Hitachi

Hitachi Advanced Database Setup and Operation Guide


16.11.4 Index operations (when the multi-node function is being used)

This subsection explains how to use B-tree indexes, text indexes, and range indexes when the multi-node function is being used.

When the multi-node function is being used, the following operations that are unique to the multi-node function are required for handling B-tree indexes, text indexes, and range indexes:

All other operations are the same as when the multi-node function is not used. For details, see 11.3 Index operations.

Organization of this subsection

(1) Creating an index (when the multi-node function is being used)

When the multi-node function is being used, you can execute the commands listed below on the master node only. You cannot execute these commands on a slave node.

Targeted commands
  • adbimport command

  • adbidxrebuild command

  • adbmergechunk command

  • adbunarchivechunk command

For details about these commands, see the manual HADB Command Reference.

(2) Releasing a B-tree index from unfinished status (when the multi-node function is being used)

If abnormal termination of the adbimport or adbidxrebuild command causes a B-tree index to be set to unfinished status, re-execute the command that terminated abnormally, and release the B-tree index from its unfinished status.

If the master node's HADB server terminates abnormally and the master node is switched over, execute the adbidxrebuild command on the new master node, and release the B-tree index from its unfinished status. At this time, execute the adbidxrebuild command with the --create-temp-file option specified.

For details about the --create-temp-file option, see Specification format for the adbidxrebuild command under adbidxrebuild (Rebuild Indexes) in the manual HADB Command Reference.

For details about the unfinished status of B-tree indexes, see 15.9.1 Steps to take when unfinished status is applied to a B-tree index.

(3) Releasing a text index from unfinished status (when the multi-node function is being used)

If abnormal termination of the adbimport or adbidxrebuild command causes a text index to be set to unfinished status, re-execute the command that terminated abnormally, and release the text index from its unfinished status.

If the master node's HADB server terminates abnormally and the master node is switched over, execute the adbidxrebuild command on the new master node, and release the text index from its unfinished status. At this time, execute the adbidxrebuild command with the --create-temp-file option specified.

For details about the --create-temp-file option, see Specification format for the adbidxrebuild command under adbidxrebuild (Rebuild Indexes) in the manual HADB Command Reference.

For details about the unfinished status of text indexes, see 15.10.1 Steps to take when unfinished status is applied to a text index.

(4) Releasing a range index from unfinished status (when the multi-node function is being used)

If abnormal termination of the adbidxrebuild command causes a range index to be set to unfinished status, re-execute the adbidxrebuild command, and release the range index from its unfinished status.

If the master node's HADB server terminates abnormally and the master node is switched over, execute the adbidxrebuild command on the new master node, and release the range index from its unfinished status. At this time, re-execute the adbidxrebuild command with the --create-temp-file option specified.

For details about the --create-temp-file option, see Specification format for the adbidxrebuild command under adbidxrebuild (Rebuild Indexes) in the manual HADB Command Reference.

For details about the unfinished status of range indexes, see 15.11.1 Steps to take when unfinished status is applied to a range index.