This page explains how to define maintenance windows for Cloud Data Fusion instances.
Maintenance windows are supported for Cloud Data Fusion instances in versions 6.8.0 and later.
About maintenance windows
A maintenance window in Cloud Data Fusion is a time period where you permit it to perform maintenance operations, reducing the risk of service disruptions during peak usage times. Scheduling maintenance windows during off-peak hours, or times when your activity is lower, minimizes their impact on your Cloud Data Fusion instances and your pipeline executions.
Configuring maintenance windows lets you control when maintenance operations occur in your instance:
- If the maintenance window is defined, maintenance operations are performed by Cloud Data Fusion within the defined periods.
- If the maintenance window isn't defined, maintenance can be performed at any time.
About maintenance operations
To ensure system security and compliance, Cloud Data Fusion conducts regular maintenance operations. These operations include applying critical updates and security patches to mitigate vulnerabilities and protect user data. Scheduled maintenance windows ensure that the updates are efficiently applied, preserving system integrity and minimizing disruption to instances and pipeline executions.
- When you configure the maintenance windows, you must allocate at least 12 hours per week for maintenance operations. Each window must be between 4 and 12 hours.
During the maintenance window hours, certain operations might be impacted. We recommend having maintenance windows during non-peak usage times. The maintenance operation can affect Cloud Data Fusion in following ways:
- The Cloud Data Fusion Studio interface might be temporarily unavailable.
- Pipeline execution might be impacted, causing failure.
- Cloud Data Fusion services might be down temporarily, causing API calls to fail.
- Instances might not be fully functional for up to an hour during maintenance window hours. We recommend that you set up maintenance windows when your pipelines either aren't running or aren't scheduled to run.
Specify maintenance windows while creating an instance
Console
To set the maintenance window when you create an instance, follow these steps:
- Go to the Cloud Data Fusion Instances page, and click Create instance.
- Enter an Instance name and Edition. For more information about each field, see Create an instance.
- Expand the Advanced options section.
- Expand the Maintenance section and follow these steps:
- Select the Enable maintenance window checkbox.
- In the Start time field, enter a time in a supported format.
- In the TimeZone list, select a time zone.
- In the Days and Length lists, select days of the week and duration of window. For more information about these settings, see About maintenance operations.
- Click Create.
For more information, see Configure maintenance window times and recurrence (RRULE).
REST API
- Set the
Instance
configuration using theinstance.create
API:
{
"name": string,
"description": string,
…
…
"maintenancePolicy": {
"maintenanceWindow": {
"recurringTimeWindow": {
"window": {
"startTime": "START_TIME",
"endTime": "END_TIME"
},
"recurrence": "MAINTENANCE_RECURRENCE"
}
},
}
}
Replace the following:
START_TIME
: time window start time, in the RFC 3339 format.END_TIME
: time window end time, in the RFC 3339 format.MAINTENANCE_RECURRENCE
: an RRULE for the maintenance window's recurrence. Cloud Data Fusion only supports the following formats:FREQ=WEEKLY;BYDAY=SU,MO,TU,WE,TH,FR,SA
orBYDAY=SU,MO,TU,WE,TH,FR,SA,MO;FREQ=WEEKLY
. This specifies the weekly recurrence of the window on the selected days, starting at thestartTime
, and for the specified duration.
For more information, see Configure maintenance window times and recurrence (RRULE).
The following example creates an instance with a six-hour maintenance window between 10 AM to 4 PM (UTC) on Saturday and Sunday every week. For more information, see the instances API documentation.
// POST https://datafusion.googleapis.com/v1/projects/example-project/locations/us-central1/instances?instanceID=example-instance
{
"maintenancePolicy": {
"maintenanceWindow": {
"recurringTimeWindow": {
"window": {
"startTime": "2024-01-01T10:00:00Z",
"endTime": "2024-01-01T16:00:00Z",
},
"recurrence": "FREQ=WEEKLY;BYDAY=SA,SU"
}
},
}
}
gcloud
You can specify the maintenance window while creating an instance using the gcloud beta data-fusion instances create
command.
