[Date Prev][Date Next]
Re: BDB memory cache problem
Luca Scamoni wrote:
Szuts Attila wrote:
On Fri, Dec 03, 2004 at 02:32:51PM +0100, Luca Scamoni
<email@example.com> wrote:I'd run it only if last slapd shutdown was not clean (i.e.: slapd.pid
exists while slapd process is dead). Without looking at the thread I
believe the suggestion to be bit "strong".
Do you run db_recover after system crash before starting OL?
Before starting OL?
My OL starts from /etc/init.d/slapd srcipt and never run db_recover.
Is it correct if I make a shell script that always run
db_recover on all BDB databases before OL starts?
There was discussion on devel list about it:
I've read it. It suggest: run db_recover before any slapd instance
Remember that running db_recover destroys the contents of the __db*
files (that contain the cache and other db environment) and so should
be done only if really needed.
Agreed. After the discussion on that email thread died out we (Symas)
went ahead and developed a locking system of our own that detects
unclean shutdowns and handles recovery automatically. It seemed there
was no interest from Sleepycat or anyone else so we have kept this as a
proprietary feature of Connexitor Directory Services.
-- Howard Chu
Chief Architect, Symas Corp. Director, Highland Sun
Symas: Premier OpenSource Development and Support