November 2026
Continuous Data 2026.6.0.0 expands Oracle support and strengthens data protection with role-based SnapSync for Oracle Data Guard, support for Oracle Exadata Exascale, shared external TDE keys across live vPDBs, and OCI WORM offline replication. DCT 2026.6 will add VDB recovery, centralized engine upgrades, and performance dashboards as Continuous Data interfaces are phased out; this release also improves engine security and Elastic Data stability and begins the staged retirement of SQL Server 2012 and Windows Server 2012/R2 support.
New features
Oracle Data Guard support
Continuous Data now supports Oracle Data Guard configurations, so SnapSync keeps working when your primary and standby databases change roles. When you link a dSource to an Oracle database that belongs to a Data Guard configuration, the engine detects the topology, records each member and its last known role as a Data Guard group, and continues SnapSync from the member you choose: the primary, the standby, or no preference. At the next SnapSync after a switchover or failover, the engine reattaches to the member in the role you specified, so ingestion continues without manual reconfiguration. Single-instance, RAC, and multitenant (CDB and PDB) databases are supported with physical standbys.
This feature is always on and changes the behavior of existing dSources on Data Guard databases. Upgrading does not create Data Guard groups for existing dSources. After you upgrade, refresh the environment of each linked Data Guard database, or run refreshDgRoles on its source config, to create the group. Detection during an environment refresh requires the environment OS user to be a member of the OSDBA or OSBACKUPDBA group. When Delphix creates the group, it sets syncFrom to the linked database’s role at that time. After a later switchover or failover, SnapSync continues from the member that holds that role, which earlier releases did not do.
-
Each Oracle database must have a unique
DB_UNIQUE_NAMEacross all environments added to the Delphix Engine. -
Data Guard groups are managed through the Delphix Engine API and CLI. Data Guard support in DCT (API and UI) will be available in the DCT 2026.6.0 release.
-
Data Guard group details, including each member's last known role, are available through the source configuration API.
For details, read Create the Data Guard group for a dSource linked before the upgrade.
Shared external TDE master keys across live Oracle vPDBs
You can now use the same externally managed Transparent Data Encryption (TDE) master key, such as one issued by an HSM or Oracle Key Vault, for multiple live Oracle virtual pluggable databases (vPDBs). This lets you keep one centrally managed key across your non-production environments while continuing to stop, refresh, and disable vPDBs as usual. When another vPDB already uses the same key, the engine posts a warning to the job so an unintentional overlap stays visible without blocking the operation.
This release also fixes an ORA-46659 error that could occur when two vPDBs on the same CDB with a FILE or AUTOLOGIN keystore were disabled at the same time.
For more information, read A closer look at TDE provisioning and Refreshing and rewinding a TDE-enabled vPDB.
Oracle Exadata Exascale Vault support
Continuous Data now supports Oracle Database sources that store their database files in Exascale Vaults. For Oracle Database 23ai and 26ai using native Exascale Vault storage, Delphix recognizes the @vaultname alias paths that are used in place of ASM disk group paths. Earlier Oracle Database versions are also supported when using Exascale Direct Volumes that provide ASM disk group aliases.
For Oracle support details, read the Oracle matrix.
OCI WORM support for offline replication
Continuous Data now supports Oracle Cloud Infrastructure (OCI) Write Once Read Many (WORM) Object Storage as a destination for offline replication. A source engine in one segregated zone writes to the WORM storage, and a target engine in another zone reads from it, with no direct connection between the engines. Replicated data can’t be altered or deleted by anyone, including administrators, until its retention period ends. This keeps golden copies and backups tamper-proof against ransomware and malicious deletion, and helps you meet regulatory data-retention requirements. Available from the CLI and Data Control Tower. To learn more, see Offline replication (air gap).
For more information, read OCI WORM offline replication.
NFS server disabled by default on engines that do not use it
The NFS server (port 2049) is now disabled by default on Delphix Engines that do not use NFS exports, addressing security scanner findings for an open RPC service. Choosing the Virtualization engine type at setup enables it. On Continuous Compliance-only engines and on Data Control Tower, Hyperscale, and PUC appliances it remains disabled. No user action or configuration change is required.
