Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.5, 10.6, 10.3(EOL), 10.4(EOL), 10.7(EOL), 10.8(EOL), 10.9(EOL), 10.10(EOL)
-
None
Description
Federated
There was the bug report MDEV-15969 in early days of system versioning saying basically the same. It was closed as "fixed".
As a comment there suggested the discovery now does not add a system versioning clause to the federated table, so the final topology is local (federated) non-versioned <=> remote (underlying) versioned.
But Federated tables are still allowed to be created with an explicit with system versioning clause, and of course they don't work because nothing was fixed in that regard. It's not a weird use case, one might very well want the local table to be system-versioned because only this way the history can be queried. If it isn't supported, it should be disabled explicitly.
Connect
There is still an open bug MDEV-15968. It cannot work with versioning because it doesn't store microseconds, so everything is history there; and of course just like federated, it doesn't pass over for system_time all clause to the remote, so history querying doesn't work either.
Spider
There is a new-ish bug MDEV-28413 about versioned partitioned Spider, that's the only one I've found. It seems complicated, in the essence it's all the same – the Spider table cannot be versioned, as it does not query the history (and, depending on the visibility of columns in the underlying table, may even not be able to do normal updates).
Attachments
Issue Links
- relates to
-
MDEV-15968 System versioning and CONNECT engine don't work well together: current data is not returned
- Confirmed
-
MDEV-15969 System versioning and FEDERATED don't work well together: DML and discovery fail
- Closed
-
MDEV-28413 System versioning is incorrect in Spider
- Stalled