On Wednesday March 7 2007 3:13 am, Martijn van Oosterhout wrote: > On Wed, Mar 07, 2007 at 02:29:08AM -0700, Ed L. wrote: > > Perhaps my question was not clear enough. Let me rephrase: > > Does the fix for this problem comes from a *fresh* DB > > structure resulting from the initdb, or from a software fix > > in 8.1.8, or both? The answer makes a big difference with a > > 200gb database. If it's in the software alone, we can simply > > restart. If it's in the DB structure, we have to migrate > > 200gb of data from one PGDATA to another. > > In your case you could also get running by creating the > missing file, but whether that's appropriate depends on the > cause. I'd check the release notes to see if any related bugs > have been fixed since then. How would I go about correctly creating the missing file? That sounds appealing, as if it were something I could do without taking downtime. Is it? In reviewing the release notes between 8.1.2 and 8.1.8, there are a number of vacuum fixes along with many others. It is unclear if any of them are related. Given this error is occurring while doing a "VACUUM FREEZE" of the template0 database, I wonder if that creates any options for me? Frankly, I'd be happy to simply drop the template0 DB if I could, since I'm not aware that we ever need it for anything. Ed
