New job and IBM SPSS Modeler/CADS

I started a new job in April 2012. My old one was relatively geeky but working in a very highly specialised field. "Group Technical Support" was the job title but in reality I was doing all-sorts, from development support, software & hardware validation, and a fair whack of PHP/MySQL/Bash/Python troubleshooting on a distributed system getting telemetry data from buses at depots through a wireless network.

My new job is for a company that does what essentially comes down to data processing. Lots of stuff like address cleansing, identifying duplicate users with different accounts, fraud prevention, data pooling. My job title is "Application Support Engineer" and formally I sit between the service desk and infrastructure, mainly tasked with monitoring the data flow in, through and out of the business. In reality there's a lot more of the tedious stuff than I really want (service-desk type stuff) and far too much being on-call for my liking, but overall I really am enjoying it.

My skill set is definitely much improved after 10 months in this role. Now I get to play with enterprise databases (e.g. Oracle, massive MySQL datasets), a vast number of cool data manipulation tools and I am getting both MySQL administrator and ITIL v3 Foundation training in the next few months.

One of the new things that we've got our hands on is IBM's SPSS Modeler & CADS combination. Modeler is essentially a GUI-based tool that allows users to manipulate data. It's hefty and a really nice piece of software. CADS is "Collaboration and Deployment Server", which is a repository for data streams (programs) and results and allows users to schedule work or run jobs in a collaborative fashion. It's Java-based and runs within the JBoss Application Server. My experience of both JBoss apps and Java is limited, to say the least.

We got an IBM consultant in for a week to help with the install, and overall they were good, they obviously knew the product well from a user's point of view, but they were NOT technical experts on the product. They didn't know what was going on behind the scenes or why X failed or what we had to do Z before Y or it all fell over. I got quite frustrated because I like to know why things happen, not just that they do happen, and ended up doing lots of the install on my own intuition. It worked, mostly... then when we found the bits that didn't, IBM's support were really patient and helpful, and got it all working as it should.

Fast-forward to today. Today I was required to update the version of Java used on the CADS host. So I stopped the service, updated the Java install, updated a few environment variables and started the service again. Then I couldn't connect using the client. Bugger. Double-checked all my work. Nothing wrong, all the right variables changed to the right values. Restarted the server. Still no connection. The error message was cryptic: "Authentication information not specified". really helpful.

I removed the new version of Java. Changed the variables back, and still had no success. WTF? I've put it all back to exactly how it was and still the service doesn't work.

At this point I bit the bullet and contacted IBM support, who appeared as baffled as I did. After a couple of emails back and forth, and at the end of the day (1730) I think they gave up. I ended up restarting my PC for an entirely unrelated reason, and went to give the users an update on the problem. Just to double-check the error message I tried to connect again.

BAM! It worked.

It connected, instantly. No fuss, no errors.

What. The. Actual?

So. The moral of this story is, if your connection to CADS through the Deployment Manager fails with the error "Authentication information not specified", don't blame the server. It could be your PC.

Comments

Popular posts from this blog

Data Guard with Transparent Application Failover (TAF)

RMAN-05531 During RMAN Duplicate from Active Data Guard Standby

Data pump - "ORA-39786: Number of columns does not match between export and import databases"