I am a little confused by the dblog situation. I am working with a
subversion repository that has been running for about a year. I recently
had a wedging situation and had to do a recover from the admin. It took
about 8 hours on good hardware. This makes me think that it recovered from
revision 0.
I'm a former perforce administrator and am used to the model of journaling
with checkpoints. I start a db and a transaction log capture all
transaction. At a certain point (a nightly backup perhaps), I make a
checkpoint and start a new journal file. I can now reconstitute my db from
the beginning using the two journal files or from day 2 using the checkpoint
and the journal file. If I am 600 days in all I need have is checkpoint.599
and the current journal to reconstitute my database.
Is this how the Berkeley db is operating? Is there a checkpoint like
process I should be doing to my db on a regular basis? I believe I may not
be, because a wedged db shouldn't need to go back to ground zero should it?
Thanks for the help....
--
Peter Kahn
pkahn@connected.com
Received on Fri Sep 3 21:27:56 2004