Recycle bin for deleted VDBs (DCT 2026.6)
DCT 2026.6 will include a Recycle Bin, so you can recover a VDB after an accidental deletion. When you delete a VDB, you will be able to choose Move to Recycle Bin instead of deleting it permanently. The new Admin > Recycle Bin page will show each deleted VDB and how long it remains recoverable, and let you restore it, delete it permanently, or empty the Recycle Bin. VDBs will be purged automatically when their retention period ends. The Recycle Bin will also be available through the DCT API.
Engine upgrades from Data Control Tower (DCT 2026.6)
In DCT 2026.6, you will be able to upgrade Continuous Data and Continuous Compliance engines without signing in to each engine's SysAdmin UI. Upload an upgrade image to DCT once, push it to one or more engines, review each engine's upgrade verification results, and apply the upgrade when you are ready. DCT checks replication order so targets are upgraded before sources, and you can track each engine's upgrade progress in DCT. Engine upgrades will be available in the DCT UI and API for engines running 2026.4.0.0 or later.
Engine performance dashboards in Data Control Tower (DCT 2026.6)
DCT 2026.6 will show real-time performance dashboards for each engine, so you can monitor engine health without signing in to the engine. The new Performance tab on the engine details page will chart CPU, disk I/O, network, NFS, iSCSI, and ZFS activity over time ranges from 5 minutes to 30 days. You will also be able to export an engine's raw metrics as CSV through the DCT API. Performance dashboards will be available for engines running 2026.6.0.0 or later.
Fixed issues
Release 2026.6.0.0 (November 2026)
Fixes that take effect immediately after upgrading
| Issue number | Description |
|---|---|
| DLPX-39280 | The dSource TimeFlow missing-logs error incorrectly suggested the GUI Snapshot repair tool, which does not apply to TimeFlow gaps, instead of the CLI TimeFlow Log Fetch command. |
| DLPX-63082 | When a job failed because the network session to a host was interrupted, the error message showed an empty "[]" instead of naming the affected host. |
| DLPX-73465 | Oracle pluggable database (PDB) dSources on a mounted physical standby container database always showed "Inactive" status, even though the PDB could be successfully synced. |
| DLPX-82259 | Detailed error messages could disclose internal application information. |
| DLPX-89030 | After a manual upgrade performed by Delphix Support, the engine could be missing the install date for its running version, causing replication jobs to fail with an internal error. |
| DLPX-90470 | The Content-Security-Policy header did not sufficiently restrict the script-src and style-src directives. |
| DLPX-94922 | A failed upgrade could leave behind an upgrade container that held significant root pool space and could not be found or removed. |
| DLPX-96439 | MSSQL Validated Sync fault messages for missing backup files did not include the backup set UUID or remediation guidance. |
| DLPX-96605 | On Elastic Data engines with very large storage pools, the storage agent could repeatedly run out of memory and restart shortly after starting up, making the engine unstable or unreachable. |
| DLPX-97582 | The Content-Security-Policy header allowed unsafe inline styles and scripts in the management UI. |
| DLPX-97714 | iSCSI LUN mounts could fail with ERROR_ASSIGN_MOUNTPATH on non-English Windows staging and target hosts because GPT disk detection relied on English diskpart output. |
| DLPX-98132 | After converting a single-tenant Oracle dSource to multitenant, the resulting dSource could appear correctly configured with an effective policy, while silently having no scheduled SnapSync, Retention, or other policy enforcement running against it. |
| DLPX-98174 | Elastic Data engines using some S3-compatible object storage, such as Dell ObjectScale, could fail to write to object storage after upgrading to 2026.4.0.0. A configuration option is now available to restore compatibility. |
| DLPX-98284 | The engine responded to ICMP timestamp requests, which could disclose the engine's system time to remote hosts (CVE-1999-0524). |
| DLPX-98535 | SMTP alert delivery failures caused by a TLS certificate trust issue did not produce a meaningful error message. |
