Skip to content

RDSC-3622 static and dynamic ttl examples #1620

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Changes from all commits
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
@@ -0,0 +1,164 @@
---
Title: Set custom expiration times / TTL
alwaysopen: false
categories:
- docs
- integrate
- rs
- rdi
description: null
group: di
linkTitle: Set expiration times / TTL
summary: How to set expiration times / TTL for keys
type: integration
weight: 40
---


You can configure custom key expiration times (TTL) for keys written to Redis by using the `expire` parameter in the `output` section of the job file. This parameter specifies the duration, in seconds, that a newly created key will remain in Redis before being automatically deleted. If the `expire` parameter is not provided, the keys will persist indefinitely.

There are two ways to set the expiration time:

- as a static value
- as a dynamic value using a JMESPath or SQL expression


## Static expiration time

The following example sets the expiration time to 100 seconds for all keys:

```yaml
output:
- uses: redis.write
with:
data_type: hash
expire: 100
```

## Dynamic expiration time

You can use a JMESPath or SQL expression to set the expiration time dynamically when it is based on a field in the source data. For example, you can set the expiration time to the value of a `ttl` field in the source data:

```yaml
output:
- uses: redis.write
with:
data_type: hash
expire:
expression: ttl
language: jmespath
```

## Dynamic expiration time based on a date, datetime, or timestamp field

In some cases, you can also set the expiration time based on a field that contains a date, datetime, or timestamp value, but it depends on the source database and the data types it supports. See the examples below for your specific source database and data type.

### Oracle examples

The transformation depends on the data type of the field in the source database:

- `DATE` - represented by debezium as a 64-bit integer representing the milliseconds since epoch
```yaml
output:
- uses: redis.write
with:
data_type: hash
expire:
# To set the expiration time to a date field, convert the value to seconds and subtract the current time in seconds since epoch
expression: EXPIRES_DATE / 1000 - STRFTIME('%s', 'now')
language: sql
```
- `TIMESTAMP` - the value is represented by Debezium as a 64-bit integer and depends on the number of decimal places of precision of the column, representing fractions of a second. For example, if the column is defined as `TIMESTAMP(6)`, there are six decimal places and so the value is represented as microseconds since epoch (since there are 10^6 microseconds in each second).
```yaml
output:
- uses: redis.write
with:
data_type: hash
expire:
# To set the expiration time to a date field, convert the value to seconds (divider differs based on the fractional second precision) and subtract the current time in seconds since epoch. Example below is for 6 digits of precision.
expression: EXPIRES_TIMESTAMP / 1000000 - STRFTIME('%s', 'now')
language: sql
```
- `TIMESTAMP WITH TIME ZONE` - the value is represented as string representation of the timestamp with time zone information.
- `TIMESTAMP WITH LOCAL TIME ZONE` - the value is represented as string representation of the timestamp with local time zone information.

For both `TIMESTAMP WITH TIME ZONE` and `TIMESTAMP WITH LOCAL TIME ZONE`, a two-step approach is needed. First, calculate the difference between the given time and now in seconds and then invert the value.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe needs a bit more explanation here?

Suggested change
For both `TIMESTAMP WITH TIME ZONE` and `TIMESTAMP WITH LOCAL TIME ZONE`, a two-step approach is needed. First, calculate the difference between the given time and now in seconds and then invert the value.
Both `TIMESTAMP WITH TIME ZONE` and `TIMESTAMP WITH LOCAL TIME ZONE` require a two-step approach. First, calculate the difference between the given time and the current time in seconds (using `time_delta_seconds ()`). This returns a negative value if the time is in the future, so you must then also negate the returned value.

```yaml
transform:
- uses: add_field
with:
fields:
- field: expire_seconds
language: jmespath
expression: time_delta_seconds(EXPIRES_TS_TZ)
output:
- uses: redis.write
with:
data_type: hash
expire:
# `time_delta_seconds` Returns the number of seconds between a given dt and now.
# A negative value means that the given dt is in the future, so we need to invert it.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Again, "negate" might be a bit more specific than "invert".

Suggested change
# A negative value means that the given dt is in the future, so we need to invert it.
# A negative value means that the given dt is in the future, so you must negate it.

# A positive value means that the given dt is in the past, so set the expiration to -1 (expire immediately).
expression: CASE WHEN expire_seconds < 0 THEN -expire_seconds ELSE -1 END
language: sql
```

----

### SQL Server examples
SQL Server supports the following date and time data types:

- `date` - represented in Debezium as number of days since epoch (1970-01-01). Please note that due to the lack of time information, this method is not very accurate.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Minor point, but you shouldn't normally use "please" in the docs.

Suggested change
- `date` - represented in Debezium as number of days since epoch (1970-01-01). Please note that due to the lack of time information, this method is not very accurate.
- `date` - represented in Debezium as number of days since epoch (1970-01-01). Note that due to the lack of time information, this method is not very accurate.

```yaml
output:
- uses: redis.write
with:
data_type: hash
expire:
# Calculate the number of seconds equivalent to the number of days and subtract the current time in seconds since epoch.
expression: (event_date * 86400) - strftime('%s', 'now')
language: sql
```

- `datetime`, `smalldatetime` - represented in Debezium as number of milliseconds since epoch.
```yaml
output:
- uses: redis.write
with:
data_type: hash
expire:
# Since event_datetime is in miliseconds, you must divide it by 1000 to convert it to seconds.
expression: event_datetime / 1000 - strftime('%s', 'now')
language: sql
```
- `datetime2` - similar to `datetime` but with higher precision. For `datetime2(0-3)` the representation is the same as for `datetime`. For `datetime2(4-6)` it is the number of microseconds since epoch. and for `datetime2(7)` it is the number of nanoseconds since epoch. You can use the same approach as for `datetime` but you need to divide by 1000, 1000000 or 1000000000 depending on the precision.

- `time` - the time of milliseconds since midnight.
```yaml
output:
- uses: redis.write
with:
data_type: hash
expire:
# Convert the time to seconds and subtract the current time in seconds since midnight.
expression: (event_time / 1000.0) -
(
CAST(strftime('%H', 'now') AS INTEGER) * 3600 +
CAST(strftime('%M', 'now') AS INTEGER) * 60 +
CAST(strftime('%S', 'now') AS INTEGER)
)
language: sql
```
- `datetimeoffset` - represented as a timestamp with timezone information, where the timezone is GMT
```yaml
output:
- uses: redis.write
with:
data_type: hash
expire:
# Convert the time to seconds and subtract the current time in seconds since epoch.
expression: strftime('%s', event_datetimeoffset) - strftime('%s', 'now')
language: sql
```

<!-- TODO [ilianiliev-redis]: Test and document the dynamic expressions for the rest of the supported databases - MySQL, PostgresSQL, MongoDB -->
Loading