Sysdom Blog

I dislike suprises. Not the nice type of suprise like a suprise birthday party or unexpected windfall; I mean the type of suprise where a contractor insists that the work is ‘easy’ and ‘wont take long’, but then a week later you have still to see any output from him.

If I had the time (and the skills) to do it all myself, I perhaps might.  However, do I really want to have to learn all new tools just to do something that would take someone else perhaps a few hours… and even if I ignore the time factor, often the tools will cost more than the cost of paying someone to do it for me and worse still – my design aptitude isn’t the best. (Lime green and neon pink for the new colour-way? sure, sounds good <grin> )

So where does that leave me? I could try to find better freelancers. I could try to manage them better (or be an unreasonable hard-ass client) or I could try something else… the question is what?

I suppose it’s just something I will have to learn to cope with and build it into my project timelines more.

With still being prelaunch, I still have most of my data at home on the server. I do encrypted online backups of my critical data to my ‘free’ backup space at Dreamhost, but most of my data is stored on 6 drives on my home server and a couple of external drives for backups…

I had a few RAID issues this past week. One of my RAID arrays (1TB of RAID 6 in case you were wondering) had already lost a drive and was running degraded. Ordinarily this would have been a problem; I run RAID 6 to allow me to keep running in the event of a single drive loss. But as most geeks know, problems seem to crop up together…

One of the other drives in the array started failing its daily SMART health checks and eventually dropped out of the array in such a way that it crashed it. No data was lost at that point, merely just irritating. So I ordered two new 1.5Tb drives and got back to work.

When they arrived, the rebuild went well, I shrunk my RAID 6 down to a 2 drive RAID 5 after moving most of the data off and swapped it over to RAID 1 using mdadm. Kudos to Neil Brown, it worked really well but as with most tasks involving large amounts of data, was just a little slow.

At this point I was struck with a choice… do I choose Linux software RAID again or move to something a little newer and more fully featured?

ZFS is nice, pretty stable, fast and when it works, it works really well… The only two problems I have with it on Linux – no native support (you can use it under FUSE if you wish, but its CDDL license just isn’t compatible to bundling in the kernel) and no real fsck and recovery tools.

Btrfs while not as far along developmentally, does look promising, but as with ZFS has the same issues with no real fsck or recovery tools – and the website insists that it is still only experimental even though the next Fedora release will be using it as default.

So where does that leave me?

With Linux Software RAID I suppose…

The new(ish) hybrid RAID 10 level is rather tempting, but mdadm only has basic support for resizing arrays based on this level, but it was tempting… In the end I just went with bog standard RAID 1. I know it well and have no problems with repairing it if it goes wrong…

There is a planning method called a critical path analysis which shows you exactly how quickly you can get from start to finish by understanding which tasks need to come before each other and which can be done in parallel.

When you are the main person doing the work it is hard to do many tasks in parallel, but you can decide which tasks don’t have to be done right away by prioritising certain required end results of the project and move them into a second phase.

Not all products will be live straight away on the site, but it is hoped that we should be able to roll out around a new product every month after we have launched.

Learning something new isn’t easy. You have many preconceived ideas you need to rid yourself of before you can start to actually learn… and so it has been with dealing with most of the details of setting up Sysdom. Continue reading →

I know a lot of people put a lot of thought into their mission statements but so far there are only a few that I have actually seen that bear more than a passing resemblance to the business they are supposed to represent. Continue reading →