Contains the configuration for CLDB nodes.
The file
/opt/mapr/conf/cldb.conf specifies the configuration parameters for
the CLDB nodes and the cluster topology.
- cldb.containers.cache.entries
- Default Value: 1000000
- Description: The maximum number of read/write containers available in
the CLDB cache.
- cldb.default.topologyfileserver
- Default Value:
/data
- Description: The default topology for newly-created volumes.
- cldb.detect.dup.hostid.enabled
- Default Value:
false
- Description: When
true, CLDB disables all nodes
with duplicate hostid, including new nodes that try to register with duplicate
hostid and the existing node. Alarm NODE_ALARM_DUPLICATE_HOSTID is
raised. This case requires administrator intervention to correct the hostid
confusion. If duplicate hostid occurs on nodes running CLDB, the cluster may
fail to start. Therefore, the alarm is not raised, but the
cldb.log file in /opt/mapr/logs/ contains
an error message.
- cldb.enable.memory.tracker
- Default Value:
false
- Description: Utility that monitors CLDB for memory usage and deadlocks.
If
true, memory allocations in CLDB are tracked. If memory
usage of CLDB goes above certain limits, the utility generates core and shuts
down the CLDB. Memory limit is configured as:
Xmx+non heap memory costant(default:3072MB)
You can
change the non-heap memory usage by setting
cldb.memory.max.nonheap.mb to any custom value.
If
CLDB memory usage goes beyond 130% of this limit, the utility dumps and shuts
down CLDB. The default value is false.
- cldb.ignore.posix.only.hb.alarm
- Default Value: 1
- Description: By default, this parameter is set to
1 to
consider all nodes except edge nodes (nodes that have only POSIX clients and
loopback NFS installed) for the No
Heartbeat alarm . Set this parameter to 0 to
include both edge as well as cluster nodes for the No Heartbeat alarm.
Note: The edge nodes that went down before changing this parameter are not
visible in alarms. However, edge nodes that go down after changing this
parameter to
0 will be visible in alarms.
See the
-nfsnodes option of the node list
command to view edge nodes.
- cldb.ignore.stale.zk
- Default Value:
false
- Description: When this setting is
true, the CLDB
ignores the ZooKeeper's information regarding the most recent copy of CLDB
data. Change this setting to true when the ZooKeeper
information is stale. Restart the CLDB with this setting. After the CLDB
starts, change the setting back to false then restart the CLDB
again.Only change this setting on CLDB nodes that are known to have the
most recent copy of the CLDB data. Shut down all CLDB processes before
changing this variable.
- cldb.jmxremote.port
- Default Value: 7220
- Description: The CLDB JMX remote port
- cldb.max.security.policies
- Default Value: 10000
- Description: Defines the maximum number of configured security
policies. To prevent users from arbitrarily creating numerous security policies
and draining CLDB performance, the maximum number of security policies is
limited to 10000 by default.
- cldb.min.fileservers
- Default Value: 1
- Description: Number of file servers that must register with the CLDB
before the root volume is created.
- cldb.numthreads
- Default Value: 10
- Description: The number of threads reserved for use by the CLDB.
- cldb.pbs.global.master
- Default Value: 0
- Description: Indicates the global primary cluster for the global
namespace. Only the global primary security policy cluster can create/modify
security policies. All other secondary security policy clusters can only view
or import security policies.
- cldb.port
- Default Value: 7222
- Description: The port on which the CLDB listens.
- cldb.security.blacklist.cleanup.duration.seconds
- Default Value: 36000
- Description: Ticket blacklist cleanup interval.
- cldb.security.resolve.user
- Default Value: 0
- Description: Resolve UID:GID on client OS or on CLDB.
- cldb.security.user.ticket.duration.seconds
- Default Value: 1209600
- Description: Default ticket duration
- cldb.security.user.ticket.max.duration.seconds
- Default Value: 2592000
- Description: Maximum ticket duration
- cldb.security.user.ticket.renew.duration.seconds
- Default Value: 2592000
- Description: Default ticket renew duration
- cldb.security.user.ticket.renew.max.duration.seconds
- Default Value: 7776000
- Description: Maximum ticket renew duration
- cldb.snap.cntr.count.alarm.threshold
- Default Value: 100000000
- Description: The threshold (in minutes) for raising the
CLUSTER_ALARM_TOO_MANY_SNAPSHOT_CONTAINERS alarm.
