routinely forgotten.
  3. Confirm the backup agent is not reading the passive database files directly.
  4. Reseed once the interference has stopped, not before.

Full reference

The reseed switches, as documented

Switch What Microsoft says it does
-DeleteExistingFiles Removes the existing log files at the target location. It removes only the files it checks for and fails if other files are present; database-related files must be removed manually
-SourceServer The Mailbox server with a passive copy of the database to use as the source for the seed
-CatalogOnly Seeds only the content index catalogue for the database copy
-BeginSeed Starts the seeding operation asynchronously and exits, which is useful for scripting reseeds
-ManualResume Does not automatically resume replication on the copy; you resume it yourself
-Network Specifies which DAG network to use for seeding. Content index catalogue seeding always uses the MAPI network regardless

A reseed with -DeleteExistingFiles removes files at the target before it starts. Make sure you are reseeding the copy you think you are, and that a healthy copy exists elsewhere. Because the switch fails when it finds files it did not expect, a half-cleaned target is a common way to have a reseed stop before it has begun.

Where the explanation actually is

Microsoft publishes no descriptions for the store and replication event IDs this article covers, so the numbers are labels rather than answers. The explanation is in the crimson channels: MailboxDatabaseFailureItems logs events associated with any failure affecting a replicated mailbox database, and HighAvailability logs the replication service’s own components, Active Manager role monitoring and database action events such as mounts and log truncation. Both sit under Applications and Services Logs, then Microsoft, then Exchange.

Before you reseed

  1. Confirm the copy is genuinely the problem, and not a symptom of a cluster or network fault affecting every copy.
  2. Find the error. A copy that failed on storage will fail again on the same storage.
  3. Suspend the copy. Microsoft requires this before Update-MailboxDatabaseCopy will run.
  4. Choose a passive source with -SourceServer so the active server keeps serving users.
  5. Decide whether the whole copy needs rebuilding or only the catalogue, and use -CatalogOnly if it is the latter.
  6. Resume, then watch the queues until they settle.

After it is healthy

  • Confirm log truncation has resumed on the active server by watching the log folder over the next backup cycle. A copy that was suspended for a long time was also pinning logs.
  • Run Test-ReplicationHealth on the member, which checks the replication pipeline, Active Manager and the underlying cluster together.
  • Use CollectOverMetrics.ps1 from the Exchange Scripts folder if you want to know how often this database has moved or failed over recently; it reads member event logs and writes one CSV per DAG.
  • If the same copy fails repeatedly, stop reseeding it and deal with the member instead.

Every code this article covers

Code What it points at Source
Event ID 231 No description is published by Microsoft. Read the database, copy and error your own entry names, then confirm the state with Get-MailboxDatabaseCopyStatus not published by the vendor
Event ID 234 No description is published by Microsoft. Read it alongside the previous entry for the state the copy was left in not published by the vendor
Event ID 218 No description is published by Microsoft. Match it by database and timestamp against the MailboxDatabaseFailureItems channel not published by the vendor
Event ID 9877 No description is published by Microsoft. Check the content index state for the copy directly rather than inferring it from the event number not published by the vendor

Confirm the fix worked

  1. Get-MailboxDatabaseCopyStatus shows the copy healthy with copy and replay queues near zero.
  2. The content index state for that copy is healthy.
  3. Test-ReplicationHealth passes on the member.
  4. Log truncation has resumed on the active server over the next backup cycle.

Questions people ask about this

Does this cost anything to fix?

No. Reseeding is a built-in operation and costs you time and bandwidth rather than money. If the underlying cause is a failing disk then the disk costs money, but nothing about Exchange licensing enters into it.

Why is there no official description of this event?

Microsoft does not publish one. That is why the useful evidence is the MailboxDatabaseFailureItems channel and the copy’s own ErrorMessage rather than the event number.

How long does a reseed take?

It depends entirely on the size of the database and the speed of the path between the two servers, so measure it on your own kit. Start with a small database if you want a rate to extrapolate from.

Can I reseed while the database is in use?

Yes. The active copy stays mounted and users stay connected. The load falls on the source server and the network, which is why naming a passive source with -SourceServer is worth doing.

Do I have to suspend the copy first?

Yes. Microsoft states you must suspend a database copy before you can update it with Update-MailboxDatabaseCopy.

Related error codes

Was this article helpful?

Your feedback helps us improve our documentation.

Related articles

Free Fix HTTP 500 on Microsoft-Server-ActiveSync: every mobile device stops syncing License Error MigrationPermanentException: the move failed in a way retrying cannot fix License Error HCW8054 and HCW8039: hybrid organisation and address space configuration fails Free Fix HTTP 404.0 on Autodiscover: Outlook cannot retrieve its own configuration
โ† Back to Knowledge Base