Support for Apache Cloudberry and usage guidance in gpdemo environment

[tuhaihe]

Hi,

I am from the Apache Cloudberry community. Apache Cloudberry is an open-source MPP database originally derived from Greenplum, but it has since evolved independently with a more modernized kernel and ongoing development.

First of all, thank you for creating this project. From my understanding, gpbackman provides an open-source solution similar to a gpbackup-manager for managing and inspecting gpbackup backups. This is very valuable for gpbackup users. Please correct me if my understanding is inaccurate.

I am currently trying to use gpbackman in a Cloudberry gpdemo environment. I was able to successfully build gpbackman, but I encountered an error when running the command:

[gpadmin@cdw gpbackman]$ ./gpbackman report-info \
  --backup-dir /home/gpadmin/cloudberry/gpAux/gpdemo/datadirs/qddir/demoDataDir-1/backups/20260225/ \
  --timestamp 20260225190002

20260225:23:20:06 gpbackman:gpadmin:cdw:081329-[ERROR]:-Unable to get info for backup 20260225190002. Error: no such table: backups

The backup files exist at the expected location:

$ ls ~/cloudberry/gpAux/gpdemo/datadirs/qddir/demoDataDir-1/backups/
20260225

$ ls ~/cloudberry/gpAux/gpdemo/datadirs/qddir/demoDataDir-1/backups/20260225/
20260225190002

I would like to ask:

• Is gpbackman expected to work directly with gpbackup backups created in a gpdemo environment?

In addition, I would like to ask whether you might be interested in supporting Apache Cloudberry. Since Cloudberry uses gpbackup (and our community maintains a fork at https://github.com/apache/cloudberry-backup), it would be great to explore compatibility or integration.

We would also be happy to collaborate or help test if needed.

Thanks again for creating this useful project!

[woblerr]

Hi, thank's for interesting to this project.

You absolutely right, gpbackman (GP backup manager) is designed as open-source alternative for gpbackup-manager.

The error was caused because the history database was not specified via the --history-db option. By default, gpbackman looks for a SQLite database named "gpbackup_history.db" in the current working directory to store backup metadata. Since this database does not exist, it cannot find the "backups" table, resulting in the error.

gpbackman should work correctly while db schema of gpbackup history database is compatible with Greenplum. If in the future Cloudberry's gpbackup history database schema diverges significantly from Greenplum's, then gpbackman may require modifications to support it.

I quickly tested gpbackman with Sandbox of Apache Cloudberry and it works.

$ psql -tc "select version();"
 PostgreSQL 14.4 (Apache Cloudberry 3.0.0-devel+dev.2615.ge304d3eb67b build dev) on aarch64-unknown-linux-gnu, compiled by gcc (GCC) 11.5.0 20240719 (Red Hat 11.5.0-11), 64-bit compiled on Mar  1 2026 13:2
2:46

$ gpbackup -version
gpbackup version 2.1.0
$ gpbackman -v
gpbackman version master-56830c5

Some local backups were created:

$ ls  /data0/database/coordinator/gpseg-1/backups/20260301
20260301133940	20260301135506	20260301135517	20260301135522	20260301135533	20260301135537	20260301135548

Examples of using gpbackman:

$ gpbackman  backup-info --history-db /data0/database/coordinator/gpseg-1/gpbackup_history.db  --failed --deleted
 TIMESTAMP      | DATE                     | STATUS  | DATABASE | TYPE          | OBJECT FILTERING | PLUGIN | DURATION | DATE DELETED             
----------------+--------------------------+---------+----------+---------------+------------------+--------+----------+--------------------------
 20260301135548 | Sun Mar 01 2026 13:55:48 | Success | gpadmin  | full          |                  |        | 00:00:02 |                          
 20260301135537 | Sun Mar 01 2026 13:55:37 | Success | gpadmin  | incremental   |                  |        | 00:00:02 |                          
 20260301135533 | Sun Mar 01 2026 13:55:33 | Success | gpadmin  | full          |                  |        | 00:00:02 |                          
 20260301135517 | Sun Mar 01 2026 13:55:17 | Success | gpadmin  | data-only     |                  |        | 00:00:01 |                          
 20260301135302 | Sun Mar 01 2026 13:53:02 | Success | gpadmin  | metadata-only |                  |        | 00:00:01 | Sun Mar 01 2026 13:56:42 
 20260301133940 | Sun Mar 01 2026 13:39:40 | Success | gpadmin  | metadata-only |                  |        | 00:00:01 |                          

