[infinispan-dev] ISPN-501

[Manik Surtani]

Sanne,

The AtomicMap approach will work but as you mentioned on the JIRA, you won't get all of the benefits of distributing chunks.

I think order of operations is a better approach.  If you are using sync communications then the order os guaranteed across the cluster.

Just to confirm, the problem has to do with:

T1:
Read metadata
Read chunks

T2:
delete chunks
delete metadata?



On 21 Jun 2010, at 21:05, Sanne Grinovero wrote:

> Hello all,
> I'd appreciate some thoughts or suggestions about how to solve ISPN-501,
> I was thinking to try an AtomicMap to contain all chunks from a File,
> including the file metadata, or should I use transactions in read mode
> too?
> 
> Or should I just make sure to send the operations in a safe order?
> (not sure if that's possible but I could look into it - also I'm not
> sure if it's safe to assume the order will be the same as received by
> other nodes)
> 
> I have been looking into GridFS, which BTW seems to have a similar
> design than Lucene's Directory, but didn't notice any special trick to
> address this kind of issues.
> 
> Regards,
> Sanne
> _______________________________________________
> infinispan-dev mailing list
> infinispan-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/infinispan-dev

--
Manik Surtani
manik at jboss.org
Lead, Infinispan
Lead, JBoss Cache
http://www.infinispan.org
http://www.jbosscache.org