Movatterモバイル変換

DBDICT: Contains data dictionary information. This is stored in:

Directories namedD8,D9,D10, andD11, each of which contains a directory namedDBLQH. These contain the redo log, which is divided into four parts that are stored in these directories. with redo log part 0 being stored inD8, part 1 inD9, and so on.

Within each directory can be found aDBLQH subdirectory containing theN redo log files; these are namedS0.Fraglog,S1.FragLog,S2.FragLog, ...,SN.FragLog, whereN is equal to the value of theNoOfFragmentLogFiles configuration parameter. The default value forNoOfFragmentLogFiles is 16. The default size of each of these files is 16 MB, controlled by theFragmentLogFileSize configuration parameter.

The size of each of the four redo log parts isNoOfFragmentLogFiles * FragmentLogFileSize. You can find out how much space the redo log is using withDUMP 2398 orDUMP 2399.

DBDIH: This directory contains the filePX.sysfile, which records information such as the last GCI, restart status, and node group membership of each node; its structure is defined instorage/ndb/src/kernel/blocks/dbdih/Sysfile.hpp in the NDB Cluster source tree. In addition, theSX.FragList files keep records of the fragments belonging to each table.

The format used for thesysfile was updated from version 1 to version 2 in NDB 8.0.

LCP: When using full local checkpoints (LCPs), this directory holds 2 subdirectories, named0 and1, each of which which contain local checkpoint data files, one per local checkpoint. In NDB 7.6 (and later), when using partial LCPs (EnablePartialLcp set totrue), there can be as many as 2064 subdirectories underLCP, named0,1,2, ...,2063, with a data file stored in each one. These directories are created as needed, in sequential order; for example, if the last data file used in the previous partial LCP was numbered 61 (inLCP/61), the next partial LCP data file is created inLCP/62.

These subdirectories each contain a number of files whose names follow the patternTNFM.Data, whereN is a table ID and M is a fragment number. Each data node typically has one primary fragment and one backup fragment. This means that, for an NDB Cluster having 2 data nodes, and withNoOfReplicas equal to 2,M is either 0 or 1. For a 4-node cluster withNoOfReplicas equal to 2,M is either 0 or 2 on node group 1, and either 1 or 3 on node group 2.

For a partial local checkpoint, a single data file is normally used, but when more than 12.5% of the table rows stored are to be checkpointed up to 8 data files can be used for each LCP. Altogether, there can be from 1 to 2048 data files at any given time.

When usingndbmtd there may be more than one primary fragment per node. In this case,M is a number in the range of 0 to the number of LQH worker threads in the entire cluster, less 1. The number of fragments on each data node is equal to the number of LQH on that node timesNoOfReplicas.