I think it’s time to re-wire the house. Running everything through 2 floors up to my office isn’t really practical and would exhausting…so I’m thinking 2x10gb down to a 24x PoE in the basement with a local battery backup for it. The white pipe here goes almost to the attic: ImageImage
There will be a rack up in my office closet later (or maybe sooner than later)...I'm thinking a small 6U conditioned rack down here. All it'll have for now is the 1U switch and a 1-2U battery backup. And maybe a media converter for fiber...all of this is in prep for a fiber swap.
I'm *assuming* AT&T won't let me plug into this switch from the fiber converter on it's own VLAN before going to their gateway and use the 20gb backhaul upstairs where the gateway will live...but if anyone is doing this rather than running a dedicated line I'm super curious.
Alrighty, I've decided to do this in phases, because I don't have a chunk of time off like the original sabbatical plan either.

So, phase 1:
Adding a 9U rack to the basement, which will contain a USW-Pro-24-POE for switching, a 1500VA UPS, and the AT&T gateway to start.
Next will be to connect this upstairs - the USW-Pro-24-POE (store.ui.com/collections/un…) has 2x 10Gb uplinks - I'll drop 1 more cat6a then bring AT&T in on a 1Gbps port and route upstairs to another 48 port PoE switch over the 20gb LACP backbone - still using the USG for phase 1.
I'm still researching the upstairs - probably a full depth 15U rack so I have plenty of space, and a 1500VA + expansion mount UPS solution. I need to add up all the weight though - need something with beefy wall mounts. With a decent UPS, the hanging capacity can vanish fast.
If this sounds strange: remember my entire network is PoE, from Flex switches in remotes to cameras to APs - so a central UPS at the switch is the strategy to keep internet, wireless, and everything up when power is out. Much cleaner than multiple backups spread out, IMO.
For example, instead of wireless for devices like a Chromecast, I run a PoE switch behind the TV, powered over the Cat6a. It powers a AP across it sometimes (via PoE) as well. This means everything at a TV is wired, not wireless...leaving much more wireless capacity open.
"Hey Nick, that sounds ridiculously over-engineered and overkill for a house"

Yeah, probably. But we all have our vices.
Alrighty! Rack, UPS, 24 port PoE switch, and 10Gb SPF+ modules all arrived today. This weekend shall be some basement fun on phase 1 of the network upgrade in prep for moving from Spectrum gig to AT&T fiber. Image
Rack LEGO time! ImageImageImageImage
It’s been a very long time since I assembled a rack. Good thing I cleaned the workbench because geez you need a lot of space: Image
General assembly tip: you can get a stack of magnetic bowls off Amazon or whatever for < $10 usually. They’re fantastic for this sort of thing. If they’re lettered, pin the bag under. They also store easily underneath a shelf with a simple metal plate approach: ImageImageImage
Base assembly together: ImageImageImageImage
4 posts in and all assembled. In retrospect, I guess it’s obvious it’s about as heavy as the original box was. Gonna need to recruit help to mount this thing easily. ImageImageImageImage
Oh, rack was completed last night - roll of painter’s tape for scale…and here’s where it needs to be mounted: ImageImageImage
Unboxed the switch earlier and got it provisioned and 10Gb transceivers in - just need to find some lumber for mounting the rack properly. Might get to that tonight. ImageImageImageImage
Side supports added and UPS test fit. Lumber is on the way so I plan to mount all this up tonight. Not sure if I’ll get to cabling or not, but hey it’s the weekend! ImageImageImageImage
Fun fact: the rack fans eat about 40 watts. UPS smoke tested…time to figure out dinner then start mounting things. ImageImageImageImage
First up: back mounts (for something beefier than stud hanging) and the French clear style hanger: ImageImageImageImage
Rack hung, UPS and fans are going. Tomorrow I’ll get into mounting the 24x PoE 1Gb + 2x 10Gb switch and network re-routing - relaxing for this evening. ImageImageImageImage
Alrighty - picking this back up. Right now internet runs down to a PoE relay switch behind the basement TV and then from there to basement AP, camera, and living room (all PoE-powered). Time to re-route most of that: ImageImageImageImage
Luckily, I planned ahead here and the distances I ran on original cables were to go to this rack location, with slack in the wall. However, with 4 in that conduit the heads aren’t going to come out cleanly. ImageImageImageImage
Eeek yeah, time to re-crimp, this will be one of the 10Gb pair coming from the office rack/stack upstairs: ImageImageImage
Times this door has hit me in the face: 6 Image
Bad door! Timeout. Image
Well, shit. I had no trouble provisioning it but after plugging up downstairs I’ve got a firmware error. Very bad sign with a new device. Will try to troubleshoot by have to RMA to @ubnt :( ImageImage
@ubnt Yeah it's cooked. Dammit.