| DLPX-98566 | Concurrent Oracle database operations could incorrectly share a single query timeout setting, causing one operation to override another's timeout value and potentially leading to unexpected query timeouts. |
| DLPX-98647 | Refreshing a Windows environment that was missing the required dlpxrunas file produced an unclear error message that did not name the missing file. |
| DLPX-98649 | CDB log retention used only the CDB's log retention policy instead of the maximum retention policy across the CDB and its associated PDBs. |
| DLPX-98657 | On Elastic Data engines, the storage agent could stop unexpectedly while reporting cache status, making the engine and its VDBs unavailable until the agent restarted. |
| DLPX-98722 | Updated Spring as part of routine dependency maintenance, which includes fix for multiple CVEs with Spring version 6.2.19 and older. |
| DLPX-98785 | On KVM-based hypervisors such as OpenShift Virtualization, multiple storage devices could be displayed with the same name. Device names on these platforms now include the SCSI LUN, so each device has a unique name. |
| DLPX-98826 | Object storage cache performance could degrade under high load. |
| DLPX-98851 | An Oracle PDB entry in oranfstab could overwrite the CDB's or another PDB's entries, resulting in incorrect NFS version being used. |
| DLPX-98892 | In-place export to physical (V2P) of an Oracle VDB created with file mapping rules could fail with an RMAN-06172 error during setup. |
| DLPX-99078 | SnapSync on a physical standby database as a dSource after switchover could result in the snapshot being permanently non-provisionable. |
| DLPX-99308 | During an upgrade, the engine console could refresh repeatedly instead of showing a login prompt, making the engine appear to fail to boot. |
| DLPX-99309 | Resuming an interrupted upgrade could fail with "failed to reload delphix-platform.service", requiring manual steps to complete the upgrade. |
| DLPX-99310 | Each upgrade image upload could leave a full-size temporary copy of the image on the engine, so repeated attempts could use up the space needed to upgrade. Uploads now need about twice the image size in free space, not three times. |
| DLPX-99425 | Provisioning a new Oracle vCDB on an on-premises Exadata target could fail with an ORA-27138 error because Delphix did not correctly detect Exadata platforms outside Oracle Cloud Infrastructure. |
| DLPX-99620 | Enabling iSCSI encryption could fail without any errors in the logs. |
| DLPX-99657 | Discovering or refreshing a SQL Server cluster environment failed when a source Availability Group had two or more listeners. |
| DLPX-99823 | Receiving a replication stream that freed a large range of a sparse dataset could cause the engine to run out of memory or the replication receive to stall. |
| DLPX-99949 | Disk I/O and NFS data could be missing from the Performance Analytics screen on engines running 2026.5.0.0 or later. |
| DLPX-100051 | Updated Node.js as part of routine dependency maintenance, including fixes for multiple CVEs such as CVE-2026-111497. |
| DLPX-100190 | Updated the JDK to 17.0.21+9 to address eMudhra-related vulnerabilities. |
| DLPX-100271 | Removed eMudhra emSign roots from the engine truststore. |
| DLPX-100390 | Removed eMudhra emSign roots from the JDK 17.0.20.1 used on AIX hosts. |
Fixes that take effect after upgrading and rebooting
| Issue number | Description |
|---|---|
| DLPX-98640 | The engine could crash with a kernel panic when an NFSv4.1 or later client retransmitted requests to the engine, such as during network disruptions. |
| DLPX-98642 | The engine could crash with a kernel panic when NFSv4.1 or later clients, such as Oracle Direct NFS, repeatedly disconnected and reconnected to the engine. |
| DLPX-98671 | On Elastic Data engines, the engine could crash with a kernel panic when the storage agent restarted several times in quick succession. |
| DLPX-100246 | Engines could crash, or be exposed to further attack, when remote TCP traffic carrying IPv4 options caused the engine to use memory it had already released (CVE-2026-98432). |
Known issues
| Issue number | Summary | Workaround |
|---|---|---|