$ gpbackman  report-info --timestamp 20260301135548 --history-db /data0/database/coordinator/gpseg-1/gpbackup_history.db  
Apache Cloudberry Backup Report

timestamp key:         20260301135548
database version:      PostgreSQL 14.4 (Apache Cloudberry 3.0.0-devel+dev.2615.ge304d3eb67b build dev) on aarch64-unknown-linux-gnu, compiled by gcc (GCC) 11.5.0 20240719 (Red Hat 11.5.0-11), 64-bit compiled on Mar  1 2026 13:22:46
gpbackup version:      2.1.0

database name:         gpadmin
command line:          gpbackup -dbname gpadmin --leaf-partition-data
compression:           gzip
plugin executable:     None
backup section:        All Sections
object filtering:      None
includes statistics:   No
data file format:      Multiple Data Files Per Segment
incremental:           False

start time:            Sun Mar 01 2026 13:55:48
end time:              Sun Mar 01 2026 13:55:50
duration:              0:00:02

backup status:         Success

database size:         84 MB
segment count:         3

count of database objects in backup:
access methods               0
aggregates                   0
casts                        0
collations                   0
constraints                  0
conversions                  0
default privileges           0
database gucs                0
event triggers               0
extensions                   1
functions                    0
indexes                      0
operator classes             0
operator families            0
operators                    0
policies                     0
procedural languages         0
protocols                    0
resource groups              3
resource queues              1
roles                        1
rules                        0
statistics_ext               0
storage server               0
storage user mapping         0
schemas                      1
sequences                    0
tables                       2
tablespaces                  0
text search configurations   0
text search dictionaries     0
text search parsers          0
text search templates        0
transforms                   0
triggers                     0
types                        0
views                        0

It would be great to have more deeper testing gpbackman with Cloudberry. Please let me know if you have any questions or need any help.

[tuhaihe]

Cool! I can run the gpbackman following your guide! Thanks!

gpadmin@cdw:~/gpbackman$ ./gpbackman backup-info --history-db ~/cloudberry/gpAux/gpdemo/datadirs/qddir/demoDataDir-1/gpbackup_history.db --failed --deleted 
 TIMESTAMP      | DATE                     | STATUS  | DATABASE | TYPE          | OBJECT FILTERING | PLUGIN | DURATION | DATE DELETED 
----------------+--------------------------+---------+----------+---------------+------------------+--------+----------+--------------
 20260303060800 | Tue Mar 03 2026 06:08:00 | Success | postgres | metadata-only |                  |        | 00:00:02 |              
 20260303060757 | Tue Mar 03 2026 06:07:57 | Success | postgres | metadata-only |                  |        | 00:00:02 |              
 20260303060557 | Tue Mar 03 2026 06:05:57 | Success | postgres | metadata-only |                  |        | 00:00:02 |              
 20260303060549 | Tue Mar 03 2026 06:05:49 | Success | postgres | metadata-only |                  |        | 00:00:02 |              

gpadmin@cdw:~/gpbackman$ ./gpbackman backup-info --history-db ~/cloudberry/gpAux/gpdemo/datadirs/qddir/demoDataDir-1/gpbackup_history.db --deleted 
 TIMESTAMP      | DATE                     | STATUS  | DATABASE | TYPE          | OBJECT FILTERING | PLUGIN | DURATION | DATE DELETED 
----------------+--------------------------+---------+----------+---------------+------------------+--------+----------+--------------
 20260303060800 | Tue Mar 03 2026 06:08:00 | Success | postgres | metadata-only |                  |        | 00:00:02 |              
 20260303060757 | Tue Mar 03 2026 06:07:57 | Success | postgres | metadata-only |                  |        | 00:00:02 |              
 20260303060557 | Tue Mar 03 2026 06:05:57 | Success | postgres | metadata-only |                  |        | 00:00:02 |              
 20260303060549 | Tue Mar 03 2026 06:05:49 | Success | postgres | metadata-only |                  |        | 00:00:02 |              

gpadmin@cdw:~/gpbackman$ ./gpbackman report-info --timestamp 20260303060800 --history-db ~/cloudberry/gpAux/gpdemo/datadirs/qddir/demoDataDir-1/gpbackup_history.db
Apache Cloudberry Backup Report

timestamp key:         20260303060800
database version:      PostgreSQL 14.4 (Apache Cloudberry 2.1.0-incubating build bdf90c5518f) on aarch64-unknown-linux-gnu, compiled by gcc (Ubuntu 11.4.0-1ubuntu1~22.04.2) 11.4.0, 64-bit compiled on Mar  3 2026 03:32:41
gpbackup version:      2.1.0

database name:         postgres
command line:          gpbackup --dbname postgres
compression:           gzip
plugin executable:     None
backup section:        Metadata Only
object filtering:      None
includes statistics:   No
data file format:      No Data Files
incremental:           False

start time:            Tue Mar 03 2026 06:08:00
end time:              Tue Mar 03 2026 06:08:02
duration:              0:00:02

