pg_upgrade: work around recent test failures
We currently have a bug where a heap partition that is exchanged out of
the partition hierarchy is not given an array type. When this table is
restored in the new database, it will be given an array type -- but
there will be no preassigned OID for it, so it'll steal one from the
pool. If some other object collides with that stolen OID, upgrade will
fail.
Some recent test additions appear to have shifted OIDs around enough to
expose this latent issue, and it's not immediately clear how to fix it,
so add this to the drop script for now so the pipeline isn't blocked.
Co-authored-by: NTaylor Vesely <tvesely@pivotal.io>
Showing
想要评论请 注册 或 登录