Upgrade of 3.4.d to 4.0.c Risks and Recommendations

[jollymon]

I have been toying with the idea of doing the upgrade from 3.4.d to the latest community stable 4.0.c I have read that some people have has some issues with this upgrade and that the SOLR vs Lucene issue is often at the heart of it.

It looks like the 4.0 branch will be the major transition point for all future upgrade, so should I just bite the bullet and get it done now or wait for a further future upgrade?

Feature wise its not revolutionary specific to our application, except the .psd and (limited) .ai file support, which doesnt seem worth it in and of itself.

Thoughts and Opinions?

Thanks,

J

[jpotts]

Some thoughts:

• You don't have to use Solr if you don't want to.


• 4.0 adds drag-and-drop and many other usability features that people really like.


• 4.0 should perform better. Benchmarking is underway, but indexing is definitely improved if you do choose to use Solr.


• If you need to do any Share customization, some of the new extension points are real time/code savers.


• We'll probably do another 4.0 drop soon so you might want to wait for 4.0.d.

With that said, my 4.0.c work has always been on a fresh repo, so while I can list some benefits of the release, I can't weigh them against the cost of a real-world upgrade.

Jeff

[jollymon]

Thanks Jeff I will wait for the D drop and toss it in at that point on some newer in-house hardware we are looking at. Our current specs are a bit weak for the newer versions

Old hardware—-
2.2 dual core
2gig
100mbit - not local/fast at all but reliable dedicated server
xp 32

But its stable as pie, and cheap as dirt!

Any chance of getting one of the newer versions to run off a platform with this basis? We use it mainly for workflow and graphics management for ~8 people a day.

[mrogers]

2 Gig is pushing it.   However you don't have many users, so you should be able to reduce the number of threads.

[jollymon]

With the D drop out I am going to attempt a migration to the 4.0.d win-32 platform and keep the luce index system. I can see no real reason for us to worry about SOLR. I will also try to update all the code patches for workflow and datalist mods on the forums.

WISH ME LUCK!