I ordered patch panel (keystone style) and panel cables (now that I know how I want this to be setup)...but that'll have to wait. I'll work on re-patching through the USW-Flex in the rack for now. ImageImage
For now, getting the basement and living room back online while I deal with RMA mess. ImageImage
As a final insult from the DOA switch, I evidently cross-threaded the lower right screw. It was so jammed I broke the screw in half with an impact driver before it came free. So…time to order more rack nuts and such as well. Image
Unfortunately with shipping delays my switch RMA hasn’t even arrived yet (shipped with 2-day…over a week ago), so who knows when I’ll get it back. I may stop dealing with @ubnt on that front and switch to MikroTik for these switches (e.g. CRS354-48P-4S+2Q+RM).
For now, patch for downstairs assembled - using cat6 keystones (can’t really find 6a) and cat7 for the 2x 10Gb drops to this location. Makes it easy to tell what’s where: ImageImage
Actually dealing with a DOA/RMA device is enough to turn me off from directly buying from @ubnt's store again. From Amazon, I could just return it. I ordered direct for warranty...but dealing with their service just doesn't seem worth it. Oh well, lesson learned the hard way.
RMA replacement has finally arrived - 11 day turnaround time. Let’s try this again… ImageImage
We’re also going to give Rack Studs a go this round. They snap in pretty easy, and the yellow plate fills the gap keeping the spring out. Rated for 44lbs…let’s see how this goes. ImageImageImageImage
Had a few people ask me about crimping previously. FWIW, I swear by pass-through connectors which are so much easier to deal with. Here’s how they work: wires through the end and crimper has a guillotine. Simpler and a lot less frustrating: ImageImageImageImage
Alrighty - ignore that camera cable that’s about a foot too short in the back (dammit!). Mounted…let’s fire it up and…go open all those patch cables I forgot to bring downstairs: ImageImageImage
And we’re live! 10Gb SFP+ transceivers up and running fine: ImageImageImage
Someone asked what it looks like on an end - here ya go: ImageImageImage
Cables! White is primary/ trunk network, orange is 10Gb, blue is any guest, red will be for external VLAN (fiber will terminate in this switch when it’s installed). I know I’ll need >6in later - just got enough for now and then I’ll see what distances I need much more easily: ImageImage
God. Fucking. Dammit. They shipped me back a non-PoE switch from the RMA. I just now realized it when wondering why my cabling wasn’t working and looking for power controls in the controller. The cables are fine, the devices just aren’t powered. Incredibly frustrated with @ubnt. ImageImage
It looks like the RMA lookup by MAC identified my old switch as a non-PoE for some reason. Still not sure what's going on yet, but the @ubnt folks are helping me track down what's wrong (so we can improve this process, for everyone).

I do appreciate them working to debug here.
@ubnt I am quite (I think understandably) frustrated at a series of events, which happens. Support folks who help debug issues despite that: you are saints. I have to walk away from the computer sometimes to write a level-headed response to things, where y'all deal with this all day.
@ubnt Props to @spencersoo for helping me figure this out on the backend and try to fix whatever process gaps led to paying DOA shipping (I've been told that's *not* normal, which is good), and getting an incorrect replacement. I hope some of our time spent here saves time for many.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Nick Craver

Nick Craver Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @Nick_Craver

8 Oct
I cannot wait to *not* maintain a build server anymore.
Building on the latest software on agents you maintain is fine when stable, and not fine when moving fast on platforms.