backup status:         Success

database size:         80 MB
segment count:         3

count of database objects in backup:
access methods               0
aggregates                   0
casts                        0
collations                   0
constraints                  0
conversions                  0
default privileges           0
database gucs                0
event triggers               0
extensions                   1
functions                    0
indexes                      0
operator classes             0
operator families            0
operators                    0
policies                     0
procedural languages         0
protocols                    0
resource groups              3
resource queues              1
roles                        1
rules                        0
statistics_ext               0
storage server               0
storage user mapping         0
schemas                      1
sequences                    0
tables                       0
tablespaces                  0
text search configurations   0
text search dictionaries     0
text search parsers          0
text search templates        0
transforms                   0
triggers                     0
types                        0
views                        0

[tuhaihe]

Thanks again for your support @woblerr — with your guidance, I was able to successfully run gpbackman in the Cloudberry gpdemo environment. Really appreciate your help!

If you have time, I was wondering whether you would be open to a short online meeting? It would be great to briefly introduce ourselves and exchange some thoughts about potential collaboration or future ideas around gpbackman and Apache Cloudberry.

It doesn’t need to be long — around 30 minutes would be perfectly fine.

If convenient, you can pick a suitable time here:
https://calendly.com/wangdianjin/45min

Looking forward to further discussion, and thanks again for building this useful project.

[woblerr]

@tuhaihe
I've added gpbackman to the cloudberry-backup project as a POC - https://github.com/woblerr/cloudberry-backup/tree/add_gpbackman

In this implementation gpbackman t provides the following commands:

• backup-info — display information about existing backups (with filtering by type, tables, schemas, etc.)
• backup-delete — delete specific backups by timestamp (supports cascade deletion for incremental chains, plugin-based backups)
• backup-clean — bulk delete backups older than a given condition (by timestamp or age in days)
• history-clean — clean deletion history from the gpbackup_history.db
• report-info — display backup reports (local and plugin-based)

history-migrate command hasn't been ported, there is no need to support YAML db format.

Unit tests and e2e pass.

Unit tests:

make unit
TEST_DB_TYPE=CBDB TEST_DB_VERSION=2.999.0 ginkgo -r --keep-going --randomize-suites --randomize-all --no-color backup/ filepath/ history/ helper/ options/ report/ restore/ toc/ utils/ testutils/ plugins/s3plugin/ gpbackman/cmd/ gpbackman/gpbckpconfig/ gpbackman/textmsg/ 2>&1
[1774216441] backup tests - 585/586 specs •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••SS••••S••••••••••••••••••••••••••••••••••••••••••••S•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••S•••••••••••••••••••••••••••••P•••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••S••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• SUCCESS! 707.696916ms PASS
[1774216441] Cmd Suite - 14/14 specs •••••••••••••• SUCCESS! 3.151542ms PASS
[1774216441] Textmsg Suite - 14/14 specs •••••••••••••• SUCCESS! 424.542µs PASS
[1774216441] testutils tests - 8/8 specs •••••••• SUCCESS! 705.042µs PASS
[1774216441] Report Suite - 34/34 specs •••••••••••••••••••••••••••••••••• SUCCESS! 11.828042ms PASS
[1774216441] utils tests - 118/118 specs •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• SUCCESS! 63.025375ms PASS
[1774216441] History Suite - 8/8 specs •••••••• SUCCESS! 15.495916ms PASS
[1774216441] Filepath Suite - 31/31 specs ••••••••••••••••••••••••••••••• SUCCESS! 3.961417ms PASS
[1774216441] Gpbckpconfig Suite - 47/47 specs ••••••••••••••••••••••••••••••••••••••••••••••• SUCCESS! 4.415333ms PASS
[1774216441] Options Suite - 27/27 specs ••••••••••••••••••••••••••• SUCCESS! 8.07025ms PASS
[1774216441] restore tests - 118/118 specs •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• SUCCESS! 104.096167ms PASS
[1774216441] TOC Suite - 45/45 specs ••••••••••••••••••••••••••••••••••••••••••••• SUCCESS! 3.393708ms PASS
[1774216441] s3_plugin tests - 32/32 specs •••••••••••••••••••••••••••••••• SUCCESS! 4.023833ms PASS

Ginkgo ran 13 suites in 15.574166458s
Test Suite Passed

End to end for gpbackman:

ginkgo -r --keep-going --randomize-suites --randomize-all --no-color   --timeout=1h --poll-progress-after=0s --focus "gpbackman" end_to_end   -- --custom_backup_dir /tmp 2>&1

Running Suite: EndToEnd Suite - /tmp/cloudberry-backup/end_to_end
=================================================================
Random Seed: 1774215116 - will randomize all specs