The following parameters must be defined for a maintenance window:
--maintenance-window-start
sets the window's start time.--maintenance-window-end
sets the window's end time.--maintenance-window-recurrence
sets an RRULE for the maintenance window's recurrence.
gcloud beta data-fusion instances create INSTANCE_NAME \
--project=PROJECT_NAME \
--location=LOCATION \
--maintenance-window-start=START_TIME \
--maintenance-window-end=END_TIME \
--maintenance-window-recurrence=MAINTENANCE_RECURRENCE
Replace the following:
INSTANCE_NAME
: the name of the instance.PROJECT_NAME
: the name of the project.LOCATION
: the name of the region of the instance.START_TIME
: the window's start time in the date/time input format. Only the specified time of the day is used, the specified date is ignored.END_TIME
: the window's end time in the date/time input format. Only the specified time of the day is used, the specified date is ignored. The specified date and time must be after the start time.MAINTENANCE_RECURRENCE
: an RRULE for the maintenance window's recurrence. Cloud Data Fusion only supports the following formats:FREQ=WEEKLY;BYDAY=SU,MO,TU,WE,TH,FR,SA
orBYDAY=SU,MO,TU,WE,TH,FR,SA,MO;FREQ=WEEKLY
. This specifies the weekly recurrence of the window on the selected days, starting at the--maintenance-window-start
time, and for the specified duration.
For more information, see Configure maintenance window times and recurrence (RRULE).
The following command creates an instance with a six-hour maintenance window between 12 PM to 6 PM (UTC) on Saturday and Sunday every week.
gcloud beta data-fusion instances create example-instance \
--project='example-project' \
--location='us-east1' \
--maintenance-window-start='2024-01-01T12:00:00Z' \
--maintenance-window-end='2024-01-01T18:00:00Z' \
--maintenance-window-recurrence='FREQ=WEEKLY;BYDAY=SA,SU'
Update the maintenance windows for an instance
Console
To update the maintenance window in an existing instance in the Google Cloud console, follow these steps:
Go to the Cloud Data Fusion Instances page.
Click the instance name to view the instance details.
On the Instance details page, navigate to the Maintenance window field and click Edit.
Maintenance window fields are opened in the edit mode. Update the values, as needed:
- Select the Enable maintenance window checkbox.
- In the Start time field, enter a time in a supported format.
- In the TimeZone list, select a time zone.
- In the Days and Length lists, select days of week and duration of window. For more information about these settings, see About maintenance operations.
Click Save.
For more information, see Configure maintenance window times and recurrence (RRULE).
REST API
For existing instances, update the maintenance windows using the
instances.patch
API request. This overrides the existing maintenance window configuration, if
present. Otherwise, a new configuration is set, as mentioned in the request
body.
Construct an API request as follows:
- In the
updateMask
parameter, specify themaintenance_policy
mask. - In the request body, specify the maintenance window in the maintenance policy configuration:
{
"maintenancePolicy": {
"maintenanceWindow": {
"recurringTimeWindow": {
"window": {
"startTime": "START_TIME",
"endTime": "END_TIME"
},
"recurrence": "MAINTENANCE_RECURRENCE"
}
},
}
}
For more information, see Configure maintenance window times and recurrence (RRULE).
The following example updates the maintenance policy configuration of the instance with a four-hour maintenance window between 11 AM to 3 PM (UTC) on Friday, Saturday, and Sunday every week.
// PATCH https://datafusion.googleapis.com/v1/projects/example-project/locations/us-central1/instances/example-instance?updateMask=maintenance_policy
{
"maintenancePolicy": {
"maintenanceWindow": {
"recurringTimeWindow": {
"window": {
"startTime": "2024-01-01T11:00:00Z",
"endTime": "2024-01-01T15:00:00Z",
},
"recurrence": "FREQ=WEEKLY;BYDAY=FR,SA,SU"
}
},
}
}
gcloud
You can update the maintenance window of an instance using the gcloud beta data-fusion instances update
command.
