[DocDB] [DB Clone] [Colocation] DDLs in cloned db with inheritance tables fail with LookupByIdRpc : Old tablets of table deleted.

[dhruva4869]

Jira Link: DB-16862

Description

Version: 2.27.0.0-b56
Steps:

1. Create parent and child table in colocated db (parent non colocated, child colocated)
2. Create indexes and load data, create snapshot schedule and take time t1
3. do some Alters and drop indexes, create new indexes
4. Load more data
5. Clone to time t1
   Clone passes. However if we try to run SELECT query (SELECT count(*) from employees_parent;) it fails with error:

ERROR:  LookupByIdRpc(tablet: ff537d68a9fe4ed9bc66a6a54256dd1e, num_attempts: 2) failed: Tablet deleted: Old tablets of table 0000401e000030008000000000004008.colocation.parent.uuid deleted at 2025-05-23 06:48:45 UTC

Issue Type

kind/bug

Warning: Please confirm that this issue does not contain any sensitive information

• I confirm this issue does not contain any sensitive information.

[yamen-haddad]

• The issue is related to the clear metacache RPC which is failing in all teservers as per the following logs:

I0604 14:23:56.155977 38652 clone_state_manager.cc:132] Scheduling clear metacache entries task for namespace: 00004001000030008000000000000000 and tserver with UUID: 414fc6ec15d840668c7931fd216b37f0
...
W0604 14:23:56.157450 50059 clone_tasks.cc:130] Clear Metacache entries for namespace 00004001000030008000000000000000 failed: Invalid argument (yb/common/entity_ids.cc:201): Not a YSQL ID string: 94c4d16574504243bb69877df1e54601

• However, we don't abort clone as the only bad behavior is that the user will see an error when running a query for the first time. Afterwards, the metacache should refresh itself in the following runs, and they should succeed.
• The issue happens when YCQL tables are cached in the metacache. Clear Metacache RPC fails as these YCQL table entries are not valid YSQL IDs.
• I will fix the issue by hardening clear metacach RPC to deal with YCQL IDs.