For flexibility and size I like external m.2 enclosures. I have some from Sabrent, Orico and Rosewill. Of them all the Rosewill is the smallest, has the nicest build quality, and seems to dissipate heat the best.

So I would recommend a Rosewill 9SIA072GJ92919, and add an NVMe SSD of your choice.

I think your MacBook is Thunderbolt 2, so you won’t get full speed but it should still be plenty fast. And this enclosure will give you TB3 speeds if you upgrade your PC later.

No. 1970 is 0 in Unix time. The NTP RFC specifies 1900. I had to look it up!

Better to represent it as a 64-bit unsigned fixed-point number, in seconds relative to 0000 UT on 1 January 1900. It’s how he would have wanted it.

In terms of language you are correct. But in terms of SI usage it seems to me OP is expressing it correctly. The SI unit prefixes have a name, a symbol and a multiplier. The prefix is a concept that encompasses all three of those attributes. So “kilo” is one way of identifying the 10^3 unit prefix, but the name kilo is not the prefix itself. It’s just the name we use to refer to it. And the symbol k in km is certainly the unit prefix portion of that unit of measure.

I’m late to the party on this, but I would agree with the others that Raspberry Pi is not only overkill, but will make this more difficult than it needs to be.

It’s a great job for a basic microcontroller, and the code needed for that will be simpler. You just need something like an Arduino, some wire, a few resistors and 4 buttons. Look at any intro to Arduino introduction that gets to button presses (and debouncing). Here’s a good guide from the start: https://learn.adafruit.com/ladyadas-learn-arduino-lesson-number-1/introduction

If you want to follow a guide end to end and can customize the code, this might be a helpful starting point with a very compact board: https://learn.adafruit.com/arcade-button-control-box/overview

It can help to download your local map for offline use. The default basemap doesn’t have details like house numbers, but the downloaded maps should.

OsmAnd will do that. If you edit the destinations you can manually specify their order. Click sort there and choose door-to-door to get the most efficient routing.

The app takes some getting used to, but it works very well, and can act as a front-end for contributing to OpenStreetsMap.

See what’s using the space. This will list any dirs using >100MiB:

sudo du -h -d 5 -t 100M /var

I use LibreNMS, which is a fork of Observium. It is primarily SNMP polling, so if you haven’t worked with SNMP before there can be a bit of a learning curve to get it set up. Once you get the basics working it’s pretty easy to add service monitoring, syslog collection, alerting and more. And since it’s SNMP you can monitor network hardware pretty easily as well as servers.

The dashboards aren’t as beautiful as some other options but there is lot to work with.

Interesting. In NC here. Not sure if there’s a difference regionally. I was seeing that kind of RTT on ipv4, but ipv6 was slower. I’ll need to give it another try. The last time I did was at my last place where I had the BGW210. I have the BGW320 now and haven’t tried on that. Maybe that, or changes in their routing since then will make a difference.

Did I read right that it doesn’t use systemd?

AT&T is the same. And the last time I looked they don’t give you enough address space to host your own subnet. You get a /64 instead of a /56. And it’s slower than ipv4.

Every few months I try it out, complain and then switch it off.

Monit works for me. Good basic monitoring solution that can also restart a service/interface.

I also use LibreNMS to do alerting for a variety of conditions (syslog events, sensor conditions, outages and services via nagios). But this is more work to get set up.

I tried draw.io, but ended up liking LibreOffice Draw better for hand-drawing.

If you want to get a live map of the connections on your network you may want to check out netdisco.org or librenms.org. Both are open source network management tools that have mapping.

Obligatory: Debian.

But I’d be tempted to put Proxmox on it and then run containers for each function. Then you get purpose-crafted solutions for each use case, but can easily plug new functions in or shut them down based on what you decide later.

So. Much.

Wasted

Space

If the playback device (tv in this case) doesn’t support the codec used for audio or video it must be transcoded for playback to work at all. If this is the cause it could be the tv doesn’t have the right codecs and a standalone device may be better. It depends on the codecs you are using.

Also, if the network connection is slow, it will transcode so that that playback is smooth but lower resolution or quality.

In either case, if the Plex server isn’t strong enough it will struggle. You can investigate this on the dashboard as well, it shows a live cpu usage graph.

Exactly. I would expect a pi to struggle with transcoding. The quality of apps on a lot of smart tvs is poor (particularly older ones that aren’t roku or android, and that Samsung is neither).

I bet it would work a lot better on the Samsung if you get an Android TV device (Onn 4k is a good inexpensive one) or roku, if you prefer that platform.

But first it’s best to know if transcoding is the problem. It might be direct stream but having network problems.

While something I playing, go to the Plex dashboard for the server. It will show the video and audio stream format/codec. Under each it should say Direct Play. That means it’s not transcoding.

Keep the tv dumb. Don’t connect it to the internet.

I like to check rtings.com for model specs and comparisons. Like, some panel types work well in a bright room, some work better than others when you are watching with a bright light source behind you. The warehouse clubs (Costco, BJ’s, Sam’s) tend to have good deals on midrange tvs.

Then pair it with a streaming stick of your choice. A generic Android TV stick/box would work.