The following parameters must be defined for a maintenance window:
--maintenance-window-start
sets the window's start time.--maintenance-window-end
sets the window's end time.--maintenance-window-recurrence
sets an RRULE for the maintenance window's recurrence.
gcloud beta data-fusion instances update INSTANCE_NAME \
--project=PROJECT_NAME \
--location=LOCATION \
--maintenance-window-start=START_TIME \
--maintenance-window-end=END_TIME \
--maintenance-window-recurrence=MAINTENANCE_RECURRENCE
Replace the following:
INSTANCE_NAME
: the name of the instance.PROJECT_NAME
: the name of the project.LOCATION
: the name of the region of the instance.START_TIME
: the window's start time in the date/time input format. Only the specified time of the day is used, the specified date is ignored.END_TIME
: the window's end time in the date/time input format. Only the specified time of the day is used, the specified date is ignored. The specified date and time must be after the start time.MAINTENANCE_RECURRENCE
: an RRULE for the maintenance window's recurrence. Cloud Data Fusion only supports the following formats:FREQ=WEEKLY;BYDAY=SU,MO,TU,WE,TH,FR,SA
orBYDAY=SU,MO,TU,WE,TH,FR,SA,MO;FREQ=WEEKLY
. This specifies the weekly recurrence of the window on the selected days, starting at the--maintenance-window-start
time, and for the specified duration.
The following command updates the instance example-instance
with a six-hour
maintenance window between 12 PM to 6 PM (UTC) on Saturday and Sunday every
week.
gcloud beta data-fusion instances update example-instance \
--project='example-project' \
--location='us-east1' \
--maintenance-window-start='2024-01-01T12:00:00Z' \
--maintenance-window-end='2024-01-01T18:00:00Z' \
--maintenance-window-recurrence='FREQ=WEEKLY;BYDAY=SA,SU'
Configure maintenance window times and recurrence (RRULE)
When you configure start and end times for the maintenance window, the date components of the provided timestamps are ignored. Only the time of the day and duration between start and end time are considered. Times are stored in UTC, but you can use UTC or your time zone offset when scheduling maintenance windows using REST API or the gcloud CLI.
When you configure recurrence, keep in mind that RRULE follows UTC days, so
adjust accordingly if your time zone differs. For example, if your time zone is
PDT (UTC-7) and you want a maintenance window to start at 21:00 on Wednesdays in
your time zone (which corresponds to 4:00 UTC on Thursdays), specify TH
instead of WE
in the BYDAY
option of the RRULE.
Remove the maintenance window for an instance
Console
To remove the maintenance window in an existing instance in the Google Cloud console, follow these steps:
Go to the Cloud Data Fusion Instances page.
Click the instance name to view the instance details.
On the Instance details page, navigate to the Maintenance window field and click Edit.
To remove the maintenance window from the instance, clear the Enable maintenance window checkbox.
Click Save.
REST API
To remove the maintenance window from an instance, pass an empty maintenance policy object in the request body.
// PATCH https://datafusion.googleapis.com/v1/projects/example-project/locations/us-central1/instances/example-instance?updateMask=maintenance_policy
{
"maintenancePolicy": {}
}
gcloud
To remove the maintenance window from an instance, use gcloud beta data-fusion instances update
command with the --clear-maintenance-window
flag.
gcloud beta data-fusion instances update INSTANCE_NAME \
--project=PROJECT_NAME \
--location=LOCATION \
--clear-maintenance-window
Replace the following:
INSTANCE_NAME
: the name of the instance.PROJECT_NAME
: the name of the project.LOCATION
: the name of the region of the instance.
What's next
- Refer to the instances API documentation.