Ron <ronljohnsonjr@gmail.com> writes:
> I can associate these dat names with their source tables through a bunch of
> bash and vim manual operations, but I was wondering if there's any automated
> method (maybe some SQL query of some catalog table; pg_class didn't seem to
> have the relevant data) of making the association.
Those numbers are the "dump object IDs" generated by pg_dump. They don't
have any significance on the server side, and typically would vary from
one pg_dump run to another. You have to look at the dump TOC (table of
contents) to figure out what corresponds to what. For example,
$ pg_dump -Fd -f dumpd regression
$ ls -1 dumpd
6143.dat.gz
6144.dat.gz
6145.dat.gz
...
blob_3001.dat.gz
blobs.toc
toc.dat
$ pg_restore -l dumpd
;
; Archive created at 2018-09-02 22:14:48 EDT
...
6573; 2613 119655 BLOB - 119655 postgres
6574; 2613 3001 BLOB - 3001 postgres
6603; 0 0 COMMENT - LARGE OBJECT 3001 postgres
6247; 0 100933 TABLE DATA public a postgres
6212; 0 89417 TABLE DATA public a_star postgres
6180; 0 88516 TABLE DATA public abstime_tbl postgres
6218; 0 89444 TABLE DATA public aggtest postgres
6446; 0 121383 TABLE DATA public alter_table_under_transition_tables postgres
...
The numbers before the semicolons are the dump IDs. In particular
$ pg_restore -l dumpd | grep 6143
6143; 0 88018 TABLE DATA public int4_tbl postgres
so 6143.dat.gz contains the data for table public.int4_tbl. There
will only be separate files in the dump directory for TABLE DATA and
BLOB dump objects ... other stuff is just embedded in the toc.dat file.
regards, tom lane
