Joe Conway wrote:
> We then spent most of the next 24 hours reviewing the recovered
> database. The bulk data loading process was well instrumented, so we
> knew exactly which data should have been committed prior to the server
> hang, and which files were inprocess (we had been doing 10 loads in
> parallel) at the time of the hang. The results of the investigation
> indicated complete recovery, with no missing or unwanted (i.e.
> uncommitted records looking committed) data.
Good analysis. Seems you were lucky in that only WAL was corrupt and
not the heap or index files.
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
In response to
pgsql-hackers by date
|Next:||From: Tatsuo Ishii||Date: 2004-12-18 23:24:59|
|Subject: Re: Call for port reports|
|Previous:||From: Joe Conway||Date: 2004-12-18 22:28:51|
|Subject: Re: production server down|