Software, technology, sysadmin war stories, and more. Feed
Friday, June 17, 2011

Test your equipment before recycling it

Let's say your business model involves using hardware until it won't work any more. This means parts will be recycled across customers and through the years until they finally give up. If done properly, this might work out fine. On the other hand, if not, it can wind up really badly.

Imagine the situation: every time a machine is built, a tech reaches for a drive in the recycle pile. If it takes the kickstart, it is put into production. Let's say something bad happens to it later. It'll come back, be wiped, and assuming it's not completely dead, will be put back in the pile. Also assume that the stresses of actual working life are nothing like whatever quick tests might happen while it's being built.

If you follow this to its logical conclusion, you should get a small army of flaky drives slowly rotating through the pile, going into customer machines, going nuts, coming back into the pile, and so on. It'll be a grand rotation of refuse.

Eventually, all of the good disks should stay put in production, and all of the bad disks will just keep rotating. The only time you'll get something fresh is when all of the bad disks are currently installed, at which point new stock will be introduced. If you never drain it to that point, it may never happen.

Given all of this, thorough testing seems mandatory every time a disk comes back from anywhere. Why leave it to chance?