Tuesday, March 20, 2012

Deploy Woes Continued, Bad Data

As I've mentioned earlier, issues happen with deployments.
And also post-deploy, both big and small.

Code assumptions
Assuming the data is in a good state. But there is always bad data.
Especially when using a bad UI framework, an old database model, and legacy code written under old assumptions.
And in trying to slowly cleanup a database for example, scripts can run fine testing in the Dev or QA environments. But if these aren't synced up with Production, it can and will fail there.
Suddenly a value is NULL unexpectedly, error error. Oops. Fail.

Code management
Ambiguity for who owns the code and who is responsible is killer.
And on that same note, who understands deployment steps.
Having a lone wolf coder on fundamental code doesn't help.
It's far to easy to be heads down on your own work and ignore it, until it bites you.
More code reviews, more shared knowledge.
So you don't deploy to production and find it isn't working because it's only half deployed. Opps. Fail again.

No comments:

Post a Comment