This is an automated email from the ASF dual-hosted git repository.
gfphoenix78 pushed a commit to branch sync-with-upstream
in repository https://gitbox.apache.org/repos/asf/cloudberry-gpbackup.git
The following commit(s) were added to refs/heads/sync-with-upstream by this
push:
new 3a1b2575 fix(test): skip AO table test on Cloudberry due to checksum
persistence issue (#23)
3a1b2575 is described below
commit 3a1b2575b64146deea730433734d1518381bc53e
Author: Robert Mu <[email protected]>
AuthorDate: Wed Aug 27 17:04:13 2025 +0800
fix(test): skip AO table test on Cloudberry due to checksum persistence
issue (#23)
The integration test for creating append-optimized (AO) tables fails
on Cloudberry due to an upstream bug in storage options handling.
Cloudberry shows checksum=true in gp_default_storage_options but fails
to persist this default to pg_class.reloptions when creating tables.
This causes gpbackup to read metadata without the checksum option,
leading to test failures when compared against GPDB 7+ expectations.
The test is temporarily skipped for Cloudberry until the upstream
issue is resolved.
See: https://github.com/apache/cloudberry/issues/1254
---
integration/predata_relations_create_test.go | 25 +++++++++++++++++++++++++
1 file changed, 25 insertions(+)
diff --git a/integration/predata_relations_create_test.go
b/integration/predata_relations_create_test.go
index 0e530b3a..7f36da97 100644
--- a/integration/predata_relations_create_test.go
+++ b/integration/predata_relations_create_test.go
@@ -101,6 +101,31 @@ var _ = Describe("backup integration create statement
tests", func() {
structmatcher.ExpectStructsToMatchExcluding(testTable.TableDefinition,
resultTable.TableDefinition, "ColumnDefs.Oid", "ExtTableDef")
})
It("creates a basic append-optimized column-oriented table",
func() {
+ /*
+ * This test is expected to fail on Cloudberry due to
an inconsistency
+ * in its handling of default table storage options.
+ *
+ * The test logic groups GPDB 7+ and Cloudberry
together, expecting both
+ * to default to `checksum=true` for append-optimized
tables and reflect
+ * this in the table's metadata. While GPDB 7+ behaves
this way,
+ * Cloudberry does not.
+ *
+ * Although Cloudberry's `gp_default_storage_options`
configuration shows
+ * `checksum=true`, this default is not persisted to
`pg_class.reloptions`
+ * when a table is created. `gpbackup` correctly reads
the actual metadata
+ * (which lacks the checksum), causing this test to
fail due to the
+ * mismatch with the hardcoded expectation for GPDB 7+
behavior.
+ */
+
+ // TODO: Remove this skip once Cloudberry fixes the
inconsistency between
+ // gp_default_storage_options configuration and
pg_class.reloptions persistence.
+ // The issue is that checksum=true appears in default
config but is not stored
+ // in table metadata, causing backup verification to
fail due to mismatched
+ // storage options expectations.
+ if connectionPool.Version.IsCBDB() {
+ Skip("Test temporarily disabled for Cloudberry
due to checksum storage option persistence issue")
+ }
+
rowOne := backup.ColumnDefinition{Oid: 0, Num: 1, Name:
"i", NotNull: false, HasDefault: false, Type: "integer", Encoding:
"compresstype=zlib,blocksize=32768,compresslevel=1", StatTarget: -1,
StorageType: "", DefaultVal: "", Comment: ""}
rowTwo := backup.ColumnDefinition{Oid: 0, Num: 2, Name:
"j", NotNull: false, HasDefault: false, Type: "character varying(20)",
Encoding: "compresstype=zlib,blocksize=32768,compresslevel=1", StatTarget: -1,
StorageType: "", DefaultVal: "", Comment: ""}
testTable.StorageOpts = "appendonly=true,
orientation=column, fillfactor=42, compresstype=zlib, blocksize=32768,
compresslevel=1"
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
