If your hybrid identity sync is slow or throwing errors, or you’ve inherited a Microsoft Entra Connect Sync server (still better known as Azure AD Connect) and want to know whether it’s healthy, check the version before anything else. In the hybrid environment I look after, it’s the first thing I look at whenever sync misbehaves, because it rules out a whole category of problems in a couple of minutes.
Microsoft set a minimum version of 2.5.79.0, released in May 2025 with a back-end service change that hardens the service. From 30 September 2026, every Connect Sync server below that version stopped synchronising, and it stays broken until the installation is upgraded. If sync has stopped recently and you’re trying to work out why, this is the first thing to rule in or out.
Being on 2.5.79.0 isn’t the end of it, either. Microsoft retires older releases on a rolling basis, and the Microsoft Entra Connect version release history lists 2.5.79.0’s own end of support as 23 October 2026. The release history is the page to check for what’s current.
The rest of this article covers how to check your version properly, how to work out what the sync engine is actually doing, which filtering and rule changes are worth making and which ones will bite you, and whether you should still be running Connect Sync at all.
Check your Entra Connect Sync version first
Before changing the scheduler, filtering objects or modifying sync rules, find out what you’re actually running. The simplest way is Control Panel > Programs and Features on the Connect Sync server, looking for Microsoft Entra Connect Sync. You can also check from PowerShell, which works whichever product name the installed version uses:
powershell
Get-ItemProperty "HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\*" |
Where-Object {
$_.DisplayName -like "*Entra Connect*" -or
$_.DisplayName -like "*Azure AD Connect*"
} |
Select-Object DisplayName, DisplayVersionIt’s also worth checking the version recorded in the sync configuration. The installed application version and the configured version can differ after an interrupted or incomplete upgrade:
powershell
Import-Module ADSync
(Get-ADSyncGlobalSettings).Parameters |
Where-Object { $_.Name -eq "Microsoft.Synchronize.ServerConfigurationVersion" } |
Select-Object Name, ValueIf the two don’t match, resolve that before doing anything else.
If you’re below 2.5.79.0, sync has already stopped and the upgrade is urgent. Even if you’re at or just above it, I wouldn’t stop at the minimum. Upgrade to the latest release available for your environment. The release history also notes that version 2.6.79.0 was recalled after release, so if you installed that one, move to the latest version too.
Upgrading properly
The installation file is now only available through the Microsoft Entra admin centre rather than the old Download Centre, and current versions require .NET Framework 4.7.2 and TLS 1.2. On an old server, those prerequisites are often the real blocker rather than the installer itself.
Microsoft’s upgrade guidance describes two approaches: an in-place upgrade, or a swing migration where you build a new server alongside the old one and switch over. Microsoft recommends the swing migration if you only have one server or you’re coming from an old version, and it’s the safer of the two. During an upgrade the normal delta sync scheduler is suspended, but password hash sync carries on.
Treat the upgrade as a normal production change:
- Check the current version, both installed and configured.
- Export or document the current sync configuration.
- Confirm whether you have a staging server you can use.
- Upgrade the staging server first if you have one, or build the swing server.
- Upgrade or switch the active server.
- Confirm that the installed and configured versions match.
- Verify that sync resumes and the Operations tab shows clean runs.
Don’t make unrelated sync rule changes at the same time. If something goes wrong, you want to know what caused it.
Should you still be using Entra Connect Sync?
This is the other question I’d ask before spending time tuning an existing installation. Microsoft Entra Cloud Sync replaces the full sync engine on your own server with a lightweight provisioning agent, and manages the configuration from Microsoft Entra ID. Microsoft’s Connect to Cloud Sync decision guide describes Cloud Sync as its strategic direction and says new sync capabilities are focused there.
That doesn’t mean every Connect Sync deployment should be ripped out. The same guide shows Connect Sync still covers things Cloud Sync doesn’t, such as the full advanced sync rule engine, while Cloud Sync adds things like group provisioning back to AD and on-demand provisioning. Treat Cloud Sync as something to evaluate rather than an automatic replacement for a working installation. The sensible order is:
- Make sure your existing Connect Sync installation is supported.
- Understand what it currently does.
- Document any custom sync rules and filtering.
- Check whether Cloud Sync supports those requirements, using the Cloud Sync overview and its comparison table.
- Decide whether migration makes sense.
That’s a much safer approach than changing a working identity platform simply because a newer product exists.
Understand what the sync engine is actually doing
Connect Sync takes objects from on-premises Active Directory, processes them through its sync rules and exports the resulting changes to Microsoft Entra ID. At a high level, it imports changes from AD, processes the objects through the sync engine, applies the sync rules and attribute flows, and exports changes to Entra ID.
