I'm still really interested in understanding what should be the standard
approach to effectively managing UGC in a clustered Magnolia
environment. We’ve used Magnolia for a smaller project and the
experience was excellent. Our developers and users both think it’s great.
I think you are mixing two different problems.
Public clustering works, and works pretty fine. I used it in a few
project. UGC handling is a completely different problem and is not
related to clustering. If there is a standard handler for UGC, that
should work both in clustered and non clustered enviromnents. I suggest
you to first fix the clustering problem and then see the best UGC
handling solution.
But now I'm in a position where in the next couple of weeks I need to
make a decision as to whether or not Magnolia is a suitable product for
largish (clustered) Web 2.0 websites that have user generated content
(UGC). Think comments, rating, recommendations, etc. I know the latter
aren't yet supported but we can extend. I really want to be convinced it
is suitable because the general CMS features are great but I’m still
searching for the evidence.
Unfortunately, I agree that a complete UGC solution is lacking in
Magnolia. Just my two cents.
I’ve spent the last few hours trying to do exactly as gjoseph has
explained by reading http://wiki.apache.org/jackrabbit/Clustering
combined with
http://wiki.magnolia-cms.com/display/WIKI/MySQL+Persistence. I got it
working... Kind of. But it seems to be unreliable. Sometimes content in
one cluster will show up instantly in the other. Other times, not at
all. When I restart clusters sometimes data entered via one cluster is
completely lost even though it was there the entire time the server was
up. I get no exceptions in the logs and the Jackrabbit repositories
appear to be all setup as proper clusters with journals and all... I
even tried clustering just the DMS workspace to simplify things and only
activating to one of the public instance... I got the exact same
unreliable behaviour. I’ve also found partially clustering a repository
(selected workspaces only) causes lots of complications with module
installations as pointed out by rah003 in the forum comments above.
Start with one step a time. First, use db persistence (two separate db,
one for author and the other for public), and see that a single instance
is working well with mySQL o whatever you are using. Check drivers and
mysql configuration. Some times ago the default installation was in
swedish language and not UTF-8, that lead to a few problems for us...
Then kill the public db and start again with clustering, configuring two
magnolia public instances. The most common problem is usually that both
instances have the same node id configured, so they'll behave VERY
stangely. You can also try completely disabling the cache on public
instances.
But it's very hard to give a solution without any info about your test
configuration...
Regards, Danilo.
----------------------------------------------------------------
For list details see
http://www.magnolia-cms.com/home/community/mailing-lists.html
To unsubscribe, E-mail to: <[email protected]>
----------------------------------------------------------------