I also just received my Turing Pi RK1 32GB compute modules, which I’m hoping to eventually have run all of the services that don’t rely on GPU transcoding directly. There’s also rumors of Plex supporting RockChip GPUs (these RK1s use the Mali G610 GPU) at some point in the future, and JellyFin already does, so I may take a more serious look at JellyFin as my video streaming solution - so watch out for invites to the service as I make it available. One downside is that I’ll have to manage user accounts, but since I make regular offsite backups now, restoring those accounts (especially via containers) shouldn’t be an issue. The eventual goal would be to eliminate (almost) all services being run on the storage host (Hemlock) and move those to the Turing Pi and NUCs where I need baremetal for database and other purposes.
If you’re wondering where the name Hemlock came from, it was my Dad’s handle back in the BBS times all the way up to the internet. We had started building the storage host out when he passed away, so I felt it was appropriate to name the host after his persona. It’s been a rock solid machine and I forsee it keeping the Hemlock monicker even when I do some physical hardware upgrades on it, which will be happening eventually.
To give you an idea of what Hemlock is:
Pixelfed went down, and I couldn’t get it back up without a complete wipe. Due to that, I felt it was not stable enough for production use yet (read: reliable low-impact upgrades) so I’ve abandoned it for the time being. Second, I’ve abandoned my diaspora instance as it had a very low user count as a platform in general and it was not providing any value.
I have maintained my Mastodon instance - there are many millions of users and I follow quite a few users. I am not very active, but there’s some great threads and great people to follow.
Peertube is still running and for the most part, it’s working pretty well. I’ve ran into an issue with large file uploads that I need to sort out, but it’s turned out to be a pretty reliable home for my personal videos to share. I don’t have federation turned on at the moment, but I’ll eventually turn that on once I sort out my storage transition to an internal S3-style model via Minio.
Facebook is certainly hard to quit. The groups function is very good primarily because of the users. Same with marketplace. There needs to be a solid replacement for those. That could include Friendica, but I have not explored that platform on my own.
Lemmy is a mixed bag. Lemmy intentionally does not support privacy options such as GDPR. There’s no respectful “delete” option that spreads across instances, and I think this is where I would diverge when creating my own platform.
More later!
]]>I ended up working in the first run group, resting during the second and running in the third. Rain was light in the morning, mostly non-existent in the second, and started coming down heavier in the third, so my times were much slower than the earlier groups. Out of twenty-five competitors in my run group, I ended up coming in fifth and taking first in NS2.
I spent some time getting SoloStorm to work. I got everything working with the OBDLink MX+ telemetry dongle gathering car metrics, GoPro Hero 4 recordingin 1080p60, and QStarz 818GT GPS. The MX+ is slow - roughly 1hz, so throttle input telemetry is fairly laggy in the overlay video. Also, there’s definitely a process that you need to go through to setup SoloStorm to record your course walk:
Links to video and results are over on the Motorsports page!
]]>I plan on restarting my Matrix instance in the next couple of months on local hardware - hopefully using K3s and the Turing Pi configuration, with dedicated hardware for database hosts. My plan is to leverage a couple of the existing NUCs I have as database hosts for all of the container-based services I run on the domain, and then have the services themselves run in containers orchestrated by K3s. Once that is done, I’ll start exploring running a Lemmy instance as well.
If you don’t know, Lemmy is a federated alternative to Reddit. A while back, Reddit wanted to lock down it’s third-party REST API access with exorbitant fees. This spurred a large migration away from Reddit and to a variety of different Lemmy providers. This stressed the ecosystem pretty significantly and it’s also highlighted some dangers and additional steps SysOps need to take if they plan on running federated services. Similar events occurred on Twitter. Elon Musk (hereto referred to as elmu) purchased Twitter, which he renamed to X, in order to shift it into service right-wing extremist content. This has caused a large migration from Twitter to Mastodon providers, causing similar issues to occur.
It’s 2023 folks! Get away from the corporate social media apparatus! A quick breakdown:
Those are some alternatives. I’ll have a future post on how to choose a proper social media provider.
]]>So this year I’m going all in on a set of #open-source, #decentralized social media platforms. My hope is that I’ll adopt one of them to host varied interests that I have without having to interact with FAANG products in a regular basis.
Right now, I have #diaspora, #mastodon, #pixelfed, #peertube and #matrix (Synapse and Element) running on my domain; Several of them are running either as #Docker containers or on VMs, with Matrix on a VPS that is more than I can supply at home currently. So far, the experience has been fairly smooth. Diaspora, Mastodon and Matrix were the most straight forward to setup with Pixelfed being the most difficult. That could be improved with up to date documentation, though. As far as additional platforms go, I’m interesting in taking a look at #lemmy as a decentalized #reddit replacement. There is a slightly larger project I want to tackle though, and that is named migrating my entire setup into a #k8s cluster so I can more easily manage each application, centralize database resources and reduce service overlap. In other words, improve efficiency.
My plan is to post to my pixelfed instance at least once everyday with photos under the #project365 hashtag. I’d like to document the technical aspects of being a social media platform #sysop and better understand how to scale #synapse once I have my #kubernetes cluster up and running.
For #hardware, I’m running a #xeon E3 1245v5 w/64GB of ram and (3) #mdadm #raid 6 arrays. These arrays consist of (1) (8) 3TB drives; (1) (8) 4TB drives and (1) (8) 512GB SSD drives. This gives me a safe amount of storage to be able to rip my bluray media and stream it via #plex to my TV, as well as host large datastores like Elasticsearch, my #nikon #d610 photo archive and #gopro videos that I take while racing.
I’ve also recreated a GPG for the new year. I lost my private key years ago and bungled everything up, so my picture is on the internet in keyservers forever! But I have a new, stronger key that I’ve saved offline properly, and I’ll be leveraging that moving forward. It is currently on the Ubuntu keyservers and Keybase if you want to download the public key.
Keep an eye out for more posts! I’ll be documenting the #k8s journey as much as I can.
-sfunk1x
]]>You can reach me on varied federated social media:
I have a variety of interests: #cars, #linux, #floss/#oss, #robotics, #rocketry, #film, #music, #diy, #docker, #photography, #geneology and a bunch of others I haven’t mentioned here.
If you are from the Portland, Oregon USA area and are interested in #decentralized social media, drop me a line. I’m happy to welcome more people to the #fediverse, even if you aren’t from #pdx!
]]>