It isn’t simply copying every AD attribute into the cloud. The sync rules decide which objects are in scope, how objects are joined and which attributes flow between the connected systems. That matters as soon as you start trying to optimise the environment.
Check the scheduler before changing it
Connect Sync normally runs a sync cycle every 30 minutes. Check the current configuration with:
powershell
Get-ADSyncSchedulerThe output shows whether the scheduler is enabled, the configured interval, when the next cycle is due, whether a sync is running now, and details of the last cycle.
For most environments I’d leave the scheduler at its default. You’ll often see a 15-minute interval suggested as a way to get changes into Entra ID faster, but that won’t work, because 30 minutes is the supported minimum. If you have a legitimate reason to sync less often, you can increase the interval. This example sets it to one hour:
powershell
Set-ADSyncScheduler -CustomizedSyncCycleInterval 01:00:00For most organisations, though, the scheduler isn’t where I’d start when looking for performance improvements.
Run a delta sync when you actually need one
If you’ve just created or changed an object and don’t want to wait for the next scheduled cycle, trigger a delta sync:
powershell
Start-ADSyncSyncCycle -PolicyType DeltaThis is the command for the normal “I changed something in AD and need it in Entra ID” situation. A delta cycle processes changes since the previous sync rather than reprocessing the whole directory. Run Get-ADSyncScheduler first, and if a sync is already running, don’t keep firing off more commands. Find out what’s happening first.
Be very careful with an initial sync
You’ll sometimes see this recommended for troubleshooting:
powershell
Start-ADSyncSyncCycle -PolicyType InitialThis isn’t simply a bigger delta sync. An initial sync makes the engine do the full processing normally needed after configuration changes, and on a large environment it can take considerably longer than a delta cycle. I wouldn’t run it just because a user hasn’t appeared in Entra ID yet. Find out why the object isn’t syncing first. If you’ve changed sync rules or scope, an initial sync may genuinely be required, but that’s a different situation.
Reduce the number of objects in scope
One of the more useful optimisations is making sure you only sync objects that actually need to exist in Entra ID. If the directory holds years of legacy objects, there’s little benefit in pushing every one of them into the cloud. Connect Sync supports domain filtering, OU filtering, group-based filtering for pilots, and rule-based filtering. In a normal production environment, OU and domain filtering are usually the simplest places to start.
For example, you might have:
OU=Users
OU=Service Accounts
OU=Computers
OU=Legacy
OU=DisabledIf Legacy contains objects that have no business in Entra ID, excluding that OU is better than syncing everything and trying to clean it up afterwards.
Be very careful when removing objects from scope
This is where “optimising” can create a much bigger problem. If an object is already synced and you remove it from scope, Connect Sync can process that as a deletion from the cloud directory, with knock-on effects on group memberships and access.
Don’t make a large OU filtering change on a production server just because an OU holds objects you don’t think you need. Before changing scope, work out which objects are currently synced, which will leave scope, what they’re used for, whether they’re members of synced groups, whether applications depend on them, and what exports the change will generate. This is where a staging server earns its keep, because you can see what the engine intends to do before it becomes a production export.
Don’t edit the built-in sync rules
You’ll often hear that disabling unnecessary attribute flows will reduce sync overhead. Be careful with that advice. Connect Sync contains built-in rules that Microsoft can update between versions, and you shouldn’t modify them directly. If you need to customise attribute flow or object processing, create custom rules in the Synchronization Rules Editor instead.
This matters for more than supportability. Imagine you modify one of Microsoft’s standard rules because you don’t need a particular attribute. Six months later you upgrade, and Microsoft has changed that rule to support a new feature or fix a problem. Your modified copy won’t necessarily pick up that change. If you have custom rules, document them, and review them before every upgrade rather than discovering afterwards that a rule from three years ago is behind the behaviour you’re seeing.
Staging mode is worth having
If identity sync matters to your organisation, a second Connect Sync server in staging mode is worth considering. A staging server imports and processes sync data but doesn’t export changes to Entra ID. That gives you a safety net for upgrades, configuration and rule changes, and recovery from a failed primary server.
A staging server isn’t load balancing. Its value is in high availability and controlled change management, and in letting you validate significant changes before they reach the active server. For filtering and rule changes in particular, that’s valuable, because a scope mistake can remove objects from Entra ID.
Check sync health properly
When someone says “Entra Connect is broken”, several different things could actually be wrong. Start with Get-ADSyncScheduler, then open the Synchronization Service Manager on the server. The Operations tab is the most useful view, because it shows the actual import, sync and export runs. Look for failed runs, big jumps in processing time, unusually large numbers of adds or deletes, export errors, objects that fail repeatedly, and connectors that haven’t completed successfully.
