OSS tries to duplicate hdb_event_log_cleanups on startup

Version Information

Server Version: v2.16.0

Environment

OSS

What is the current behavior?

They are running Cloud Enterprise in parallel with OSS v2.16.0 Both instances (Cloud E & OSS) work using the same metadata file after manually applying. Error occurs when OSS version restarts it then goes into a restart loop with the error that hdb_catalog.hdb_event_log_cleanups table already exists and is trying to be creating again on startup.

What is the expected behavior?

Both instances (Cloud E & OSS) should be able to run using the same metadata file and the local OSS version should be able to restart without error.

How to reproduce the issue?

1. Metadata change occurs.
2. Changed metadata is applied to OSS v2.16.0.
3. OSS v2.16.0 container is manually restarted.
4. restart loop begins, each time throwing duplicate hdb_event_log_cleanups error.

Screenshots or Screencast

Please provide any traces or logs that could help here.

{"detail":{"info":{"code":"unexpected","error":"database query error","internal":{"arguments":[],"error":{"description":null,"exec_status":"FatalError","hint":null,"message":"relation \"hdb_event_log_cleanups\" already exists","status_code":"42P07"},"prepared":false,"statement":"CREATE TABLE hdb_catalog.hdb_event_log_cleanups\n(\n id TEXT DEFAULT hdb_catalog.gen_hasura_uuid() PRIMARY KEY,\n trigger_name TEXT NOT NULL,\n scheduled_at TIMESTAMP NOT NULL,\n deleted_event_logs INTEGER,\n deleted_event_invocation_logs INTEGER,\n status TEXT NOT NULL,\n CHECK (status IN ('scheduled', 'paused', 'completed', 'dead')),\n\n UNIQUE (trigger_name, scheduled_at)\n);\n\nALTER TABLE hdb_catalog.event_invocation_logs\nADD COLUMN trigger_name TEXT;\n\nALTER TABLE hdb_catalog.event_invocation_logs\nDROP CONSTRAINT IF EXISTS event_invocation_logs_event_id_fkey;\n"},"path":"$"},"kind":"catalog_migrate"},"level":"error","timestamp":"2022-12-08T15:14:15.572+0000","type":"startup"}

Any possible solutions/workarounds you’re aware of?

Currently the workaround is to only use Cloud directly, or to not restart the OSS version. This isn’t feasible for production use.

Keywords

hdb_catalog hdb_event_log_cleanups hdb_catalog.hdb_event_log_cleanups restart

[dsandip]

@adas98012 Could you please try and repro this?

[adas98012]

I updated the metadata json file (removed the backend_configs) and then imported it into the OSS project. I don’t see the event trigger (created in the Cloud Standard project) in there.