As usual i'll explain why I constructed this tower and thought process. During my ongoing power saving observations I noticed both DAS units were drawing more power than they should have been for the number of disks in them, this is to be expected naturally as there will be loses in the conversion and cable resistance of power from the mains. I'm using relatively high powered PSUs in 800watt range for a lower powered job, their efficiency drops off at the lower power ranges, I was burning almost 40 watts more per DAS than I needed, thats enough for another four drives per unit at least! Smaller rated PSUs are available but many aren't modular and it's another cost. This lead me to the thought of powering all the disks from one PSU which should put the average power consumption well within the PSUs efficient range. I had a brief look online for cases that would allow me to max out as many HDDs as possible, of course the easiest thing to do would be to buy a large HDD caddy plus rack, needless to say an extremely expensive business!

Another consideration I had was my old ARM machine (automated DVD ripper), while it works perfectly, the 1 min boot up process wasn't ideal lol, first world problems. I wanted something that was always ready plus it was yet another stack taking up space and wasn't power efficient. The original plan was always to add optical drives to the DAS units then run ARM in a VM/Container.

The last consideration was space, why go horizontally when I can go vertically? It then occurred to me why not stack two of the DAS units, bolt them together, and have one PSU powering the whole lot? This is what i decided to do!

Materials/Cost list...

• 2 x Corsair Obsidian 750d - £50 (roughly)
• 1 x 20 Pack M6 (6mm) x 12 mm nut/bolt set - £6
• 1 x Corsair AX850 PSU - £100 (it was bought for a gaming system in 2003) YMMV
• 1 x LSI 9201-16e SAS HBA card - £50
• 2 x HP 487738-001 24-Port PCI Express x8 3Gb SAS Expander - £10 each
• 2 x Mini SAS26P SFF-8088 to SFF-8088 External Cable - £20 each
• 1 x 24 Pin PSU tester - £5
• 10 x SATA to 4 SATA Splitter power cable 45cm - Two pack £10
• 10 x 1 metre mini SAS 36P SFF-8087 to 4 SATA - £9 each
• 2 x Molex to 5 x 3 Pin connector for fans - £9 each
• 1 x Fan PVM Y splitter cable -£3
• 2 x 60cm 4-pin Molex extension cable - £6 each
• 2 x PCIe Riser, 3 in 1 1x to 16x PCI-E Riser Board - £30 for six pack
• 4 x 140mm Corsair Red LED fans - £25 pack of two
• 4 x 140mm Corsair Blue LED fans - £25 pack of two
• 2 x 120mm Corsair White LED fans - £28 pack of three
• 2 x Akasa PCI Slot Bracket for Mounting One/Two 120mm Fans - £6
• 2 x I/O Shield Replacement Backplates - £3 pack of two
• 12 x Corsair HDD upgrade cages, case comes with two - £10 each

Above is a list of everything required plus bare costs, I already had pretty much everything I needed already cases, fans, disks, disk cages from the other DAS units etc so my expenditure for this project was minimal. SAS cards and SAS cables plus the Molex extensions was all I needed. This just gives a rough idea if you wanted to build something similar. Total cost excluding drives £767, way, way less than a bare metal 32 chassis would cost you! More adaptable and configurable, with slightly reduced convenience due to no hot swap trays bit a lot quieter than the jet fans they put in these chassis. Ambient noise is about the same as a small PC on idle.

If you read my previous articles on building the DAS units you'll see I used 8088 to SATA cables connected to 2 x 16e LSI HBA cards on Ultron. If I continued like this I would need more SAS-8088 to SATA cables for all the HDDs possible plus the optical drives, a total of eight would be required with an additional two for the stacks of 3 x optical drives. Apart from being cable heavy, 2 x 16e HBA cards only gives me 32 possible SATA connections, another HBA would be needed! The number of HBAs required was starting to get a little silly plus slot heavy using up three out eight of my precious PCIE slots on Ultron. Thankfully my friends at IBRACORP had a better suggestion...use a SAS Expander!

