Ilya Eremin
Ilya Eremin
What happens on the last update: Page 4754 is the root of the index (1-level). 240 and 4711 are 0-level pages. Page 4711 splits. 4752 is the new 0-level page...
> IMHO, both scripts should fail on "drop table test", long before commit. I agree but to achieve this we need to compile a procedure/view, update and check dependencies at...
> Yes. What's the problem? Do you suggest to move compiling and updating dependencies from dfw to dsql stage? It's not effective to do these actions at dsql stage. Especially...
> I would also suggest to add a commit where the phase 4 of delete_relation DFW is moved to the phase 6 (modifyRoutine DFW has the phase 5, and exceptions...
OK, thanks.
> However, if the first call of `notify_shutdown()` returns true (which means we already have an exclusive access), then `DBB_shutdown_*` flags are not set properly. This is probably the original...
> if it is `Next` on disk is wrong (less than actual value) The actual value - do you mean it as the number of the last non-active transaction on...
> Yes, exactly. If such corruption is detected, is it possible to fix it safely? Which page can we trust more - the header or the TIP?
Conflicts are solved. Please review.
> As for SS, where the real problem happens, I think it is not good to re-open database file with a (lot of) user connections - it could lead to...