Large backup size of pg_dump
Hello,
I am using PostgreSQL 18.4 x64 on Windows Server 2022. There is a very
small single database in the cluster.
There are hourly pg_dump backups scheduled and database backup size is
around 10GB.
command line is like below
pg_dump.exe -p 5432 -U dbuser --exclude-table=app -F p -b -c -f "hourly.bak"
When I check the cluster directory size it is 4.1 GB.
Database has one BLOB saved in a single record and it is 16MB in size and
that is in the "app" table which is excluded from the backup file.
I didn't understand about 2.5 times bigger backup sizes than the total
cluster size. I do not know what to check either. Is there a way for me to
make the hourly backup size smaller?
Thanks & Regards,
Ertan
When you store large objects, the actual data resides on pg_largeobject
table (https://www.postgresql.org/docs/current/catalog-pg-largeobject.html).
So your "app" table might not be the only thing to exclude in your dump.
Regards,
Priyanka Chatterjee
On Wed, May 20, 2026 at 9:18 AM Ertan Küçükoglu <ertan.kucukoglu@gmail.com>
wrote:
Show quoted text
Hello,
I am using PostgreSQL 18.4 x64 on Windows Server 2022. There is a very
small single database in the cluster.There are hourly pg_dump backups scheduled and database backup size is
around 10GB.command line is like below
pg_dump.exe -p 5432 -U dbuser --exclude-table=app -F p -b -c -f
"hourly.bak"When I check the cluster directory size it is 4.1 GB.
Database has one BLOB saved in a single record and it is 16MB in size and
that is in the "app" table which is excluded from the backup file.I didn't understand about 2.5 times bigger backup sizes than the total
cluster size. I do not know what to check either. Is there a way for me to
make the hourly backup size smaller?Thanks & Regards,
Ertan
On Wed, May 20, 2026 at 10:17:57AM +0300, Ertan Küçükoglu wrote:
Hello,
I am using PostgreSQL 18.4 x64 on Windows Server 2022. There is a very
small single database in the cluster.There are hourly pg_dump backups scheduled and database backup size is
around 10GB.
1. pg_dump is not the best choice for backups.
2. When using pg_dump, use at least -Fd, and -jX to make the dumps work
in parallel
3. Check what is using the most space in dump, and compare it with db
4. What exactly do you mean by "BLOB"? What is the actual datatype of
the field?
5. What is `\l+ your_db_name` output from psql?
I didn't understand about 2.5 times bigger backup sizes than the total
cluster size. I do not know what to check either. Is there a way for me to
make the hourly backup size smaller?
Consider compressing it? Or use some backup tool that handles
incremental/differential backups, like, for example, backrest.
depesz
On Wed, May 20, 2026 at 3:18 AM Ertan Küçükoglu <ertan.kucukoglu@gmail.com>
wrote:
Hello,
I am using PostgreSQL 18.4 x64 on Windows Server 2022. There is a very
small single database in the cluster.There are hourly pg_dump backups scheduled and database backup size is
around 10GB.command line is like below
pg_dump.exe -p 5432 -U dbuser --exclude-table=app -F p -b -c -f
"hourly.bak"
1. Note that -Fp generates plain SQL files.
2. Where are you specifying the database name? Or is everything going into
"postgres"?
3. No need to specify the default port 5432.
When I check the cluster directory size it is 4.1 GB.
Database has one BLOB saved in a single record and it is 16MB in size and
that is in the "app" table which is excluded from the backup file.
Is 16MB *that* big?
I didn't understand about 2.5 times bigger backup sizes than the total
cluster size. I do not know what to check either. Is there a way for me to
make the hourly backup size smaller?
Taking full backups every hour is suboptimal.
But if you *must*, then
pg_dump -Fp --compress=zstd $db > ${db}.sql.zst
--
Death to <Redacted>, and butter sauce.
Don't boil me, I'm still alive.
<Redacted> lobster!