Whats is a SAS expander? A SAS expander will allow you split the ports of your HBA card even further, in my case 24 drives per 8088 connector on the HBA! Each cable can support four lanes of 12G, 6G, or 3G, and the standard maximum transfer speed for hard drives is around 135MB/s. This means that it will take many hard drives to saturate the bandwidth of a single cable. SSDs, on the other hand, are much faster and can saturate the bandwidth of a single cable with fewer drives. I choose the HP 24 port model as it gives me plenty of drive options (it will actually support 36 drives) runs at about 11 watts and is stupidly cheap on eBay coming at £10-£15! I bought two of these, one for primary service and one as a spare. These cards were produced over a decade ago specifically for use in HP systems so they're getting old now hence the spare. In the end I ended up using both as it simplified the number of cables running between the two cases and halved the bandwidth requirements on one card, not a major issue give HDD speeds. I'll pick up another two cards as spares in the future. It's important to upgrade the firmware to the latest version, the cards are driver less, once powered and connected to your HBA use the command lsscsi -g | grep HP which will return the FW version, 2.10 being the latest and probably last ever revision. Instructions on updating the firmware here. The card has two ports you can use to internally (one system) link up to a HBA card, these can also be used for SATA drives increasing the possible number to 36. As my DAS is a separate/external unit i'll be using the external port. If you're buying one of these get the green PCB version not the yellow, the yellow can't be upgraded.

In an HP system they're normally powered by the PCI interface and rely on passive cooling provided by the HP enclosure. As we're using these in a non HP environment we'll have to cool and power the device without a motherboard, It's actually very easy to do! Thankfully the crypto mining industry has provided PCIE power adapters that many enthusiasts use in their mining rigs normally used with GPUs. I found a pack of six on Amazon for £30, these have a MOLEX interface which was important for me, covered later in this article.

As for cooling, I purchased a three pack Corsair 120mm white LED fans and mounted them below the SAS expanders using Akasa PCI slot brackets. This keeps the expander cool as they come with passive heatsinks only. Heatsink is on the underside of the card as normal so mount the fan beneath directing air on to the card.

One critical aspect of this build is the PSU, as I'm using only one, I need to power as many drives as possible plus the fans and expander cards. My old Corsair AX850 Gold PSU was recycled from DAS One as it had 5 x Peripheral/SATA outputs, 4 x would be used to power the HDDs/Optical Drives with 1 x used to power the expander cards and fans via MOLEX. As you may have worked out already, normally the standard SATA cable only comes with 4 x SATA outputs. I need 8 x per cable so the purchase of SATA to SATA splitters was required. It's extremely important to only use the cables supplied by your PSU manufacturer and good quality splitter cables. Not all the PSU cables have the same internal arrangement...DON'T MIX N'MATCH from different manufacturers! As a general rule, and depending on the power draw of your drives, 10 x drives per PSU SATA cable should be okay. I'll be using 8 HDDs per cable plus one or two optical drives powered by the same cable. I prefer doing it this way instead of powering the drives from MOLEX which can be a very dodgy thing to do, they have a habit of going on fire if you don't use good cabling! I'd rather not take the risk as old saying goes "MOLEX to SATA...loose your data!" See this link for determining good cables if you must.

Moving on to building the stack, the Obsidian cases would have to be modified as they were not designed to be stacked with most of the work required on the upper case. As you know previous articles, the Obsidian case I used for DAS Two was a wreck bought from eBay for next to nothing, this was the natural choice chopping! I started by cutting off the plastic legs at the rear before removing the front legs. On the Obsidian case range, the front legs are actually part of the front shroud so a smooth level cut is needed otherwise it will look terrible, cosmetics aren't really a consideration, functionality over form is.......but still. I purchased a small battery powered Dremel kit for the cutting and rounding off jobs. Remember to grind off any sharp areas!

Once chopped and the edges smoothed, I placed one on top of the other to start drilling holes for the bolts. Tape the cases together using duct tape for this part to ensure the holes match up. It was originally intended to use 8 x bolts spaced evenly around the top/base of the case, alas this wasn't possible, while there was plenty of space on the left side of the case, the right hand side is very limited and access for drilling was a nightmare! I ended up taking the laissez-faire approach drilling/bolting where it was easiest. Next, a cable channel was cut in the front/right side of the tower. I did this using the Dremel tool, make damn sure the edges are smooth, if not it will tear the power/data cables to shreds as this part of the case is metal. in retrospect I should have drilled the cable channel in the more towards the centre as some of the power cables are a wee bit tight. Tighten up the bolts and we're now ready to install all the gear.

Okay so now it's time to add the gear. I have the PSU installed in the upper case as this is closet to the middle of the tower, best place for cable length. Install the expander cards in the PCI slots along with the PCI MOLEX adapters and add the 120mm fan plus brackets underneath. Fan layout is as follows...

