MapR 5.0 Documentation : Table-Replication Alarms

You can view table-replication alarms in any of these locations:

MCS

Alarms for replication appear on the Volume Properties window in the MapR Control Service (MCS). They appear on that screen because they are issued per volume rather than per source table. An alarm indicates that one or more source tables in a volume have encountered problems.

To open the Volume Properties window for a volume:

    1. Select Volumes under MapR-FS in the Navigation bar on the left side of the screen.
    2. On the right side of the screen, select one of the listed volumes. The Volume Properties window opens.

maprcli

You can also view the alarms by using the maprcli alarm list command.

Log files

The log files mfs.log-5 and cldb.log also display these alarms. These files are located in the directory /opt/mapr/logs.  

Alarms display the paths and names of the source tables for which the alarms were issued. You can look up the error that is specific to the source table either in MCS or by using the command maprcli table replica list.

Table Replication Error

Logged As

VOLUME_ALARM_TABLE_REPL_ERROR     

Meaning

These alarms display the paths and names for up to ten source tables that have encountered an error. Errors might be caused by the following conditions:

A missing table or column family on the destination cluster

If a column family no longer exists in the replica, pause replication with the maprcli table replica pause command, recreate the column family with the maprcli table cf createcommand, run the CopyTable utility to copy data from the source table into the column family, and then resume replication with the maprcli table replica resume command.

A mismatch in column family names for the table on the destination cluster

If the column family still exists in the replica but the name of the column family was changed, run the maprcli table cf edit command with the parameter -newcfname set to the correct name. Replication will resume automatically.

Unreachable gateways on the destination cluster

This alarm occurs only if none of the gateways on the cluster are reachable.

Table Replication Lag High

Logged As

VOLUME_ALARM_TABLE_REPL_LAG_HIGH

Meaning

These alarms display the paths and names of the source tables for which the replication lag is high.

  • High lag times might be caused by these conditions:
  • High load on the source or replica
  • Low network bandwidth between the source and replicas
  • Miscellaneous error conditions that prevent replication from proceeding
  • Replication explicitly paused on the source cluster

Table Replication Asynchronous

Logged As

VOLUME_ALARM_TABLE_REPL_ASYNC

Meaning

These alarms display the pathnames of the source tables that are involved.

If MapR-DB is replicating synchronously and it judges the latency of the replication stream to be too high, it will switch to asynchronous replication temporarily.

After a new gateway is created, an existing gateway is restarted, or after latency is sufficiently reduced, MapR-DB switches the mode of replication back to synchronous.

You can also check whether a source table is being replicated synchronously or asynchronously by running the command maprcli table replica list.