Skip to main content



Well, we're now running on a CPU with twice the cores! System performance should be better now—I might even be able to do some ~fancy~ hypervisor shit like the Cool Kids

(Also, sorry I didn't do the upgrade two weeks ago like I said I would, I got sick.)


Going to upgrade Rosalina tomorrow, for real this time. ~KNS will be offline for an hour or so, and as usual, I'll make an announcement prior to the shutdown.



Planned maintenance announcement


~KNS Rosalina will be taken offline for a couple hours on Monday, May 22 to install hardware upgrades. A final notification will be made an hour prior to the maintenance.


Upgrades to ~KNS Rosalina


I've finally been able to purchase a CPU upgrade for ~KNS Rosalina, which should arrive in a couple weeks. Hopefully, this will help improve Rosalina's performance and mitigate the added strain on its resources since Luma's failure in January.

~KNS will be temporarily offline during the upgrade and maintenance process, which may last a few hours. I will issue additional notices closer to the time of the upgrade.


okay testing. does anyone who is Not following this account see this post in their home timeline? is this a bug that happens for all accounts with automatic follow approval, or is there something else going on that I completely do not understand


Storage upgrade complete!



Rosalina's filesystem has been successfully transferred to an SSD to increase performance (before, disk I/O on the hard drive was getting up to 100% so frequently that this Friendica server would regularly become unresponsive). I've decided to go with btrfs for its self-healing and snapshotting capabilities, but I might end up having to mount the /var partition with nodatacow—we'll see how the SQL server performs.

I... could probably also have just used a subvolume and chattr +C as needed, but at least this makes things slightly less painful if I end up needing to switch that partition to XFS or something else.


~KNS Outage


On January 28, 2023 at approximately 16:45 UTC, I became aware of a total system failure on ~KNS Luma, the primary server.

I am currently in the process of restoring ~KNS to full operation. Current recovery status is listed here, and a full post-mortem will be published later.


~KNS will never run on Cloudflare.


No VPSes, no CDNs, no MITMs. Just a physical server box connected to the Internet. That's the promise I made when I started self-hosting, and it's a promise I intend to keep.

To date, the only "outsourced" part of the ~KNS infrastructure is an e-mail relay provided by a trusted friend of mine. Every other service is fully home-grown, running entirely on real hardware that I physically own. Because privacy and digital autonomy is important to me, and I refuse to compromise on my values for the companies attempting a hostile takeover of the Internet.

Stop using Cloudflare. Start self-hosting. Let's fight for a free Internet run by people, not corporations.

reshared this



Maintenance notice for ~keith/cloud


~keith/cloud is now being updated to the latest Nextcloud version. It will be unavailable during this process, which may take a few hours. I apologize for any inconvenience.


Updates to ~keith/bytes


The ~keith/bytes code forge is now being updated from Gitea 1.17 to Forgejo's federated branch. It will be unavailable during this process, which may take a few hours. I apologize for any inconvenience.


Well, I finally got around to creating a little "old Web"-style icon for ~Keith Network Services (the collection of infrastructure and services I operate):
A low-resolution icon of a wolf paw, with the middle replaced by a spinning planet Earth.

So this account exists now! I'll probably use this to share updates about infrastructure status and ~KNS-related projects. This isn't a "professional" profile, though, just a more focused one. (This is the Friendica instance where I yearn about furry bulges, I think the professionalism ship has sailed out of the galaxy by now.)

RoseBeanNikko reshared this.