On Mon, Jul 20, 2015 at 9:18 PM, Alexander Korotkov <a.korotkov@postgrespro.ru> wrote: > attached patch allows pg_rewind to work when target timeline was switched. > Actually, this patch fixes TODO from pg_rewind comments. > > /* > * Trace the history backwards, until we hit the target timeline. > * > * TODO: This assumes that there are no timeline switches on the target > * cluster after the fork. > */ > > This patch allows pg_rewind to handle data directory synchronization is much > more general way. For instance, user can return promoted standby to old > master.
Yes. That would be useful. > In this patch target timeline history is exposed as global variable. Index > in target timeline history is used in function interfaces instead of > specifying TLI directly. Thus, SimpleXLogPageRead() can easily start reading > XLOGs from next timeline when current timeline ends.
The implementation looks rather neat by having a first look at it, but the attached patch fails to compile: pg_rewind.c:488:4: error: use of undeclared identifier 'targetEntry' targetEntry = &targetHistory[i]; ^ Nice to see as well that this has been added to the CF of September.
