The only saving grace was the cluster was able to tolerate multiple block device failures, so I was able to reformat those disks using ZFS and resync the data.
To sum up, I experimented on an experimental file system, got burnt, and then switched back to ZFS. I didn't lose data, just time and frustrations.
It sounds like a joke, but it's not. Happens all the time. These office systems are so obscure that the new vendors often time don't even bother to migrate the data, forcing manual data entry for all the relevant records. It's very sad.
It's the same when these "office" systems have thousands of connected offices. Each would still have to re-enter its records.
From the article, "The device offers the possibility of a cheap and flexible design suitable for harvesting waste heat in the 100- to 200-degrees Celsius range.". Assume it really work all the way to 10C, that could theoretically be 190/283 ~ 67%. If their efficiency is higher than a turbine, then they can use that to replace the last turbine stage. If not, then there is nothing to see.
I think the news here is that they *might* have come up with a better device than a low temperature turbine.
Cumulus' product appears to be a full blown debian port that runs directly on expensive 48 port switches. Seems pretty useful to me. Feel free to point out alternatives.
Last I look, a Quanta 1GB switch is in the 3K range, and the 10GB switch is in the 11K range.
Here is the list of supported switches: http://cumulusnetworks.com/support/hcl/
Also, for flexibility, you want SFP+ ports and adapters for each port. None of those are cheap.