Database Reference
In-Depth Information
# The maximum size of a request that the socket server
# will accept (protection against OOM)
socket.request.max.bytes=104857600
The
log.dirs
property controls Kafka's log output location. These
locations contain a number of partition directories of the form
<topic
name>- <partition number>
and a file called
replication-offset-checkpoint
:
$ ls
analytics-0 audit-1 replication-offset-checkpoint
This text file contains the “high watermark” data described in the earlier
replication section and should not be deleted. If multiple comma-separated
directories are given to Kafka, it distributes partitions evenly across each
of the directories. This is a tempting option for servers where JBOD (Just
a Bunch of Disks) configurations are the default, such as Amazon's EC2.
Unfortunately, Kafka simply distributes partitions randomly to the various
disks rather than trying to, say, distribute partitions of the same topic across
multiple disks. It is often the case that topics have very different data sizes,
and by luck it can happen that different disks have very different usage
characteristics. If possible, it is usually better to use RAID0 to combine the
disks into a single large disk and use a single log directory.
# A comma seperated list of directories under which to
store log files
log.dirs=/tmp/kafka-logs
These next settings control the creation and management of topics.
# Disable auto creation of topics
#auto.create.topics.enable=false
# The number of logical partitions per topic per
server.
# More partitions allow greater parallelism
# for consumption, but also mean more files.
num.partitions=2
