From: Joel E. L. <joe...@ve...> - 2009-09-30 20:24:42
|
On Wed, 2009-09-30 at 12:46 -0700, go...@in... wrote: > Thinks like this have happened to our customers a few times over the > years, and the effects are pretty devastating for our software. > Shouldn't protection from corruption on power loss be built into the > database? Has anything changed in 1.9 to help protect against this? > We've taken to implementing a number of measures to try to mitigate > against catastrophic failures like this, such as periodically doing a > CHECKPOINT, and copying the database off to an autobackup directory > upon startup, but none of these leave a very good taste in my mouth. > > > Thoughts? > Carl Gould > Inductive Automation > [snip] Hi Carl - I actually blame myself (not the software) for the problem because I didn't have a recent backup. I was just hoping that a manual method of recovery might be possible. As an Oracle database specialist, I always keep both physical and logical backups available and have their generation well automated. I have also practiced full disaster recovery on many occasions. I think that anything can get corrupted - including loss of a whole disk or server. I just failed to keep my business rules working in my personal space (with HSQLDB). I know - I should be very ashamed! I believe that HSQLDB does have the ability to provide both logical and physical backups via various kinds of shutdown or checkpoint, as you indicated, followed by file copy to a safe location. If I can dig out from under the current mess, I will make setting up some cron jobs, to accomplish this, my first priority. I just need to think through how best to do things with the database "live". I have started to reconstruct things by using a hex dump followed by ascii conversion. I am trying to reconstruct by re-creating records in an older database using copy and paste from the source. It seems to work but is very tedious. Cheers - Joel > |