Get-ADSyncConnectorRunStatus tells you whether a connector operation is running, but it isn’t a general health check. For troubleshooting, use the scheduler, the Synchronization Service Manager and the actual connector errors together.
The errors I’d investigate first
Duplicate proxyAddresses
Duplicate SMTP addresses are a common reason objects fail to sync. If two AD objects both contain SMTP:user@company.com, Entra ID can’t represent both with the same unique address. Find the conflicting objects in Active Directory before you go anywhere near the sync rules.
Duplicate or incorrect UPNs
The userPrincipalName also needs to make sense in Entra ID. A user might exist in AD as user@company.local when the intended cloud sign-in is user@company.com. The fix isn’t necessarily in Connect Sync. Often the underlying AD attribute is simply wrong.
Objects outside the sync scope
If a user isn’t appearing in Entra ID, check whether the object is actually inside the configured scope. OU filtering is the obvious place to look, and the Synchronization Service Manager can show whether the object exists in the connector space and whether it has been filtered.
Objects stuck in the connector space
If an object has been imported but isn’t making it through to the metaverse or Entra ID, trace its path through the sync engine. The Synchronization Service Manager lets you inspect the connector space and metaverse data for an individual object, which is far more useful than running another sync and hoping the problem disappears.
Optimisation isn’t always about making sync faster
If your sync cycle takes 45 minutes, the answer isn’t automatically to change the scheduler. First ask why it takes 45 minutes. Common causes include too many objects in scope, an unusually large number of changes, a sync rule problem, export errors causing repeated processing, a configuration change that needs more extensive processing, an ageing or undersized server, or a problem with the connected directory. Reducing the number of objects that need to sync can help, but changing the scheduler doesn’t make the sync engine itself any faster.
The bigger question
There was a time when an article about optimising Azure AD Connect could focus almost entirely on making the sync engine faster. That’s no longer the whole story. If you’re still running Connect Sync, there are two questions worth answering: is the current installation supported and healthy, and is Connect Sync still the right platform for this environment?
The first has a clear answer. You need to be on a supported release, and realistically the latest one available to you, or sync doesn’t run. The second takes more investigation. Cloud Sync is Microsoft’s longer-term direction, but that doesn’t mean every Connect Sync deployment can or should be migrated immediately. The right choice depends on the features and topology your environment actually uses.
So before spending hours trying to shave a few minutes off a sync cycle, make sure you’re on a supported version, understand what’s actually taking the time, and check whether you’re optimising a platform you ultimately want to replace.
Connect Sync health checklist
If I inherited an existing Connect Sync server, this is the order I’d work through it in.
- Check the version, both installed and configured. Anything below 2.5.79.0 has stopped syncing, and anything not on a current release is on borrowed time, so plan the upgrade to the latest version.
- Confirm the prerequisites for that upgrade, including .NET Framework 4.7.2 and TLS 1.2, and decide between an in-place upgrade and a swing migration.
- Map the current scope by recording which domains and OUs are synced, without changing anything yet.
- Run
Get-ADSyncSchedulerand confirm the interval is the 30-minute default, not a custom schedule someone set years ago and forgot about. - Open the Synchronization Service Manager and review recent operations for failures, unusually long runs and unexpected volumes of adds or deletes.
- Investigate recurring errors. A single failed object isn’t a performance problem, but hundreds of repeated export or attribute errors are worth chasing, starting with duplicate proxyAddresses and UPNs.
- Review filtering for legacy OUs and objects that don’t need to be in Entra ID, and work out the impact before removing anything from scope.
- Document every custom sync rule, and find out whether anyone has edited the built-in rules directly before you perform the next upgrade.
- Confirm whether a staging server exists, and if not, consider building one.
- Once you understand what the environment depends on, compare those requirements against Cloud Sync and decide whether to keep optimising Connect Sync or start planning a migration.
Microsoft references
- Microsoft Entra Connect version release history – current releases and end-of-support dates
- Upgrade from a previous version to the latest – in-place upgrades, swing migration and prerequisites
- Connect to Cloud Sync decision guide – Microsoft’s feature comparison and migration guidance
- What is Microsoft Entra Cloud Sync? – how Cloud Sync differs from Connect Sync

From my early days on the helpdesk through roles as a service desk manager, systems administrator, and network engineer, I’ve spent more than 25 years in the IT world. As I transition into cyber security, my goal is to make tech a little less confusing by sharing what I’ve learned and helping others wherever I can.

