On 2016-01-19 15:34:48 +0000, postgresql.org@gclough.com wrote:
> That final message was mis-interpreted as corruption in the WAL, but without
> knowing the internals of either the WAL or pg_xlogdump I can't say for sure
> that it's not corruption. It's repeatable, so I doubt it's a problem in
> PostgreSQL core.
>
> Could someone confirm my understanding, and if so, then could we please get
> pg_xlogdump updated to recognize an end-of-timeline?
Yes, that's pretty much correct. The "old" timeline doesn't even really
know about it, so it's not generally clear how we go forward. There
could be WAL from several parallel timelines present.
You can manually choose the timeline you want to be displayed. Either by
specifying -t or by specifying files from the other timeline.
I don't think we really need to update pg_xlogdump to do any of this
automatically - for which use case is that required?
Regards,
Andres