| DLPX-92789 | Upgrading a replication target engine could fail if it has replicated TimeFlow data with a SnapSync still in progress on the source at the time of replication. | Before upgrading the target engine, wait for the in-progress SnapSync to create a snapshot on the source TimeFlow. Then replicate the snapshot to the target engine before upgrading the target. |
| DLPX-100019 | On a host that uses Oracle bequeath (BEQ) connections, a hung connection attempt can block later connections and process starts for the Delphix Connector. This can cause SnapSync, refresh, and other jobs for all Oracle dSources and VDBs on the host to hang. | Restart the Delphix management service on the engine. On the affected target host, kill any leftover connector (client.jar) process that has a single thread and parent PID 1, as well as any connector process started before the restart. Wait up to 5 minutes for the next environment-monitor cycle to start a fresh connector. The hang can recur until a fix is available. |
| DLPX-100206 | Oracle LogSync fails to fetch archived redo logs larger than 32 GiB (at the default 512-byte block size), causing the affected log sequences to be skipped and breaking dSource synchronization. | No workaround that preserves future log capture and recoverability has been verified. Disabling LogSync on the affected dSource avoids the failure, but stops future log capture and recoverability until the fix is applied. |
API changes
In Delphix 2026.6.0.0, the new API version is 1.11.52. This section describes all API changes since API version 1.11.51, which was released with Continuous Data 2026.5.0.0. All URL paths are relative to /resources/json/delphix.
Updates
| API object | Path | Type | Name | Change |
|---|---|---|---|---|
OfflineReplicationReceiveSpec
|
NA | Property | cleanupOfflineData
|
Description updated. Cleans up data from the offline storage target (NFS share or non-WORM OCI bucket) after it has been successfully received. |
OfflineReplicationReceiveSpec
|
/replication/receive
|
Property | receivePolicy
|
Added property. Supports NEXT_IN_CHAIN, which receives the next sequential serialization point and is the default, and LATEST_FULL_PREFERRED, which receives the highest available full serialization point. |
Plugin
|
NA | Property | language
|
Added enum values "PYTHON27" and "PYTHON38". Previously only "PYTHON311" was supported. |
ReplicationOfflineTraffic
|
NA | Property | fullSendPolicy
|
Description updated. Applicable to any offline storage type. For an OCI WORM bucket, a policy that sets intervalDuration is additionally bounded by the bucket's retention period. |
Additions
| API object | Path | Type | Name | Change |
|---|---|---|---|---|
FullSendPolicy
|
NA | API type | FullSendPolicy
|
New API type added. Policy controlling when a full serialization point is sent to offline storage. Properties: intervalRuns, intervalDuration. |
InfluxDBMetadata
|
/influxdb/metadata
|
API type | InfluxDBMetadata
|
New API type added. InfluxDB metadata (organization, bucket, token) for DCT Smart Proxy configuration. |
OciApiKeyAuth
|
NA | API type | OciApiKeyAuth
|
New API type added. Authenticates using a customer-supplied OCI API signing key pair, for engines that are not OCI compute instances. Required on create: region, tenancyOcid, userOcid, fingerprint, privateKey. Optional: passphrase. |
OciAuth
|
NA | API type | OciAuth
|
New API type added. Abstract specification of how an OCI object storage bucket is authenticated to. |
OciInstancePrincipalAuth
|
NA | API type | OciInstancePrincipalAuth
|
New API type added. Authenticates using the engine's own OCI instance identity. Requires the engine to be an OCI compute instance enrolled in a dynamic group. |
OciOfflineStorage
|
NA | API type | OciOfflineStorage
|
New API type added. OCI object storage bucket used as the offline replication storage location. Properties: bucketName, auth. |
OciOfflineStorage
|
NA | Property | auth
|
New property added. How the OCI bucket is authenticated to. Defaults to instance-principal authentication when omitted. |
OracleBaseDbConfig
|
NA | Property | dataguardGroupRef
|
New property added. Read-only reference to the DataGuard group this source configuration belongs to. Non-null for CDB or non-MT SI/RAC DataGuard members; null for other Oracle configs; omitted for non-Oracle configs. |
OracleBaseDbConfig
|
NA | Property | dataguardInfo
|
