Skip to content

Latest commit

 

History

History
53 lines (38 loc) · 1.95 KB

README.md

File metadata and controls

53 lines (38 loc) · 1.95 KB
Raw

Filedump for Cloudberry Database

Filedump is a utility to format Cloudberry Database heap/index/control files into a human-readable form. You can format/dump the files several ways, as listed in the Invocation section, as well as dumping straight binary.

This repo is forked from @greenplum-db/filedump. The original project is based on PostgreSQL Filedump, with modification for support of Greenplum/Cloudberry Database. Refer to the PostgreSQL Filedump wiki for more information.

The original readme of PostgreSQL Filedump is included as README.pg_filedump.

Build

This package only supports building in pgxs mode, which means you will need to have a properly configured install tree (with include files) of the appropriate Cloudberry Database major version.

You can typically build like this:

    # make sure we can find the gpdb package
    source /path/to/greenplum_path.sh

    # build the pg_filedump
    make

The build result is a binary file named pg_filedump, you could copy it to your target machines and run it without the Cloudberry Database binaries.

Cloudberry Database versions

This package supports Cloudberry Database 1.0.0+ (gpdb 7 and onward).

Configuration

There are several configurable options:

  • ENABLE_ZLIB: To use pg_filedump on zlib compressed data files, we need to enable the zlib support. It is automatically enabled if the zlib headers and libraries are detected with pkg-config, but you could force it to be enabled by make -f Makefile.gpdb ENABLE_ZLIB=y, make sure the zlib headers and libraries are available, you may also need to set CFLAGS and LDFLAGS accordingly. It is also possible to force building without zlib by setting this variable to n.

  • ENABLE_ZSTD: A similar setting for the zstd compression method.