Removing Disks from MapR Filesystem

When you remove a disk from MapR Filesystem, other disks in the storage pool are also removed automatically from MapR Filesystem and are no longer in use (they are available but off-line). Their disk storage goes to 0%, and they are eligible to be re-added to MapR Filesystem to build a new storage pool. You can either replace the disk and re-add it along with the other disks that were in the storage pool, or just re-add the other disks if you do not plan to replace the disk you removed. See Adding Disks to MapR Filesystem for more information.

Warning: Removing a disk in the storage pool that contains Container ID 1 will shut down CLDB, triggering a CLDB failover. Container ID 1 contains CLDB data for the master CLDB. From the command-line, run the maprcli disk remove command without the -force 1 option first and examine the warning messages to make sure you aren't removing the disk with Container ID 1. To safely remove such a disk, perform a CLDB Failover to make one of the other CLDB nodes the primary CLDB, then remove the disk as normal.
Before removing or replacing disks, make sure the Replication Alarm (VOLUME_ALARM_DATA_UNDER_REPLICATED) and Data Alarm (VOLUME_ALARM_DATA_UNAVAILABLE) are not raised. These alarms can indicate potential or actual data loss. If either alarm is raised, you might be able to repair the problem using the /opt/mapr/server/fsck utility before removing or replacing disks.
Note: Using the /opt/mapr/server/fsck utility with the -r flag to repair a filesystem risks data loss. Call MapR support before using /opt/mapr/server/fsck -r.

Removing Disks from MapR Filesystem Using the MapR Control System

Complete the following steps to remove disks using the MapR Control System:

  1. Log in to MCS and go to the Summary tab in the node information page.
  2. Select the disks to remove disks in the Disks pane and click Remove Disk(s) from MapR-FS.
    The Remove Disk(s) from MapR-FS confirmation dialog displays.
    Warning: One or more disks you selected may have unreplicated data on it and this action will forcefully remove the disks.
  3. Review the list and click Remove Disk.
    Wait several minutes while the removal process completes. After you remove the disks, any other disks in the same storage pools are taken offline and marked as available (not in use by MapR).
  4. Remove the physical disks from the node or nodes according to the correct hardware procedure.
  5. From a command line terminal, remove the failed disk log file from the /opt/mapr/logs directory.
    These log files are typically named like this:
    diskname.failed.info

Removing Disks from MapR Filesystem Using the CLI or REST API

  1. On the node, determine which disk to remove/replace by examining Disk entries in the /opt/mapr/logs/faileddisk.log file.
  2. Run the following command, substituting the hostname or IP address for <host> and a list of disks for <disks>
    maprcli disk remove -disks <disk names> -host <host>
    Note: This command does not remove a disk containing unreplicated data unless forced.

    For complete reference information, see disk remove.

  3. Examine the screen output in response to the command you ran in step 2.
    For example:
    # maprcli disk remove -host `hostname -f` -disks /dev/sdd
    message   host   disk
    removed.  host1  /dev/sdd
    removed.  host1  /dev/sde
    removed.  host1  /dev/sdf
    Make a note of the additional disks removed when the disk is removed. For example, the disks /dev/sde and /dev/sdf are part of the same storage pool and therefore removed along with the disk (/dev/sdd).
  4. Confirm that the removed disks do not appear in the disktab file.
  5. Remove the disk log file from the /opt/mapr/logs directory.
    For failed disks, these log files are typically named in the pattern diskname.failed.info.

When a failed disk has been replaced, it should be added back to MapR Filesystem along with the other disks from the same storage pool that were previously removed. If you add only the replacement disk to MapR Filesystem, this will result in a non-optimal storage pool layout, which can lead to degraded performance.

Once the disks are added to MapR Filesystem, the cluster allocates properly-sized storage pools automatically. For example, if you add ten disks, MapR allocates two storage pools of three disks each and two storage pools of two disks each.