Cory Zue <czue@dimagi.com> wrote:
> I was able to get the database back to a normal functional state
> using the zero_damaged_pages flag. However, after getting
> everything working and starting to use the database again, I am
> again getting "invalid page header" errors on a certain table.
>
> Does this imply there is a hardware issue on my machine? Is there
> anything else that could be causing this to come back?
In my personal experience bad hardware is the most common cause,
followed by buggy device drivers (where an OS software upgrade
prevented further corruption), followed by using incorrect
procedures for backup, restore, replication setup, or node
promotion. For example, not excluding files under pg_xlog from a
base backup or deleting (or moving) the backup_label file can cause
corruption.
For a more complete discussion, see this blog page:
http://rhaas.blogspot.com/2012/03/why-is-my-database-corrupted.html
--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
