Обсуждение: Can "on delete cascade" dependency be used in pgdump or similar ?
pg 15.15 on linux
I have a DB with a table called "projects" which has a primary key column called "project". Other child tables are linked in a primary/foreign key relationship to "projects" with "on delete cascade". Each of those may have grandchild tables linked in via other primary/foreign key relationships, all with the "on delete cascade". Etc... . If I delete project "a", it'll cascade delete the children, grandchild, etc... recs. All traces of project "a" will be gone.
But I don't really want to lose the data from project "a", I want to archive it in another DB (same DB server, different DB). The brain-numb method I've been using thus far is to copy the whole DB via pgdump then use that to create the DB "a_archive". Then (here's the wasteful part) basically delete all the projects in "a-archive" EXCEPT for project "a". Then, when that's done, go to the main DB and delete project "a". In effect, I just archived all the data for project "a" and put it in the DB called "a_archive".while relieving the main DB of the project "a" data. But what would be really neat is to leverage that cascade on delete stuff to just pgdump project "a" and use that to create "a_archive".
Can pgdump do something like that ?
Is there a better way to approach this problem of archiving one project (remembering that we do have the "on delete cascade" set up) ?
Thanks in Advance !
On Fri, Feb 27, 2026 at 3:40 PM dfgpostgres <dfgpostgres3@gmail.com> wrote:
But what would be really neat is to leverage that cascade on delete stuff to just pgdump project "a" and use that to create "a_archive".Can pgdump do something like that ?
No, it cannot.
Is there a better way to approach this problem of archiving one project (remembering that we do have the "on delete cascade" set up) ?
Nothing built into core. The only additional feature you could leverage is on delete triggers. But you'd need some custom code for maintenance to ensure all relevant tables have the cascade delete and on delete triggers.
David J.
On 2/27/26 2:39 PM, dfgpostgres wrote: > pg 15.15 on linux > > I have a DB with a table called "projects" which has a primary key > column called "project". Other child tables are linked in a primary/ > foreign key relationship to "projects" with "on delete cascade". Each > of those may have grandchild tables linked in via other primary/foreign > key relationships, all with the "on delete cascade". Etc... . If I > delete project "a", it'll cascade delete the children, grandchild, > etc... recs. All traces of project "a" will be gone. > > But I don't really want to lose the data from project "a", I want to > archive it in another DB (same DB server, different DB). The brain-numb > method I've been using thus far is to copy the whole DB via pgdump then > use that to create the DB "a_archive". Then (here's the wasteful part) > basically delete all the projects in "a-archive" EXCEPT for project > "a". Then, when that's done, go to the main DB and delete project "a". > In effect, I just archived all the data for project "a" and put it in > the DB called "a_archive".while relieving the main DB of the project "a" > data. But what would be really neat is to leverage that cascade on > delete stuff to just pgdump project "a" and use that to create "a_archive". > > Can pgdump do something like that ? The best you can do, with pg_dump, is use: https://www.postgresql.org/docs/current/app-pgdump.html --table=pattern Dump only tables with names matching pattern. Multiple tables can be selected by writing multiple -t switches. Though it will be up to you to make sure all the related tables are present in the dump. Also depending on how things are setup you might be able to use logical replication: https://www.postgresql.org/docs/current/logical-replication.html In either case you still have to clean out the original database. > Is there a better way to approach this problem of archiving one project > (remembering that we do have the "on delete cascade" set up) ? > > Thanks in Advance ! > -- Adrian Klaver adrian.klaver@aklaver.com
On 2/27/26 3:13 PM, Adrian Klaver wrote: > On 2/27/26 2:39 PM, dfgpostgres wrote: >> pg 15.15 on linux >> >> I have a DB with a table called "projects" which has a primary key >> column called "project". Other child tables are linked in a primary/ >> foreign key relationship to "projects" with "on delete cascade". Each >> of those may have grandchild tables linked in via other primary/ >> foreign key relationships, all with the "on delete cascade". >> Etc... . If I delete project "a", it'll cascade delete the children, >> grandchild, etc... recs. All traces of project "a" will be gone. >> >> But I don't really want to lose the data from project "a", I want to >> archive it in another DB (same DB server, different DB). The brain- >> numb method I've been using thus far is to copy the whole DB via >> pgdump then use that to create the DB "a_archive". Then (here's the >> wasteful part) basically delete all the projects in "a-archive" EXCEPT >> for project "a". Then, when that's done, go to the main DB and delete >> project "a". In effect, I just archived all the data for project "a" >> and put it in the DB called "a_archive".while relieving the main DB of >> the project "a" data. But what would be really neat is to leverage >> that cascade on delete stuff to just pgdump project "a" and use that >> to create "a_archive". >> >> Can pgdump do something like that ? > > The best you can do, with pg_dump, is use: Forget this idea. -- Adrian Klaver adrian.klaver@aklaver.com
On Fri, Feb 27, 2026 at 5:40 PM dfgpostgres <dfgpostgres3@gmail.com> wrote:
pg 15.15 on linux
[snip]
Is there a better way to approach this problem of archiving one project (remembering that we do have the "on delete cascade" set up) ?
Yes. postgres_fdw is what you want. It will let you:
INSERT INTO archive.projects SELECT * FROM projects WHERE project='a';
INSERT INTO archive.child1 SELECT * FROM child1 WHERE project='a';
INSERT INTO archive.child2 SELECT * FROM child2 WHERE project='a';
INSERT INTO archive.grandchild1 SELECT * FROM archive.grandchild1 WHERE ...
Death to <Redacted>, and butter sauce.
Don't boil me, I'm still alive.
<Redacted> lobster!
On Fri, Feb 27, 2026 at 5:40 PM dfgpostgres <dfgpostgres3@gmail.com> wrote:
pg 15.15 on linuxI have a DB with a table called "projects" which has a primary key column called "project". Other child tables are linked in a primary/foreign key relationship to "projects" with "on delete cascade". Each of those may have grandchild tables linked in via other primary/foreign key relationships, all with the "on delete cascade". Etc... . If I delete project "a", it'll cascade delete the children, grandchild, etc... recs. All traces of project "a" will be gone.But I don't really want to lose the data from project "a", I want to archive it in another DB (same DB server, different DB). The brain-numb method I've been using thus far is to copy the whole DB via pgdump then use that to create the DB "a_archive". Then (here's the wasteful part) basically delete all the projects in "a-archive" EXCEPT for project "a". Then, when that's done, go to the main DB and delete project "a". In effect, I just archived all the data for project "a" and put it in the DB called "a_archive".while relieving the main DB of the project "a" data. But what would be really neat is to leverage that cascade on delete stuff to just pgdump project "a" and use that to create "a_archive".Can pgdump do something like that ?Is there a better way to approach this problem of archiving one project (remembering that we do have the "on delete cascade" set up) ?Thanks in Advance !
Hi,
I think a better way to approach this problem is using "change data capture" using the WAL. Projects such as Debezium [debezium.io] will allow you to process the changes from the server and capture the deletions from tables. You can archive the deleted rows however you like.
--Justin