To canary Stack Overflow on .NET 5 I need to build on TeamCity, but it's not detecting .NET 5 right, so I need to upgrade, which means chasing a license and...
Yeah, it's yaks. I just want to run dotnet build and you think that'd be simple but no - it can't be simple. They have to do fancy detection and lots of mess.

On GitHub Actions we'll be able to just pick an SDK version to install and go anytime - a self-contained app to deploy.
Read 4 tweets
8 Oct
Was using the MacBook on my bike just fine about 20 minutes ago…opened the lid and no display (reboot, etc. doesn’t fix…trying more resets). How’s your morning going?
I’ve tried all the keyboard-based resets: no dice. Time to take it apart and yank the battery.
“The new normal” ImageImage
Read 11 tweets
2 Sep
To someone new, IMO a mentors primary jobs are:
1. Be there.
2. Be welcoming.
3. Be patient.
4. Make sure they're recognized.
5. Be a DNS server. They don't know to go or who to ask, you'll be asked for a "who...?" a lot. Documentation helps, but is rarely 100%.

#DevDiscuss
Mentoring comes first.
Your work comes second.
Ideally, pair on things to do both.

YOU NEED THE BACKING OF MANAGEMENT/PM/ETC. ON THIS.

The last part is critical. It's a marathon and you're supposed to be investing. Don't sacrifice tomorrow to ship something today.

#DevDiscuss
Mentoring cannot be an afterthought or something "on the side". It has to be a primary function and a priority. It's not fair to either side if it's anything less.

The org needs to make it a priority if they're serious about it. And if they do, it'll pay dividends.

#DevDiscuss
Read 4 tweets
7 Jul
We only cache a few things on the Question page for Stack Overflow (and all Q&A sites).

Loaded live (from SQL) are: your user (top bar), the question, answers, comments, users, your votes, flags (for mods).

We cache some sidebar: linked and related questions, for 5 minutes.
When you consider how many questions are on the network (over 27 million active), the chances of a user hitting the same question _on the same server_ (it's a local cache) within those 5 minutes is basically just the very hot or very new questions.

So is it worth caching at all?
Today, we're going to test.

I have a PR in that adds a setting to remove all caching and just query it live, as a single query with 2-3 result sets and Dapper reads them off. This means 1 SQL roundtrip for latency. I think it'll perform better, and eat far less cache:
Read 13 tweets
28 Jun
We give our new devs and SREs the ability to screw up (incl. taking down an entire service) their first week. Most manage to accidentally do it. And that's great.

They understand blame isn't a thing, more about that system in a hurry, and that decisions aren't driven by fear.
Most places have systems no one spends a lot of time on. A new person diving into those with a passion and wanting to take ownership it fantastic. If it starts with an oops, that's just fine. It's the cost of doing business and everyone's better for it. People are what matters.
We had an internal outage very few users saw this week.

What happened:
- Wrong server rebooted
- People learned:
- PTRs are in a bad state, don't trust them
- How we re-replicate Redis
- How to write a postmortem
- Lots of misc. around it

What didn't happen:
- Blame
Read 7 tweets
30 Mar
2 things:
1. Stack Overflow now has dark mode: stackoverflow.blog/2020/03/30/int… - after dropping IE 11 support, we could use CSS3 variables and make it all possible.

2. All Stack Overflow servers are running on .NET Core. We anticipate merging to master tomorrow.

It's launch day y'all.
We found a few minor glitches and one important one around duplicate cookie headers (super specific to our setup, but could happen to anyone). We rolled back to master to restore login for a small set of users, fixed the issue, and are rolling the .NET Core branch again shortly.
Nope - that approach had other issues. I will say Set-Cookie has been one of the most painful parts of the migration.

.NET Full Framework did far too much magic (to support IE 5) which we had to design around, while .NET Core swings entirely the other way: it's all totally raw.
Read 8 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!