Re: Read-ahead and parallelism in redo recovery

Aidan Van Dyk wrote:
> How difficult is it to parse the WAL logs with enough knowledge to know
> what heap page (file/offset) a wal record contains (I haven't looked
> into any wal code)?

Unfortunately there's no common format for that. All the heap-related 
WAL records, insert, update and delete, have a 
RelFileNode+ItemPointerData at the beginning of the WAL payload, but 
update records have another ItemPointerData for the tid of the new tuple 
in addition to that. And all indexam WAL records use a format of their own.

It would be nice to refactor that so that there was a common format to 
store the file+block number touched by WAL record. Like we have for full 
page images. That would useful for all kinds of external tools to parse 
WAL files, like the read-ahead restore_command you envisioned.

--   Heikki Linnakangas  EnterpriseDB   http://www.enterprisedb.com