[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: index corruption (1164) still present in 2.0.15 with db 3.1.17 (ITS#1359)

At 09:44 PM 2001-10-10, adamson@andrew.cmu.edu wrote:
>Is there something that makes thread2 wait to
>open the cache while thread1 has it open? Should there be?  Or should
>insert_key be serialized? Or serialized on some per-db-key basis?

I think you've hit on the problem.  This likely effects
DN parent/subtree indices as well.  We can either try to
add some fine grained r/w locks or one giant r/w lock.
I suggest the latter and that it initially be quite
giant.  That is, every "database" have one giant lock
which is acquired for read for LDAP read operations and
acquired for write for LDAP write operations.  This
will serialize all writes within a given "database".