Upper Case Section

• 2 x 140mm Corsair Red LED fans - Extract
• 2 x 140mm Corsair Blue LED fans at the case front - Intake
• 1 x 140mm Corsair Red LED fan rear of the case - Extract
• 1 x 120mm Corsair White LED fan cooling the expander

Lower Case Section

• 2 x 140mm Corsair Blue LED fans at the case front - Intake
• 1 x 140mm Corsair Red LED fan rear of the case - Extract
• 1 x 120mm Corsair White LED fan cooling the expander

The lower case is exactly the same layout minus to two top red fans extracting. Depending on where you cut the cable channel, it's possible to add fans to the top of the Lower case in a similar fashion as the Upper case pushing air upwards. I haven't done this as yet as I'm not convinced its required. As there's no motherboard, theres going to be a gap between the MOLEX PCI board and the case. You may want add some thick non conductive packing here such as sticky foam strips. I haven't done this yet but it is planned.

Cabling wise we'll start with the Upper case. Connect the MOLEX Peripheral cable to the PSU then to the Upper case expander PCI board and use a MOLEX to fan splitter to connect up the fans. As there's six fans in this section you'll need to use a Y Splitter for two of them, I choose to put the rear and PCI fan on this splitter. Add two SATA peripheral cables to the PSU, then 2 x SATA to 4 x SATA splitters on each, this will be enough for 16 drives. Now add the Mini SAS 36P SFF-8087 to 4 SATA cables connecting the sas expander card to the drives. I found the easiest option is to build a stack of 8 x drives in each of cages, attach the power/data cables then slide the cages into the trackways and screw into place. Do the forward stack first then rear stack. Be careful when installing the drives in the Upper section, make sure the drive sleds are secure and the cages are screwed together. If you drop a drive or the whole stack from this height it's not going to be good!

Run two SATA Peripheral cables down the cable channel to the Lower case and run down both MOLEX extension cables. Connect up everything exactly the same as above. There's only 4 x fans in this section so only 1 x MOLEX 5 x fan splitter required.

As for the optical drives, simply add some more SATA breakout cables for power and connect up to expander by using the spare 8087 to SATA cables. I have SATA power coming from the rear HDD stack as these are backup drives which are not spun up 24/7 and NOT part of the array. All my array drives/working (archive/download/media cache) drives are in the forward stacks in the Upper & Lower sections primarily as there's better cooling available and these drives are mostly on 24/7.

Don't forget to install the blank I/O shield as it will improve airflow. Lower case PSU area is empty, I've sealed the rear off using black duct tape on my build as PSU blank plates don't seem to be a thin? Lastly, install the 24 pin power cable to the PSU adding the PSU tester to the end. This will let you turn the PSU on using the switch at the rear as there's no motherboard in the DAS. It's now time to connect the 2 x Mini SAS26P SFF-8088 to SFF-8088 external cables between your expanders and your HBA card in the main system.

Turn the PSU on and start Unraid (or other linux system)!

Open up an SSH connection and run lsscsi -g

All going well everything should be detected.

You'll notice at the very top of the list are two SSDs. These are on a separate HBA internal card on Ultron (another article on that later) which is designed for SSD speeds, don't be tempted to use SSDs on these HP expanders as they're not fast enough, it's a waste!

The optical drive use will be covered in another blog post.

Some notes on this build....

1. The Super DAS in my case is located in the attic and stands proud between two narrowish cross beams which will prevent it from tipping over if accidentally knocked.
2. Filled out with multiple drives it's HEAVY..beware!
3. As mentioned earlier its possible to install fans on the top of the Lower case.
4. Rear stack drives will always be hotter than drives in the forward stack, it is recommend to put heavy use drives in the forward stack.
5. Optical drives are an option obviously, you could add more HDDs in the 5,25 bay using a caddy.
6. The rear stack is capable of 4 x 3 drive cages/12 drives if desired! I've went with 8 x drives per stack as it's easier cabling wise.
7. DVD-RW are drives purely for ripping DVDs en-masse, you could add Blu-Rays if you wanted. No experience with writing disks in this setup, no reason why you couldn't.
8. It's possible to add another PSU to the Lower case for even more drives!
9. There's space for two PCI SAS expanders/120mm fans per case.
10. Obsidian case comes with 4 x SSD trays down the right hand side, with enough power available and appropriate expander you could add SSDs. I use Ultron for hosting SSDs as there's plenty of space in it's case with no expensive expander required.
11. All the original front I/O connectors are still present coiled up and stowed out of the way for future use if needed.

While ordering some more PIRs for the attic plus a manual switch from Localbytes I decided to purchase some wifi bulbs to try them out on a whim, plus I got free postage :-)

Smart lighting isn't something I was into as yet, mainly because the lights are hardly used here for various reasons...that's changed now after trying these little things out! Needless to say I'm rather impressed. They come in either E27 or B22 (converters for smaller sockets are available on Amazon) fittings at £12.00 and draw 9 watts. Installation was a breeze, plug it in and turn it on! Connect to the bulb via wifi and enter the HA/MQTT details same as the Tasmota Smart Plugs....nothing else required! The bulbs colour and temperature settings are instantly accessible just by clicking on the dashboard icon. There's even some built in effects.