New property added. DataGuard group membership and properties for this source configuration. Populated on single-reference GET and refreshDgRoles responses; not populated on list responses. |
OracleDgGroup
|
NA | API type | OracleDgGroup
|
New API type added. Read-only Oracle DataGuard group discovered on the engine. Properties: name, description, enabled, syncFrom, skipFullSync. |
OracleDgGroup
|
/sourceconfig/oracle/dataguardgroup
|
Operation | list
|
Added operation. Returns the list of all Oracle DataGuard groups on the engine. |
OracleDgGroupMember
|
NA | API type | OracleDgGroupMember
|
New API type added. Read-only member of an Oracle DataGuard group. Properties: dgGroup, config, role. |
OracleDgGroupMember
|
/sourceconfig/oracle/dataguardgroupmember
|
Operation | list
|
Added operation. Returns the list of all Oracle DataGuard group members across every group on the engine. |
OracleDgGroupMemberView
|
NA | API type | OracleDgGroupMemberView
|
New API type added. A member of an Oracle DataGuard group as seen from a source configuration. Properties: name, role, current. |
OracleDgGroupView
|
NA | API type | OracleDgGroupView
|
New API type added. View of the DataGuard group that a source configuration belongs to, including updatable group properties. |
ReplicationOfflineTraffic
|
NA | Property | fullSendPolicy
|
New property added. Policy controlling when a full serialization point is sent. |
ReplicationSpecExecuteParameters
|
/replication/spec
|
Payload | ReplicationSpecExecuteParameters
|
Added payload. Triggers execution of a replication spec. |
ReplicationSpecRuntime
|
NA | Property | lastFullSendTimestamp
|
New property added. Timestamp of the last full send to offline storage. |
ReplicationSpecRuntime
|
NA | Property | runsSinceLastFullSend
|
New property added. Number of successful incremental sends since the last full send to offline storage. Resets to 0 after each full send. |
SerializationPoint
|
/replication/serializationpoint
|
Property | fullSend
|
Added Boolean property. |
SourceConfig
|
NA | Operation | refreshDgRoles
|
New operation added. Refreshes the DataGuard member roles for an Oracle source configuration by querying all member databases via JDBC. Also creates the DataGuard group if one does not exist yet for the source configuration. Returns the updated source configuration. |
User
|
/user
|
Property | currentCredential
|
New property added. The user's current password, required to change the principal on one's own User object. |
Deprecated and EOL features
Continuous Data user interfaces end of life
As we transition to the Data Control Tower platform experience, we are sunsetting all Continuous Data management and setup interfaces. The transition happens in stages:
-
September 2026: Starting with 2026.5, new engine installations no longer include the Management UI, while upgraded engines are unaffected.
-
July 2027: With 2027.4, all UI, CLI, and API interfaces are removed from new installs and upgrades, marking End of Maintenance; support will continue to triage issues and offer reasonable workarounds, with equivalent fixes delivered through Data Control Tower or interface-free engine upgrades.
-
January 2029: End of Maintenance and Support concludes.
As you migrate, your data, dSources, virtual databases, and configurations remain fully intact; only the interfaces change. All Delphix customers are entitled to Data Control Tower, and we recommend installing it and connecting your engines now to begin the transition.
For more information, read the full announcement and FAQ.
SQL Server 2012 and Windows Server 2012 end of life
Starting with Continuous Data 2026.6, support for SQL Server 2012, Windows Server 2012, and Windows Server 2012 R2 is deprecated. These versions have reached end of support from Microsoft, and Delphix will retire support for them in stages:
-
November 2026: With 2026.6, SQL Server 2012, Windows Server 2012, and Windows Server 2012 R2 are deprecated. Existing environments on these versions continue to work.
-
November 2027: With 2027.6, these versions reach end of life. Delphix no longer supports them as source, staging, or target environments.
To avoid disruption, upgrade affected environments to a supported SQL Server and Windows Server version before you upgrade to Continuous Data 2027.6. For the list of supported versions, read the SQL Server matrix documentation.