The brain transplant was delayed until tonight. But it should also only mean downtime of a few minutes.
But it's good downtime. No, really - still, my apologies if you can't get to sleep and try to tune into UHub for some of that smooth overnight jazz (wait, do I need some sleep myself?).
With any luck, the company that hosts Universal Hub will be taking the server down to replace the current antiquated spinning-disk hard drives that the site now runs on (yes, platters that matter) with some of them newfangled SSD things the kids are always on about.
The new SSDs will have better I/O (which means better performance) and less chance of catastrophic needle-crashing failure (also roughly four times the storage, yeeha).
Obviously, this computer brain surgery will take a bit of time to do, hence the downtime.
Equally important, a couple of very kind folks who actually know about Linux servers and MySQL databases have been pouring over the site over the past couple of days and coming up with ways to fix the site's poor performance of late, especially for visitors with accounts. The site seems to be doing better today than it has in a long time, even before the SSD transplant.
More on the changes, and these two good people on the flip side (I want to get their OK to name them first).
Like the job UHub is doing? Consider a contribution. Thanks!
Ad:
Comments
Fingers crossed!
By mg
Tue, 03/11/2025 - 8:45pm
SSD should be a big improvement.
I noticed the improved performance today. Thank you to the kind people helping you with the IT end of things!
I/O, I/O, it's off to work we go ...
By adamg
Wed, 03/12/2025 - 9:25am
I'm learning way more about server maintenance than I ever thought I would:
The way they swap out drives is first they sync the drives on the live server with the impending new drive to get all the data and then, after making sure the transfer worked and the data is all there (exclusive video showing the process), they turn off the server, disconnect the old drive and connect the new one.
They started the process last night, but it turns out there are I/O issues on the current server (don't know if its a physical disk issue or a software issue related to how the database is still reading/writing massive amounts of stuff to/from the disk), so the backup was taking way longer than expected and was still running when the tech on the project's shift ran out. So they'll do it tonight. The good news, as now mentioned in the original post: The actual server downtime should only be a few minutes.
If you ever wanna jump
By Kaz
Wed, 03/12/2025 - 12:54pm
I'm happy to help consult on using AWS if you wanna join one of the big boy providers (disk swaps there are basically instantaneous if you do it right). A little more self-determinism and a little less waiting on techs...but may not compete on the cost (not knowing how much your current provider charges).
🎶 F-disc Format Reinstall, Do
By MassMouse
Wed, 03/12/2025 - 7:52pm
🎶 F-disc Format Reinstall, Do Dah, Do Dah 🎶
Comment counts
By crispino
Thu, 03/13/2025 - 8:25am
Will this work fix the comment counts? They were broken for a while after the switch to the new software, then there was a brief period where they seemed to be working again, but now they're back to not working (or working for some articles and not for others, which seems even weirder).
Add comment