In this post I'll briefly cover the installation of my Zigbee network in HA which is only the beginning. For years I was always dreaming of the possibility of temperature sensors in each room, wireless and done cheaply. Zigbee as been around for a long time but I've been a late adopter due to not being as seamless as I would have liked in previous years. Now another issue with Zigbee, as a wireless protocol, and in this house is the terrible attenuation caused by the limestone walls. Zigbee will require routers throughout the property which usually come in the form of light bulbs or switches. Neither was an acceptable solution in this place due to lack of neutral wire and that main lights are rarely used. It should be noted Zigbee works on the 2.4Ghz range which is the same as WiFi operating frequencies...roughly WiFi Ch 6. I had to put the WiFi channels on 1 & 13 see here for more info.  Thankfully Sonoff have come out with a range of exceptionally cheap and very good coordinators/routers/sensors.

Coordinator/Router

For every installation you'll need a Coordinator attached to Home Assistant which is usually via USB. The best they have so far is the 3.0 USB Dongle which can be used as a Coordinator or Router simply by flashing the firmware. It is highly configurable and will also do +20db plus other options if so required. I bought four of them, one as a Coordinator and three routers, Coordinator is plug n play with HA using the ZHA integration. The other three were flashed into router mode using PTIO firmware (see digblurs YT video). Seen has HA runs in a VM on Ultron all I had to do was plug it in via a USB cable and pass it through to the VM. I bought a long extension cable to run it into the apex of the attic. Added the ZHA integration in HA...that was it.

Now had the challenge of getting a Zigbee signal downstairs on the ground floor, the first floor wasn't really an issue as the ceilings are all plaster work and so getting a signal from the attic wasn't too hard. The ground floor is another matter though and also which ever router was down there would need a strong signal to get through the walls, basically the same issue as the WiFi APs. This is where these dongles are great with the option of the increased signal capability.

The solution was to place one in a USB wall plug (charging port in the upwards direction) on the first floor, one almost vertically beneath it on the ground floor and one additional router in the kitchen where there's a few sensors.

Sensors

I purchased 12 x Sonoff Thermometers, one for each room which gives both humidity and temperature readings. 5 x Sonoff Door Sensors, two for the front and rear doors plus three for the fridge/freezer doors which are routinely being left open! Additionally 3 x Sonoff PIR sensors for the attic which would control a bank of LEDs set on a automation timer with HA. Once the routers were in place installation is simple, place the sensor into pairing mode and scan for new devices via the android HA app with ZHA integration. Once discovered..name them...it's that simple.

Home Assistant

Thankfully due to the simple nature of these devices there's not that much to do in HA. I currently have a page for Temp/Humidity and one for Doors plus a few are tied with automations.

As I mentioned earlier I have 3 x PIR sensors for the attic which control banks of LED lights set on a 10 minute timer, 5 x Door sensors for the house's doors and also the fridge/freezer. If any one door is opened for a predetermined time and within specific hours, HA will send notifications to any of the Android clients or even my Discord channel. Monitoring pages are a simple case of adding some entities or mini graph cards. Below is an example of the code for the mini graphs (the dual line graphs) which can be obtained via HACS.

type: custom:mini-graph-card
entities:
  - entity: sensor.back_bedroom_thermometer_temperature
    show_state: true
    show_graph: false
  - entity: sensor.back_bedroom_thermometer_humidity
    show_state: true
    show_graph: false
  - entity: sensor.back_bedroom_thermometer_temperature
    color: green
    show_state: false
    name: Temperature
  - entity: sensor.back_bedroom_thermometer_humidity
    color: orange
    show_state: false
    name: Humidity
name: Rear Bedroom
hours_to_show: 24
points_per_hour: 30
show:
  name: true
  legend: true
  icon: false
  labels: true

Automations

For the attic lights it's a breeze, just select the motion activated light from the blueprints and pick the activation device, then pick the smart switch the lights are connected too. For the door sensors its simply a "Call Service" once the trigger has be activated, which in this case is one of the sensors being opened for more than 5 minutes with a specific time period. Fridge/freezer is all day for obvious reasons where as the front and rear doors are during the hours of darkness.

