> Does it concern the Weblocks cl-prevalence datastore backend
> only, or cl-prevalence proper, too?

Both. It's the problem described here:

  http://common-lisp.net/pipermail/cl-prevalence-devel/2008-March/000097.html


> I'm concerned because I have recently deployed a non-Weblocks
> application using cl-prevalence
> and would like to prevent any problems as early as possible (no
> problems so far, though,
> but perhaps that's due to the way I've designed my transactions).

The problem doesn't occur when you always shut down your
application normally. Weblocks will create a snapshot
on closing the store and restore from snapshot works
as it should, preserving object identities and pointers.

It's only when the transaction log is used to restore
the system that the problem occurs and you suddenly
have a lot of duplicate objects.

This means that once your image exits without closing
the store (by you hitting C-d or your server crashing)
you will get an inconsistent state -- or the consistent
state of the last snapshot should you decide to delete
the txn log.

It's basically a durability issue that may easily
become a consistency issue if you don't know about it.


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"weblocks" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/weblocks?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to