You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Consider a topic with 100 partitions, retention.bytes set to 100MB, the default 1GB segment size limit, and a 50GB drive.
We would have to write 100GB of data (well beyond the drive capacity) before any segments roll, and retention rules are only applied to sealed segments.
The result is that we would hit our threshold for disk-nearly-full and start blocking writes.
Solution
When applying retention rules to a topic, we should notice if the system is under disk pressure, and if we have an opportunity to free some space by applying retention rules, then prompt the log to roll now, even if it has not yet reached the configured segment.bytes.
Impact
Users who have configured appropriate retention.bytes settings for their partition count will no longer run the risk of filling up their disk.
Problem
Consider a topic with 100 partitions, retention.bytes set to 100MB, the default 1GB segment size limit, and a 50GB drive.
We would have to write 100GB of data (well beyond the drive capacity) before any segments roll, and retention rules are only applied to sealed segments.
The result is that we would hit our threshold for disk-nearly-full and start blocking writes.
Solution
When applying retention rules to a topic, we should notice if the system is under disk pressure, and if we have an opportunity to free some space by applying retention rules, then prompt the log to roll now, even if it has not yet reached the configured segment.bytes.
Impact
Users who have configured appropriate retention.bytes settings for their partition count will no longer run the risk of filling up their disk.
JIRA Link: CORE-1019
The text was updated successfully, but these errors were encountered: