mfs.conf

The configuration file /opt/mapr/conf/mfs.conf specifies the following parameters about the MapR Filesystem server on each node.

Warning: You must restart the FileServer after making changes to this file.

Parameters

Parameter

Default Value

Description

mfs.server.ip   IP address of the FileServer. For example, 192.168.10.10.
mfs.server.port 5660 Port used for communication with the server
mfs.cache.lru.sizes

Version 4.0.1:inode:6:log:6:meta:10:dir:40:small:15

Version 4.0.2 and later versions: inode:3:meta:6:small:27:dir: 15:db:20:valc:3

LRU cache configuration. See the section, Notes on LRU Cache Configuration, following this table.
mfs.on.virtual.machine 0 Specifies whether MapR Filesystem is running on a virtual machine
mfs.io.disk.timeout 60 Timeout, in seconds, after which a disk is considered failed and taken offline. This parameter can be increased to tolerate slow disks.
mfs.max.disks 48 Maximum number of disks supported on a single node.
mfs.max.logfile.size.in.mb 1000MB The maximum amount of disk space that the MFS logs can consume before the oldest log file is deleted; based on the following calculation:

maxSizePerLogFile = maxLogSize / MAX_NUM_OF_LOG_FILES

where

  • maxLogSize = total amount of space that MFS log files can consume
  • MAX_NUM_OF_LOG_FILES = total number of MFS log files
mfs.subnets.whitelist   A list of subnets (up to 256 characters) that are allowed to make requests to the FileServer service and access data on the cluster.
mfs.disk.iothrottle.count 100 The maximum number of outstanding requests on disk.
Note: You can disable throttling by setting a high value. This is disabled if the value for mfs.disk.is.ssd is 1.

mfs.disk.resynciothrottle.factor

20

The amount of time to wait before request is submitted to disk or sent over the network. Increasing the value will reduce the wait time and decreasing the value will increase the wait time. For example, setting the value to 40 will halve the wait time and setting the value to 10 will double the wait time.

mfs.network.resynciothrottle.factor 20 Controls the amount of time to wait before sending a resync operation over the network. Increasing the value will reduce the wait time and decreasing the value will increase the wait time. For example, setting the value to 40 will halve the wait time and setting the value to 10 will double the wait time.
mfs.ssd.trim.enabled 0 Set this parameter to 1 to enable TRIM operations for SSD devices.
Note: Enable TRIM only if it is recommended by the SSD vendor.
mfs.disk.is.ssd 0 Specifies whether (1) or not (0) the drives are SSD. If 0, the drives are assumed to be rotations. The default value is 0. If the value is 1, the noop scheduler on the SSD is automatically enabled and IO throttling is disabled.
mfs.mem.debug.enabled 0 Specifies whether fileserver should (1) or should not (0) track all memory allocations. The default value is 0. If value is 1, you can determine the root cause for high memory allocation or determine the component consuming most memory.

Example

mfs.server.ip=192.168.10.10
mfs.server.port=5660
mfs.cache.lru.sizes=inode:3:meta:6:small:27:dir:15:db:20:val:3
mfs.on.virtual.machine=0
mfs.io.disk.timeout=60
mfs.max.disks=48
Notes on LRU Cache Configuration

The cache values are expressed as percentages, which vary based on the expected size of the data the node is required to cache. The goal is to achieve a state in which most of the required data comes directly from the cache. You may need to tune the cache percentages based on your cluster configuration and the workload on specific nodes. Non-default allocations tend to work better for nodes that run only CLDB and nodes that do not have CLDB but do have a heavy MapR Database workload. Note the following recommendations.

  • For CLDB-only nodes, increase the size of the cache for Dir LRU to 40%: change dir:15 to dir:40A CLDB-only node is a fileserver node that hosts only the CLDB volume mapr.cldb.internal (no user volume data is hosted on the node). Dir LRU is used to host B-tree pages.
  • For non-CLDB nodes with no MapR Database workload, optimize the cache to host as many file pages as possible. Change the value of the parameter to: inode:3:meta:6:small:27:dir:6

The remainder of the cache will be used to cache file data pages.

Note: You need to restart MFS in order for the change in mfs.conf to take effect.