Will run 32 of 221 specs
SS•S•SSSSSS•SSSS•SSSSSSSSSSSSSSSS•SSSSSSS•SSS•SSS•SS•SSSSSSSSSSSSSSSS•SSSSSSSSSSSSSSSS••SSS•S•S•SSSSS•SSS•••SSSS•SSSSSSSS••SSSSSSSSSSSSSSSSSSSSSSS•SSSSSSSSSS•SSSSS•S•SSSSSSSSSSSSSSSSSSSSS•S•S•SSSSSSSSSSSS•SSSSSSSS•SSSS•SS

Ran 32 of 221 Specs in 160.312 seconds
SUCCESS! -- 32 Passed | 0 Failed | 0 Pending | 189 Skipped
PASS

Ginkgo ran 1 suite in 2m41.333482531s
Test Suite Passed

Some examples.

$ gpbackman --history-db /data/master/gpseg-1/gpbackup_history.db backup-info --failed --deleted --detail
    TIMESTAMP    |           DATE           | STATUS  | DATABASE |     TYPE      | OBJECT FILTERING | PLUGIN | DURATION |       DATE DELETED       |        OBJECT FILTERING DETAILS          
-----------------+--------------------------+---------+----------+---------------+------------------+--------+----------+--------------------------+------------------------------------------
  20260322220743 | Sun Mar 22 2026 22:07:43 | Success | demo     | full          | include-table    |        | 00:00:01 |                          | public.test_initialization               
  20260322220558 | Sun Mar 22 2026 22:05:58 | Success | testdb   | incremental   |                  |        | 00:00:01 | Sun Mar 22 2026 22:08:25 |                                          
  20260322220553 | Sun Mar 22 2026 22:05:53 | Success | testdb   | incremental   |                  |        | 00:00:01 |                          |                                          
  20260322220443 | Sun Mar 22 2026 22:04:43 | Success | demo     | metadata-only |                  |        | 00:00:01 |                          |                                          
  20260322220416 | Sun Mar 22 2026 22:04:16 | Success | testdb   | full          |                  |        | 00:00:02 |                          |                                          
  20260322220405 | Sun Mar 22 2026 22:04:05 | Success | demo     | full          |                  |        | 00:00:02 |                          |                                          
  20260322220235 | Sun Mar 22 2026 22:02:35 | Failure | testdb   | full          | include-table    |        | 00:00:01 |                          | schemaone.measurement_peaktemp_catchall 


$ gpbackman --history-db /data/master/gpseg-1/gpbackup_history.db report-info --timestamp 20260322220743
Apache Cloudberry Backup Report

timestamp key:         20260322220743
database version:      PostgreSQL 14.4 (Apache Cloudberry 2.1.0-incubating-rc2 build dev) on aarch64-unknown-linux-gnu, compiled by gcc (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0, 64-bit compiled on Mar 22 2026 19:56:50
gpbackup version:      2.1.0

database name:         demo
command line:          gpbackup --dbname demo --leaf-partition-data --include-table public.test_initialization
compression:           gzip
plugin executable:     None
backup section:        All Sections
object filtering:      Include Table Filter
includes statistics:   No
data file format:      Multiple Data Files Per Segment
incremental:           False

start time:            Sun Mar 22 2026 22:07:43
end time:              Sun Mar 22 2026 22:07:44
duration:              0:00:01

backup status:         Success

segment count:         4

count of database objects in backup:
constraints          1
conversions          0
default privileges   0
event triggers       0
indexes              0
policies             0
rules                0
statistics_ext       0
sequences            0
tables               1
triggers             0
views                0

I would appreciate it if you could do some additional testing on your side — especially around different backup scenarios and environments.
If everything looks good, I can add docs and open a PR to the main Apache cloudberry-backup repository.

[tuhaihe]

@tuhaihe
I've added gpbackman to the cloudberry-backup project as a POC - https://github.com/woblerr/cloudberry-backup/tree/add_gpbackman

Great! Thanks for your excellent work! Will test it later. If anything needs improvement, I will let you know.

[tuhaihe]

Thanks again, @woblerr, for the great work!

I ran both the unit tests and the end-to-end tests successfully on the Cloudberry demo cluster, and the gpbackman-related tests also worked well on my side.

I think this looks ready for a PR. I will also invite other cloudberry-backup engineers to review it and share feedback in the new PR. Does that sound good to you?

Thanks again!

[woblerr]

@tuhaihe
I opened a PR apache/cloudberry-backup#79.

It makes sense to continue further discussion there. I'll be waiting for the review.

[tuhaihe]

@tuhaihe I opened a PR apache/cloudberry-backup#79.

It makes sense to continue further discussion there. I'll be waiting for the review.

Cool! Let's discuss there. Thanks again!

[woblerr]

PR is merged. Closing this issue.