Beware The Bad Crimping Job

After reorganizing my server room, I was getting really poor connectivity. Connections were dropping, dealing with 500+ms pings, etc. ¬†Thought it might’ve been too many data cables running together (HDMI and data were running together, maybe interference between the two?). That being said, they’re all twisted-pair, which is supposed to deal with that. Determined very quickly that the cable density wasn’t the issue.

Turns out the culprit was a bad crimping job on the cable line.
The cable in question:
2013-09-29 10.23.44
Pulled the end off with two fingers:

2013-09-29 10.24.15

You can’t see it in there, but the connection between the shield and the connector was only about 1/5 connected. Replaced the bad section, and problem solved!

The Investment

Due to the recent complete failure of my server, I’ve decided to plot of some of the purchases I need to make in order to secure my storage space a little more (and to make my server able to take over the world a little bit more, thereby helping along my plans for world domination). Amaretto, the name of my server, hasn’t had a proper upgrade in quite a while, and is in need of some new parts in order to ensure its well-being.

Due to the massive costs associated with building a server, I’ve built the server in small pieces rather than just dropping all the money at once. The system started off as a 500GB hard drive and a $15 case. I slowly added more hard drives (3 more), then upgraded the case twice (first to a Cosmos 1000, and then to a Norco RPC-4020) to deal with my expanding storage needs. Now, my motherboard has hit its limit in terms of SATA drive slots, so it’s time to start looking at larger hard drives and a RAID card.

On the recommendation of a number of trustworthy sources, I decided to go for a 3Ware RAID card which will handle an additional 4 drives and will get me properly started on my new RAID. The choice if hard drives is still up for grabs, though.

Because I will be using the drives in a hardware RAID, I will be buying RAID edition drives, which cost significantly more. Although I would normally go for a Seagate drive, they currently offer the worst warranty on all drives: a measley 3 years. On top of that, they are also the second-most expensive. Instead, I’m leaning towards the Samsung F1 RAID drive. Not only is it the cheapest, but it also has a whopping 7 year warranty. The user reviews were also promising and nobody had any serious problems with them.

All of that will set me back about a grand, so it will have to wait until I have paychecks coming in. Next purchase is a new motherboard and CPU, but I’ll talk about that another time.


For the past two years, I have been running a small personal server out of my house. Well, I guess “small” is a relative term. To be more specific, it was a Fedora 8 server with 2 terabytes of storage for my backups, music, and Subversion repositories. Along with my desktop, it was my pride and joy; everything was custom-built, the distro was actively maintained, and all of the configuration was done by a two-factor encrypted console.

Notice the use of the past tense. Was. On Monday I noticed that I was getting read errors on my drives, although I wasn’t sure which one (I had 4 drives). Although I went to bed with a (mostly) working server, I woke up to a system that wouldn’t boot up because the partition information. After fumbling with different recovery methods to try and get my data off, I came to the conclusion that I wasn’t going to get the data off; it was gone, possibly because of my recovery actions.

Needless to say, I was more than a little devastated. Although I can rerip my music, I can’t get back the Subversion repositories; those are all gone, along with all of the revisions I’ve made on projects.

Despite my demotivation, I took this as one huge learning experience. Although I had made backups of some of my work, it was still on the same drive and partition, and I should have considered them just as vulnerable as the originals. When you can’t make consistent backups on a different system, RAID your drives.

I’ve since reinstalled my server as a 1TB RAID10. Alrhough this means I have half the space to work with, it also means that I have 100% redundancy. So, if one of my drives fails like what happened to me, all I have to do is pull out the drive, put a replacement in, and the system will copy all of the data over to the new drive. The best part about all of this is that this all happens on-the-fly; no downtime, no rebooting, and no manual work except for adding the drive to the RAID configuration.

In short, if you take one thing away from this, remember to back up your stuff frequently, even if it’s just on a flash drive or external hard drive. When your original fails, you’ll thank yourself for that backup.