Equipment List eBay

• 4 x Sonoff USB 3.O Dongles £16 each
• 12 x Sonoff Termometers £9 each
• 5 x Sonoff Door Sensors £9 each
• 3 x Sonoff PIR Senors £9 each

I recently decided it was time to start building DAS Two as I've been collecting parts over the last few months on the cheap from eBay. At this time I don't have any new disks to add to the array however I split the current 16 disks between the two DAS units. Why am I doing this? Well the aft stack in DAS One runs about 5C hotter than the front stack of drives, I still haven't got round making fan towers for the after stack, and given it's the summer, I can easily fix that issue.

Back in May I came across another Corsair Obsidian 750d case on eBay for £30, it was mostly complete but had a 5.25 plate missing and also missing dust filters. Given these case's are getting old I wanted to snap up another one to match DAS One. Once delivered it also turned out the the forward plastic face plate was broken, easily fixed with some Gorilla Glue! For the build would need another PSU, Corsair LED fans to match DAS One plus a Molex fan splitter. I also picked up another LSI 16e card for £50 although it won't needed at this time. No additional SAS-SATA cables were required as I just took two for the aft stack out of DAS One.

PSU in the form of an RM850 was sourced off eBay for just £47 and also a 750d Airlifter kit for £25, everything else came from Amazon. Assembly was exactly the same as DAS One.

1. 1 x Corsair RM850 850 Watt 80 PLUS Gold  £47
2. 1 x Corsair 750d Airfliter Kit £25
3. 1 x Corsair Obsidian 750d case £30
4. 1 x 24 Pin PSU tester £5
5. 2 x SATA to 4 SATA Splitter power cable £10 (not used atm)
6. 1 x Molex to 5 x 3 Pin connector for fans £10
7. 2 x 140mm Corsair Red LED fans £32
8. 2 x 140mm Corsair Blue LED fans £22

Okay so I now have 8 disks per DAS, over the coming year I'll slowly start to replace the 6TB Reds in DAS One with 18Tb drives, the old drives will then be moved in DAS Two acting as back up drives for all data either in the same array or in a backup cache pool...maybe even a second array as apparently that may be on the horizon in the future! This will mean I'll need to install the other LSI 16e card in Ultron and source to more SAS-SATA cables. When it comes to that time I plan on installing SAS PCI SFF-8087 to SFF 8088 connector boards in the each of the DAS units because as it stands the units can be unplugged but will have long non-removable cables stick out the rear. The connector boards will allow removable SAS cables between the units and Ultron as internally the drives will be connected to the PCI boards. Not strictly a necessity but nice if I need to shift things about.

I've been unable to find replacement 5.25 faceplates, as you can see from the above picture one needs a serious clean, so I was wondering what could be done? Installing fan controllers was one option but wouldn't really add that much to the build. I then had the idea of installing some Optical drives for a disk ripping ARM VM! Essentially it will be the same setup as my ARM Machine but virtualized, as when ARM is out on loan I'm back to manual disk rips. I'll need more SAS-SATA cables for this so I'll probably wait until do the PCI SAS connector mod.

In this part we'll look how to add standing charges to the total cost and how to correct readings from device that only outputs wH.


Meross Plug Energy Correction

So as I said in Part 1, the Meross plugs I have don't output a kWh reading, only in wH. All we have to do is create a bank of template sensors for each one and divide the reading 1000 which will result in the kWh reading. These plugs don't have a ENERGY Yesterday sensor built in either, it's possible to get HA to calculate it if you wish. I haven't bothered as they'll all be replaced for Tasomata plugs in the coming weeks.

The wh reading is named sensor.attic_auxiliary_energy,  we then use the template sensor named attic_auxiliary_energy_kwh to perform the division sum, do this for each device.

## Meross Plug Kilowatt Correction from wH

      - platform: template
        sensors:
          attic_auxiliary_energy_kwh:
                friendly_name: "Attic Auxiliary Energy"
                unit_of_measurement: 'kWh'
                value_template: "{{ (states('sensor.attic_auxiliary_energy') | float / 1000) }}"

      - platform: template
        sensors:
          port_lamp_energy_kwh:
                friendly_name: "Port Lamp Energy"
                unit_of_measurement: 'kWh'
                value_template: "{{ (states('sensor.port_lamp_energy') | float / 1000) }}"

Total Cost Card

First of all we want to create a template sensor for the standing charge which in my case is £0.25 (due to double in the coming months!).

      - platform: template
        sensors:
          price_standing_charge_daily:
                friendly_name: "Price Standing Charge Daily"
                icon_template: mdi:currency-gbp
                unit_of_measurement: '£'
                value_template: "0.2500"

Now we want to sum together the total costs for previous day, daily, weekly and monthly. In my case I already know what the totals are for my equipment which is divided up between the Server Gear & Battle-Station as we set these up in Part 1.

