Restore
The GRAX Restore feature gives users the ability to update selected records to an earlier state which was backed up by GRAX, as well as the option to restore previously deleted records back into Salesforce. This is crucial for users who have a catastrophic event and need to restore records, but it isn't the only challenge Salesforce Administrators face. What about a data corruption that only affects a few fields, or one that has escaped the notice of the team for a few days or weeks? How would the administrator mass restore thousands, or a hundred thousand records, to a version of this record from a few weeks ago?
With GRAX Restore, administrators have the ability to restore individual field level changes, or multiple fields, to any point in time in the history of the GRAX Backup.
Restore in the GRAX Application
Open the GRAX Application and navigate to the Restore tab.
The Restore home page shows a list of all the previously configured batch and single Restore executions and the different statuses that they can be in.

Possible Restore Statuses
Status | Description |
---|---|
Pending | A Restore execution is in a Pending state if further configuration is required before the execution can be launched. |
Ready | A Restore execution is Ready when it has loaded data from the source and validated which records of this data set exist for the specified execution. |
Queued | A Restore execution is Queued when a user has submitted the job, but it is waiting for a previous Restore execution to complete. |
Running | The Restore execution is currently running. \n\nThere can only ever be one running Restore execution at a time. |
Completed | The Restore has completed with zero errors reported back. |
Aborted | The Restore has been aborted by a user. |
Error | The Restore has completed but some (or all) of the records were unable to be restored due to an error. |
Creating a New Restore Execution
The Restore process starts with the creation of a new Restore execution. From the Restore page, click the New Restore
button. This launches a small guided process that takes you through the steps to create and launch either a Point in Time Restore execution or a Create Records Restore execution.
Select Data Source
Reports (Point in Time Restore only)
You can use a report in Salesforce to indicate the records that need to be restored. To select the records that are going to be restored, create a report in Salesforce that
- is for a single object only
- contains (at least) the record ID as a column
- doesn't have any cross object joins, grouping or summary information
- is stored in a folder that the GRAX Integration User has access to
GRAX Reports
For the report to be visible in the GRAX Application it must be saved to a folder with the name GRAX in it, or to have the name GRAX in the Report Title. The report also needs to be saved to a location that the GRAX Integration User has access to. We recommend that you create a specific folder, "GRAX Reports" that you use to keep GRAX specific Salesforce reports in.
100k Record Limit using Reports
If you seed using the Salesforce Report, there is a limit of 100,000 records that can be loaded into the GRAX Application. If GRAX identifies there are more than 100,000 records in the report, you won't be able to proceed. GRAX first takes a few seconds to analyze the Salesforce report to make sure there aren't more than 100,000 records. If there are too many records, the report needs to be reduced in size before continuing.
Search (Create Records Restore only)
You can use Global Search to find records to be restored; simply select the object and record status, set the date filter and use the optional field filters to narrow down your search. The search will start by returning 10 parent records, but the remaining results can then be restored via an Auto Restore execution.
CSV
Upload a CSV file to restore records by record id. This will restore records in a batch of up to 10,000, but an Auto Restore execution can be setup to restore the remaining records on the file. Please note that the CSV must contain a header named Id
and must not exceed 500 MB.
Record IDs
You can manually enter up to 200 record ids to restore by listing them and separating each one by either a comma or a new line.
Select Restore Options for Point in Time Restore
GRAX Point in Time Restore allows the user to be specific about which data they wish to restore, allowing a partial update of a record to a new version. This can be particularly useful when a process has been incorrectly updating data, while at the same time users have been working with these records.
Select the Point in Time and Choose Restore Fields
The Point in Time looks at the date fields on the record and selects the values from the GRAX Data Vault that were the latest version at the date / time specified by the Point in Time. Depending on the object type, this would be the System Modstamp
, Last Modified Date
or Created Date
standard fields.
Fields need to be added to the Restore by selecting and moving the fields within the filtered dueling picklist control. To restore all fields, select all fields by moving them into the right hand list.

