As you might have noticed, SemiAccurate has been down for a few days now, and although it is back up, it is an older copy of the database. Here is what we know so far.
Some time Thursday morning we had a catastrophic SSD failure on our server that seems to have been building for quite a while now. Unfortunately the cause of it had been going on for some time unnoticed in the background. This of course corrupted a lot with a lot being defined by databases, backups, and more. This is why you see the old copy of the site, it is the last clean copy we have,
Where do we go from here? We are pulling the old data off the dying server as fast as possible and are fairly confident we can get most if not all of it up eventually. With Hot Chips starting Sunday, we have limited time to pull backup, right now it is a slow file by file slog with slow reboots when you hit a bad sector, what we in IT define as ‘not fun’. Hopefully in about a week, all will be restored with no loss of data.
In the mean time if you are a subscriber and your account is not recognized by this new version or there is a story that you want to read which is not up, please write charlie at semiaccurate dot com and I’ll try to send you a copy or fix the account. Hopefully all will be restored soon and it won’t be a problem.
We will update this story with any new events. Sorry for the downtime and problems.
Update 1: Aug 29, 2016 @ 5am: Sorry for the long delay but things have been happening in the background. We have decided not to change anything on the current version because if we do, when we restore the backups, the changes will be wiped out. Because of that it looks like nothing is happening but it is.
Pulling files off the old dying server is happening at a snail’s pace because when you hit a bad sector, the whole machine goes down. Reboots are slow and frequent but we have the overwhelming majority of the files back and most are 1-2 days from the crash. Hopefully these will be put back on the new server later today.
Once again sorry for the delay but things will be going forward really soon.
Latest posts by Charlie Demerjian (see all)
- HyperX ships it’s 60 millionth enthusiast memory module - Oct 15, 2018
- Bittware/Nallatech water cools 300W of Xilinx FPGA - Oct 12, 2018
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018