On May 15, 2009, at 6:22 AM, Hanno Schlichting wrote:
> I have a configuration with one ZEO database server (ZODB 3.8.1) and
> physical servers housing four single-threaded ZEO clients each.
> some 15gb of blob data in the (Plone) site.
> Setting up and maintaining NFS to share direct access to the blob data
> is beyond what I can support in production right now. So this leaves
> with a blob-cache per ZEO client.
> Am I correct to assume that sharing the blob-cache between the four
> processes on the same physical machine is not supported, as these
> have any mechanism to synchronize read/write activity to the blob-
> Or is there something in the blob design that ensures a no-conflict
It probably works in ZODB 3.8, although there are no tests for it.
I made sure it works in 3.9 and there are tests for it. This is our
(ZC's) production configuration.
> Does the blob-cache grow indefinitely or is there some mechanism to
> purge / pack it built-in somewhere?
ZODB 3.9 has an option to limit the blob cache size. We're using that
Note that the ZODB 3.9 blob cache uses it's own layout to facilitate
the management of the cache. You can't use a 3.8 blob cache with 3.9.
For more information about ZODB, see the ZODB Wiki:
ZODB-Dev mailing list - ZODB-Dev@zope.org