Time to create another bank of four template sensors!

For the previous day and today figure we just add the standing charge sensor to the already totalled sensors for both sets of equipment. Weekly and Monthly are slightly different as we need to multiply the daily charge by 7 & 28 respectively, then add the totalled sensors for both sets of equipment.

## Cost Senors
## Totals for energy card

      - platform: template
        sensors:
          total_cost_yesterday:
                friendly_name: "Total Cost Yesterday"
                icon_template: mdi:currency-gbp
                value_template: "{{ (states('sensor.cost_server_devices_yesterday_power') | float +
                                                        states('sensor.cost_monitored_devices_yesterday_power') | float +
                                                        states('sensor.price_standing_charge_daily') | float) | round(2)  }}"


      - platform: template
        sensors:
          total_cost_daily:
                friendly_name: "Total Cost Daily"
                icon_template: mdi:currency-gbp
                value_template: "{{ (states('sensor.cost_server_devices_daily_power') | float +
                                                        states('sensor.cost_monitored_devices_daily_power') | float +
                                                        states('sensor.price_standing_charge_daily') | float) | round(2)  }}"


      - platform: template
        sensors:
          total_cost_weekly:
                friendly_name: "Total Cost Weekly"
                icon_template: mdi:currency-gbp
                value_template: "{{ (states('sensor.price_standing_charge_daily') | float * 7 +
                                                        states('sensor.cost_monitored_devices_weekly_power') | float +
                                                        states('sensor.cost_server_devices_weekly_power') | float) | round(2)  }}"



      - platform: template
        sensors:
          total_cost_monthly:
                friendly_name: "Total Cost Monthly"
                icon_template: mdi:currency-gbp
                value_template: "{{ (states('sensor.price_standing_charge_daily') | float * 28 +
                                                        states('sensor.cost_monitored_devices_monthly_power') | float +
                                                        states('sensor.cost_server_devices_monthly_power') | float) | round(2)  }}"

Remember to check the .yaml as we did before in the Developer tools before rebooting HA!

The next part is easy, create a new entity card on the dashboard and add the new totalled template sensors to it. This can be done in the GUI as it's a simple table card and you can use the sensors "friendly name".

Not Included Card

This card exists purely for keeping an eye on some low powered devices. It is not included in the total cost card mainly because there's hardly any usage . Like the total cost card above it's a simple entity card with the plugs kHw usage, which we made at the start of this tutorial, and the current wattage. As you can see it's barley measurable, note this is for a 24hr period.

After publishing I've been made aware this is the legacy method of making templates, they still work and will continue to work for sometime yet I would imagine. Post 3 will probably detail how to convert the current templates to the new method.

One of the key aspects with Home Assistant is the ability for remote control & monitoring, the mobile apps make this a breeze for both Android & iOS devices. They also include a variety of sensors native to the app from monitoring the device itself such as, but not limited too...

• Battery Charge
• Charging State
• Battery Temp
• WiFi details
• IP address
• Geo Location
• Storage

If your using tablets as remote control terminals, the inbuilt sensors are great for checking on the devices charging state etc. It comes into it's own really with the geo location options as you can tie this in with HA's automations so that when HA detects your device at home, it can activate anything you wish from lights, boilers, thermostats...you name it!

You can also configure HA via the app for installing/pairing other devices, VS Code is out though along with Node-Red, much better to use a desktop system for the real magic.

So what if you want to use the device outside of your LAN? Personally I use Tailscale zero config VPN on my mobile devices which enables access to all my subnets/VLANs and also VPN subnets in the other properties. This isn't built into HA, it runs in a container on Ultron and so isn't the solution for everyone. HA offers a built in Cloud service which is paid and will allow you to use HA with Alexa or Google home. It's also possible to use a reverse proxy, while I do have one that's heavily protected by Fail2Ban, it's not a secure as a private VPN. Over the next few weeks I'll be withdrawing most of my services on my RVP for Tailscale access only. You could be a complete amateur and forward the admin port to HA...DO NOT DO THIS! Your asking to get hacked then they can turn off/on whatever they want, there isn't any brute force protection at all!

To set it up simply download the app and login, that's pretty much it. Using a desktop device head into Settings - Mobile App then select the device, here it will display all the sensors available. You may find some appear to be missing, their not, by default the Companion App doesn't enable everything it finds...you need to enable them. Using the app go into Settings and scroll down, you'll see a section called Companion App, this is where you can enable sensors, change device name, NFC tags, themes...all sorts. Once you enable the disabled sensors they will be available to add to the dashboard.