- cldb.snap.cntr.count.disable.threshold
- Default Value: 128000000
- Description: The maximum number of snapshots to allow before disabling
snapshot creation.
- cldb.snap.cntr.count.monitor.interval.minutes
- Default Value: 60
- Description: The interval of time (in minutes) to elapse between
checking the number of snapshots on the cluster.
- cldb.v2.features.enabled
- Default Value: 1
- Description: Enables new features added in
data-fabric version 2.0. Used only during the
upgrade process from v1.x to 2.x to control when new features become active.
Once enabled, cannot be disabled.
- cldb.v3.features.enabled
- Default Value: 1
- Description: Enables new features added in
data-fabric version 3.0. Used only during the
upgrade process from a pre-3.0 version to control when new features become
active. Once enabled, cannot be disabled.
- cldb.web.port
- Default Value: 7221
- Description: The port that the CLDB uses for the webserver.
- cldb.zookeeper.servers
- Default Value: Not Applicable
- Description: The nodes that are running ZooKeeper, in the format
\<host:port\>.
- hadoop.version
- Default Value: Not Applicable
- Description: The version of Hadoop supported by the cluster.
- net.topology.script.file.name
- Default Value: Not Applicable
- Description: The path to a script that associates IP addresses with
physical topology paths. The script takes the IP address of a single node as
input and returns the physical topology that should be associated with the
specified node. This association is used only at the time a node is initially
added to the cluster. To change topology for nodes already in the cluster, use
the
maprcli node move
command.
- net.topology.table.file.name
- Default Value: Not Applicable
- Description: The path to a text file that associates IP addresses with
physical topology paths. Each line of the text file is of format
<hostname/ip> <rack>, with the IP address or hostname
of one node, followed by the topology to associate with the node. This
association is used only at the time a node is initially added to the cluster.
To change topology for nodes already in the cluster, use the maprcli
node move
command.
- num.volmirror.threads
- Default Value: 1
- Description: The number of (volume mirror) threads to create to process
mirroring requests. The specified number of threads will be created to process
requests in parallel; the remaining requests will be in the queue till they are
picked up by volume mirror thread.
Example cldb.conf file
#
# CLDB Config file.
# Properties defined in this file are loaded during startup
# and are valid for only CLDB which loaded the config.
# These parameters are not persisted anywhere else.
#
# Wait until minimum number of fileserver register with
# CLDB before creating Root Volume
cldb.min.fileservers=1
# CLDB listening port
cldb.port=7222
# Number of worker threads
cldb.numthreads=10
# CLDB webport
cldb.web.port=7221
# CLDB https port
cldb.web.https.port=7443
# Disable duplicate hostid detection
cldb.detect.dup.hostid.enabled=false
# Deprecated: This param is no longer supported. To configure
# the container cache, use the param cldb.containers.cache.percent
# Number of RW containers in cache
#cldb.containers.cache.entries=1000000
#
# Percentage (integer) of Xmx setting to be used for container cache
#cldb.containers.cache.percent=20
#
#Frequency of the heartbeat interval
# Topology script to be used to determine
# Rack topology of node
# Script should take an IP address as input and print rack path
# on STDOUT. eg
# $>/home/mapr/topo.pl 10.10.10.10
# $>/mapr-rack1
# $>/home/mapr/topo.pl 10.10.10.20
# $>/mapr-rack2
#net.topology.script.file.name=/home/mapr/topo.pl
#
# Topology mapping file used to determine
# Rack topology of node
# File is of a 2 column format (space separated)
# 1st column is an IP address or hostname
# 2nd column is the rack path
# Line starting with '#' is a comment
# Example file contents
# 10.10.10.10 /mapr-rack1
# 10.10.10.20 /mapr-rack2
# host.foo.com /mapr-rack3
#net.topology.table.file.name=/home/mapr/topo.txt
#
# ZooKeeper address
cldb.zookeeper.servers=10.10.82.22:5181
# Hadoop metrics jar version
hadoop.version=2.7.0
# CLDB JMX remote port
cldb.jmxremote.port=7220
num.volmirror.threads=1
# Set this to set the default topology for all volumes and nodes
# The default for all volumes is /data by default
# UNCOMMENT the below to change the default topology.
# For e.g., set cldb.default.topology=/mydata to create volumes
# in /mydata topology and to place all nodes in /mydata topology
# by default
#cldb.default.topology=/mydata
enable.replicas.invariant.check=false