Cabling Nightmare

Over the weekend I performed a memory upgrade on one of our Poweredge servers. It needed a bit more in order to run a virtual machine in addition to it's designated role as an application server. A simple task, right?  Open the case, smack two 16GB memory cards into the free spaces, update the CMDB and the change record and Robert's your mother's brother. Yeah.  Not so much. I unplugged the server and popped the locks so I could slide the server out on the rails and was horrified to find that it wouldn't move an inch.

I haven't had to do any hardware maintenance on the servers yet so I hadn't really spent that much time looking at the cabling that had been done in the rack.  Sure, I knew that the network cabling looked okay but since nothing was labelled it's going to take some serious effort to really understand where all the network cables go.  Certainly a task for later.  What I didn't know is that the cabling in the server rack needed so much work.  What one of my predecessors did was wrap the cables around the armature that's supposed to extend as the server is pulled out and then velcroed the cables down at the length they sit when the server is flush.  The result?  The armature can't move because the cables wrapped around it catch on the devices above and below the server.  *sigh.

I know it's super easy to blame someone who's no longer on the job - I'm sure when I've transitioned to new roles my name was used in vain to explain why something isn't done.  I also don't want this blog to become a place where I rant and whine about the challenges of being a mighty IT team of one.  So instead I'll say this:

I know it's cliche, but if you're going to do something then you may as well do it right. I get that a short run time on when something needs to be deployed can reward fast and furious work but supportability and future proofing a solution - even if it is something as boring as cable management shouldn't be a luxury.  Make it right and the next person won't curse you and your ancestors.

As long as we're at it, another cliche:  If it's not written down, it doesn't exist.  There's no justifiable reason to implement something without documentation or at least labels.  Sure, the network rack doesn't look horrible as it could but it may as well be since there's no way to identify a wire coming out of the switches. I'm in the process of getting them all identified (and documented in the CMDB) but I shouldn't have had to.  Grrr..