Since 1998 I've always used Epson printers until this year. Printer models I have owned since 1998...

Epson Stylus 640 - Parallel Port only, good machine but had to move with the times.

Epson Stylus 830u - Junk! Leaked ink everywhere after a few months

Epson Stylus R240 - Still around somewhere, fairly reliable

Epson Stylus PX810FW - Someone binned it due to blocked heads...don't ask!

Epson Stylus PX830FWD - Still around but has dead/leaking print heads

Epson Expression Premium XP-830 - The final straw

Why was this the final year of Epson use? Well when I desperately needed to print something recently in black & white, the XP-830 wouldn't allow it as despite there being PLENTY of black ink, someone had used up one of the colours the week previous. As it turns out Epson gives you a few days grace between running out of one colour before it locks up! I'm not putting up with anti-consumer crap...so that was it, Epson and I are finished! On to eBay it went!

It just so happens I was looking for an upgrade anyway in the form of an AIO device that could scan/print/copy in A3. I also needed a device that could scan to FTP/SMB network folders without the use of a PC, it just makes things easier for those that struggle with operating tech lol. I also started using Paperless which is an awesome little project for electronic document management. Having the network scan option is great as I just need to load the ADF up, one set of documents/multiple pages at a time, scan to the folder then Paperless takes over which will categorize the scans according to heuristic learning/ set of instructions. Check out Paperless here.

None of the mainstream big names had such a device in an acceptable price range plus I was intending on fitting a continuous ink system, usually that meant performing invasive surgery/ permanent modification on the unit. I then decided to check out the second hand market, specially brother printers as they are in wide spread use in industry and are regularly swapped out/upgraded just like other enterprise hardware. I had used Brother units in my line of work before and while the software GUI leaves a lot to be desired, the functionality/features are excellent....On to ebay I went in search for such a unit!

MFC-J5335DW

I came across an MFC-J5335DW going for £70 which had all the features I wanted, seen as the previous Epson was sold for slightly more than £70 it was cost neutral...sweet! Anyway I bought the thing only to discover next day, after reading the manual, that Brother were a bit creative with their descriptions. As it turns out it is indeed an A3 printer but only an A4 scanner...feck! No big problem as it was more or less a straight swap for a better machine without Epson's BS firmware. So off I went in search again...I then discovered that the J5335DW has a bigger brother in the form of the J6530DW! Both from the same generation and take the exact same cartridges, always good for redundancy! Back on to eBay only to find one minutes later for £140...snapped that up instantly as they retail for a few times that.

A few days later the J5335DW arrives from ParcelForce. The seller had put the printer inside of a large, tight fitting plastic container, excellent idea. Unfortunately ParcelForce had kicked the thing all the way up from down south resulting in the plastic container smashed in one side, the printer though was okay after a few test scan/prints...or so I thought. A week later my sister came down to use the thing were she discovered the ADF on the unit wasn't working! Not surprised after ParcelForce had given it a good seeing too. It could have been that way before shipped, no idea.

I didn't want the hassle of returning it which would have been almost a 3rd of what I paid for the unit...it was surgery time. Pulled the plastic panel off to reveal the gears to see what was going on, the issue was instantly visible.

• Red Arrow - Electric Motor
• Blue Arrow -  Drive Cog
• Pink Arrow - Transmission Arm with Helical Gear
• Green Arrow - Document Pull Roller Gear

Right so the way it works is this...

• Motor engages
• Motor turns the drive cog
• Drive cog forces the helical gear in the opposite direction
• Transmission Arm is pushed in the down direction by the opposing rotation
• Helical gear engages with the Document Puller Gear drawing in the page

Why does it work like is? I assume it's so that if you need to pull jammed docs out in the opposite direction, that force isn't send back to the motor. I also means if it jams, the motor will instantly disengage. The issue was the Helical and Puller gear would not engage due to being slightly misaligned, they both had flanges which stood proud of the teeth preventing them engaging if not aligned properly. To fix the issue I tried a few things such as trying to reduce the clearances/play in the bearings by inserting thin metal plates or wrapping wire round the bearing, nothing really worked, it was hit and miss. Then I thought well all I need to do is hold that arm down for the gear teeth to stay in contact. The solution....a chopped down piece of Stadler eraser lol, it had enough rigidity to wedge itself in there and absorb a small amount of feedback plus it was easily removed if required. I had it scanning piles of blank paper for hours testing it out.....not a single hiccup, ADF has functioned perfectly in the months since.

