# Kafka Topic Configuration !!! tip "Kafka Configuration Variability" The defaults and exact list of per-topic configuration options may differ between various Kafka versions. For the most accurate information, refer to the documentation for the specific Kafka version.
Names Default Value Read-Only Sensitive Description
cleanup.policy
log.cleanup.policy		 delete ❌ ❌ The policy to use for log cleanup. The default is 'delete' which removes old log segments. 'compact' is for log compaction.
compression.type producer ❌ ❌ Specify the final compression type for a given topic. Valid values are 'uncompressed', 'zstd', 'lz4', 'gzip', and 'snappy'.
delete.retention.ms
log.cleaner.delete.retention.ms		 86400000 ❌ ❌ The amount of time to retain delete markers in log compacted topics. This is also the time to retain tombstone messages in topics without log compaction.
file.delete.delay.ms
log.segment.delete.delay.ms		 60000 ❌ ❌ The time to wait before deleting a file from the filesystem.
flush.messages
log.flush.interval.messages		 9223372036854775807 ❌ ❌ The number of messages to accept before forcing a flush of data to disk.
flush.ms 9223372036854775807 ❌ ❌ The maximum time in milliseconds to wait before forcing a flush of data to disk.
follower.replication.throttled.replicas ❌ ❌ A list of replica IDs that will be throttled on replication.
index.interval.bytes
log.index.interval.bytes		 4096 ❌ ❌ The interval with which Kafka adds an entry to the offset index.
leader.replication.throttled.replicas ❌ ❌ A list of replica IDs that will be throttled on replication.
local.retention.bytes
log.local.retention.bytes		 -2 ❌ ❌ The maximum size of the log before deleting it.
local.retention.ms
log.local.retention.ms		 -2 ❌ ❌ The maximum time to retain the log before deleting it.
log.cleaner.max.compaction.lag.ms
max.compaction.lag.ms		 9223372036854775807 ❌ ❌ The maximum time a message will remain uncompacted in the log.
max.message.bytes
message.max.bytes		 1048588 ❌ ❌ The largest record batch size allowed by Kafka. Records larger than this will be rejected.
log.message.downconversion.enable
message.downconversion.enable		 true ❌ ❌ Enables or disables the automatic down-conversion of messages to older message formats for consumers with older clients.
log.message.format.version
message.format.version		 3.0-IV1 ❌ ❌ The message format version used for the topic.
log.message.timestamp.after.max.ms
message.timestamp.after.max.ms		 9223372036854775807 ❌ ❌ The maximum difference allowed between the timestamp of a message and the log append time.
log.message.timestamp.before.max.ms
message.timestamp.before.max.ms		 9223372036854775807 ❌ ❌ The maximum difference allowed between the timestamp of a message and the current time before the message is rejected.
log.message.timestamp.difference.max.ms
message.timestamp.difference.max.ms		 9223372036854775807 ❌ ❌ The maximum difference allowed between the timestamp of a message as set by the producer and the log append time.
log.message.timestamp.type
message.timestamp.type		 CreateTime ❌ ❌ Define whether the timestamp in the message is set by the create time or the log append time.
log.cleaner.min.cleanable.ratio
min.cleanable.dirty.ratio		 0.5 ❌ ❌ The minimum ratio of dirty log to total log for log compaction to start.
log.cleaner.min.compaction.lag.ms
min.compaction.lag.ms		 0 ❌ ❌ The minimum time a message will remain uncompacted in the log.
min.insync.replicas 1 ❌ ❌ When a producer sets acks to 'all', the minimum number of replicas that must acknowledge a write for it to be considered successful.
log.preallocate
preallocate		 false ❌ ❌ Should preallocate file segments for this topic.
remote.storage.enable false ❌ ❌ Enable the usage of remote storage for this topic.
log.retention.bytes
retention.bytes		 -1 ❌ ❌ The maximum size of the log before deleting it.
retention.ms 604800000 ❌ ❌ The maximum time to retain the log before deleting it.
log.segment.bytes
segment.bytes		 1073741824 ❌ ❌ The maximum size of a single log segment file before a new log segment is rolled.
log.index.size.max.bytes
segment.index.bytes		 10485760 ❌ ❌ The maximum size of the offset index that Kafka will allow before a new log segment is rolled.
segment.jitter.ms 0 ❌ ❌ The maximum jitter to add to log segment roll time.
segment.ms 604800000 ❌ ❌ The maximum time to retain a log segment before rolling it.
unclean.leader.election.enable false ❌ ❌ Indicates whether unclean leader election is enabled for the topic.
!!! warning "Common Misconfiguration" Setting `min.insync.replicas` equal to `replication.factor` causes write failures during broker maintenance. See [Broker Failures and Fault Tolerance](https://karafka.io/docs/Infrastructure-Broker-Failures-and-Fault-Tolerance.md) for detailed scenarios and recommendations. ## Kafka Topic Configuration / Legend - **Names**: The name of the parameter or setting with its synonyms. - **Default Value**: The initial value assigned to the parameter if not explicitly set. - **Read-Only**: Indicates if the parameter is immutable and cannot be modified. - **Sensitive**: Specifies if the parameter contains sensitive information that will not be accessible or visible using Karafka. - **Description**: A detailed explanation of the parameter's purpose and usage. ## Kafka Topic Configuration / See Also - [Declarative Topics](https://karafka.io/docs/Infrastructure-Declarative-Topics.md) - For declarative topic management in application code - [Kafka Cluster Configuration](https://karafka.io/docs/Kafka-Cluster-Configuration.md) - For cluster-level configuration options - [Admin API](https://karafka.io/docs/Infrastructure-Admin-API.md) - For programmatic topic management operations - [Broker Failures and Fault Tolerance](https://karafka.io/docs/Infrastructure-Broker-Failures-and-Fault-Tolerance.md) - Impact of replication settings on availability during broker failures --- *Last modified: 2026-07-24 11:45:41*