2.53 DUMP 2398

Code

node_id 2398

Symbol

---

Kernel Block

DBLQH

Description.  Dumps information about free space in log part files for the data node with the node ID node_id. The dump is written to the data node out log rather than to the cluster log.

Sample Output.  As written to ndb_6_out.log:

REDO part: 0 HEAD: file: 0 mbyte: 2 TAIL: file: 0 mbyte: 2 total: 256 free: 256 (mb)
REDO part: 1 HEAD: file: 0 mbyte: 2 TAIL: file: 0 mbyte: 2 total: 256 free: 256 (mb)
REDO part: 2 HEAD: file: 0 mbyte: 2 TAIL: file: 0 mbyte: 2 total: 256 free: 256 (mb)
REDO part: 3 HEAD: file: 0 mbyte: 2 TAIL: file: 0 mbyte: 2 total: 256 free: 256 (mb)

Additional Information.  Each line of the output has the following format (shown here split across two lines for legibility):

REDO part: part_no HEAD: file: start_file_no mbyte: start_pos
TAIL: file: end_file_no mbyte: end_pos total: total_space free: free_space (mb)

A data node's redo log is divided into four parts; thus, part_no is always a number between 0 and 3 inclusive. The parts are stored in the data node file system D8, D9, D10, and D11 directories with redo log part 0 being stored in D8, part 1 in D9, and so on (see Section 1.1.2, “NDB Cluster Data Node File System Directory”). Within each directory can be found a DBLQH subdirectory containing NoOfFragmentLogFiles files. The default value for NoOfFragmentLogFiles is 16. The default size of each of these files is 16 MB; this can be changed by setting the FragmentLogFileSize configuration parameter.

start_file_no indicates the number of the file and start_pos the point inside this file in which the redo log starts; for the example just shown, since part_no is 0, this means that the redo log starts at approximately 12 MB from the end of the file D8/DBLQH/S6.FragLog.

Similarly, end_file_no corresponds to the number of the file and end_pos to the point within that file where the redo log ends. Thus, in the previous example, the redo log's end point comes approximately 10 MB from the end of D8/DBLQH/S6.FragLog.

total_space shows the total amount of space reserved for part part_no of the redo log. This is equal to NoOfFragmentLogFiles * FragmentLogFileSize; by default this is 16 times 16 MB, or 256 MB. free_space shows the amount remaining. Thus, the amount used is equal to total_space - free_space; in this example, this is 256 - 254 = 2 MB.

Caution

It is not recommended to execute DUMP 2398 while a data node restart is in progress.