Records Filtered by Point in Time
It's possible to filter your initial Restore seed by selecting a Point in Time that is before the created date of the records you wish to restore.
If the Restore confirmation record count doesn't match the expected record count from your Restore seed, it may be that you have excluded some of these records by the selection of this Point in Time value.
Select Restore Options for Create Records Restore
The configuration page for the Create Records restore allows you to select the level of child records to restore and provides an option to skip specified objects. There are also options to send emails regarding the job, include required parent records in the job, and disable automation which can sometimes help when testing.

Preview Restore
Once the Restore configuration has been set, GRAX begins to prepare the data set for restore from the GRAX Data Vault. This process can take some time (minutes even) depending on the size of the Restore and while this evaluation is underway you won't be able to proceed from this step.
Execution in Pending State
If you close the tab here, you can return to the job by clicking on the newly created Restore execution in the Restore home tab which will show in the "pending" state; there will also be a banner on this page for each pending job.
Cancelling a Restore execution deletes it from the list on Restore home.
Here you are presented with a preview of the records and values that are going to be restored. This data is different from the Restore seed - it's the data that is going to be written back to your Salesforce org as part of the Restore execution, so it's worthwhile ensuring that you are comfortable with your selections.
Once the data finishes the preparation phase, the Next
button activates. You can download the complete record set as a CSV file here as well.
Null or Empty Values
There are a number of scenarios that could mean your GRAX Vault contains records but not values for all fields that you are trying to restore.
- The field might have been empty at the time of the backup
- The field might not have existed at the time of the backup
- The field might not have been accessible to the GRAX Integration User at the time of the backup.
Null values appear in the Restore preview as #N/A and reset any field values in the record being restored to empty. If this isn't the intended behaviour, deselect this field before continuing.
Overrides and Relationships
When viewing a preview of the Restore job, you can change to a list view to provide options for overrides and selecting/deselecting relationships. To override a field value for a Restore job, select the object on the left hand side, click on the Overrides
section, and then select the field you want to proceed with. You can enter a new value to be written to this field, or you can leave the value blank to exclude this field from the restored record.

You can also navigate to the Relationships
section to include or exclude specific relationship attributes for the restored records.
Launch Restore
The Point in Time Restore confirmation screen informs the actual number of records that are going to be affected, the fields on those records and the point in time to which that data is going to be restored.

If this all looks good, clicking the Restore Data
button starts the process. Please note that for Create Records Restore jobs, this notification will not appear, but a graph that shows the records to be restored will be available for review before launching the restore.
Restoring Data is Writing to Salesforce
This writes data to your configured Salesforce environment; be certain you understand the possible ramifications this might have.
Please review the Restore Best Practices as these apply to both Point in Time and Create Records Restore jobs.
Monitoring a Restore Execution
Once the Restore execution has been submitted it can be monitored in the GRAX Console. This submits records in batches to Salesforce so you can see a number of different outcomes here.

Aborting a Restore
You can abort a Restore Execution but this won't perform a Rollback of data that has been successfully processed as part of this Point in Time Restore. Any data that has been successfully updated would need to be included in a new Point in Time Restore if it needed to be reverted to a previous state.
Restore Records Status
In addition to the status of the job as a whole, individual batches of records have a status while the job is running and once it has completed.
Status | Description |
---|---|
Pending | Data has been prepared but not yet submitted to Salesforce |
Submitted | Data has been submitted to Salesforce, waiting on the result from this API transaction |
Successful | Data was successfully restored. |
Error | Data failed to be restored. |
Restore Errors and the Retry Flow
When restoring records to Salesforce you may encounter errors like validation rule failures, changes to the object schema, or other users updating the same records.
GRAX Restore allows a user to retry just the failed records. This creates a new Restore execution with this subset of data as the restore seed.

Here is a Restore that has failed due to an inactive user as the value in the Owner field. This can be fixed by retrying the restore job, and setting an override with a null value to the Owner field.
Restore errors should provide the needed information to resolve the issue; typically a change to validation rules or triggers, or a field override is necessary to proceed.
Updated 18 days ago