I had a bit of time to kill last night so I thought I would use the time productively. I downloaded the latest distribution build of Nuxeo 5.8. Given that the release is at end of month I reasoned that the release must be relatively stable at this point.
Here's what I did:
- Downloaded the distribution and installed on CentOS 6.4 and Windows 7
- Used the out-of-the-box database and configuration (with a few exceptions - noted below)
- Took a quick spin through most of the screens (but this was not meant as an in-depth functional test)
- I made no attempt to apply any custom packages
- I did not add any additional add-ons (the focus was only on DM)
This was intended as a quick spin through 5.8 to note anything of interest (good, bad, or ugly :-). And really this is contrasted with Nuxeo 5.6 the previous long term release. I'm very happy to report that most of what I found was very good!
Today's post deals only with the set-up and performance. I will delve into UI and functional observations in my next post.
I downloaded the 5.8 distribution from Nuxeo's maven repository. I ended up downloading the file "nuxeo-distribution-tomcat-5.8-20131016.015541-19-nuxeo.cap.zip. Finding the distribution I wanted was the most time consuming part of this process - as there are many options and builds when you traverse the distribution sub-tree. Once downloaded, I extracted Nuxeo 5.8 on CentOS and Windows.
I tried to run Nuxeo right out of the box, on Windows and CentOS, but both failed but for different reasons. I know I should read the documentation but I am always curious as to whether I can avoid it.
On Windows, I got a Java version error which isn't surprising since Nuxeo 5.8 requires Java 7 to run. I had Java 7 loaded on my Windows workstation (but not as the default), so I edited nuxeo.conf and set JAVA_HOME to point to my Java 7 install. Thereafter Nuxeo loaded without complaint.
On CentOS Java 1.7 is my current default so I knew that wouldn't be an issue. However, when I went to issue the ./nuxeoctl start command I got an error that this command could not be run. Simple enough change chmod +x on the startup script and then Nuxeo was happy to start.
So far so good.
It should be noted that with this build I was not taken through the start-up wizard on first load. Rather, the default database and tables were created on first start-up using H2/Derby embedded databases. So, I decided to log in on Windows and CentOS, log out, and then restart the server. The reason for this was simple - I wanted to compare Nuxeo 5.6 start-up times to those in 5.8 (after all tables are created).
From a startup perspective, I was quite happy to see that (once again) Nuxeo has done a great job in reducing start-up time. Using the same systems and resources I saw start-up times roughly 20% faster in Nuxeo 5.8 than for a comparative configuration in Nuxeo 5.6. This is not in any way scientific, and I don't yet know how it's been achieved (lazy loading?) but I do know that the difference is noticeable. My Windows machine is far slower than my CentOS development box, so while the absolute load time was longer the effect of a faster load was more significant.
Other startup observations
- I noticed that there are far fewer warnings in the console/log on startup (don't know if logging levels have changed or whether Nuxeo has addressed the root issues generating those warnings)
- There are some new informational messages on startup related to what looks like CMIS service endpoints
- On CentOS, clicking the Open in Browser button on the Control Panel generated an error (most likely an issue with my install but I need to track this down since it used to work in Nuxeo 5.6), however, I could still launch Nuxeo by typing the URL into the browser. There was no issue launching Nuxeo on Windows.
- I used the Administrator/Administrator account to login and here's where I had a real surprise - the time to login is dramatically faster than in Nuxeo 5.6. I didn't delve into the mechanics of how this is achieved but it's a very big improvement.
That's it for today...