MFC-J6530DW

Arrived without any issues and performing well, does lovely pictures in full A3 glory. I bought a CIS system for fitting once the current carts run out...post to follow! Another post to follow when the CIS is fitted.

Specs

Both printers have the same features available including a highly configurable Web UI for setting up the network scan options.

MFC-J5335DW Specs

MFC-J6535DW Specs

• Print from - Facebook, Dropbox, Dropbox for Business, OneDrive, EVERNOTE, OneNote, Box, Box for Business, Flickr, Google Drive, Google Apps for Work
• Scan to - Microsoft Office, Scan to Searchable PDF, Easy scan to Email, Scan to Mobile App, Scan to SharePoint (via Control Center 4 only)

Home Assistant

There's rather good HA integration for Brother devices enabling monitoring of ink levels and such. Just install the integration and add the printers IP.

Now we're going to build a dashboard just like mine for energy monitoring. A new Energy tab was introduced to HA last year which is more designed for utility monitoring electrical mains, gas, solar panels. I'm not setup for that as yet although it will be an option in the future. It can do individual devices but it's a bit rudimentary for my likes. I wanted a dashboard that I could look at the current draw in Watts and calculate the devices usage/cost in kWh/£ for daily, weekly, monthly and the previous day. This will give me reasonably accurate figures to plan my energy usage around plus the ability to turn on/off devices. I should note this is give me a rough idea of the costs, it's never going to be 100%.

As an example we'll use the server gear in the attic as it runs off two plugs but this can be adapted for as many devices as you want, just change the names of the sensors! We'll need a few things to start with which can be obtained from the HACS plugin that we installed in the last post, install these!

• Multiple entity row
• Mini graph card
• ApexCharts Card

Tasmota Sensors

Okay so there's a few of the smart plugs sensors we will be concentrating on apart from the outlet on/off function.

• ENERGY Power in Watts - For current power draw
• ENERGY Today - For usage & cost calculations, daily/weekly/monthly etc are all based off this....it's an important one!
• ENERGY Yesterday - Same as above but only for the previous day

These sensors will all have a unique names based upon the name you have given the device in HA. For example..

• ENERGY Power (w)- sensor.ultron_energy_power
• ENERGY Today (kWh)- sensor.ultron_energy_today
• ENERGY Yesterday (kWh)- sensor.ultron_energy_yesterday

HA Configuration.yaml

You'll need to use the VS Studio Code plugin or other file editor for this part, Studio Code is the preferred option as it's runs inside HA and features syntax highlighting. I've you haven't already installed it then do so via Settings - Add-Ons - Add-On Store. Once installed and started it will show up on the left hand side bar, open it up and select the configuration.yaml. This file is where we tell HA what to do with the data from the sensors and how to turn those readings into a monetary value for example.

Utility Meters

This function allows HA to track cumulative usage for what ever the input sensor value is, for example, if I wanted to know what the weekly, monthly, yearly reading in kWh usage was for Ultron I would setup a meter to track usage for each of those time periods based upon the ENERGY Today sensor value from the plug named sensor.ultron_energy_today

 # Utility Meters
    
    utility_meter:
        ultron_energy_daily:
          source: sensor.ultron_energy_today
          cycle: daily
        ultron_energy_weekly:
          source: sensor.ultron_energy_today
          cycle: weekly
        ultron_energy_monthly:
          source: sensor.ultron_energy_today
          cycle: monthly
        ultron_energy_yearly:
          source: sensor.ultron_energy_today
          cycle: yearly

As you can see the first line features the entity ultron_energy_daily The plug sensor already tracks the daily value however I put it in for ease when it came to easy to remember names in the dashboard cards. HA will automatically sum the usage for the cycle period specified. I created a bank of these for each plug I wanted to monitor, all eight of them. I haven't listed them all here as there's too many, see below...

 # Utility Meters
    
    ultron_energy_daily:
      source: sensor.ultron_energy_today
      cycle: daily
    ultron_energy_weekly:
      source: sensor.ultron_energy_today
      cycle: weekly
    ultron_energy_monthly:
      source: sensor.ultron_energy_today
      cycle: monthly
    ultron_energy_yearly:
      source: sensor.ultron_energy_today
      cycle: yearly

    das_one_energy_daily:
      source: sensor.das_one_energy_today
      cycle: daily
    das_one_energy_weekly:
      source: sensor.das_one_energy_today
      cycle: weekly
    das_one_energy_monthly:
      source: sensor.das_one_energy_today
      cycle: monthly
    das_one_energy_yearly:
      source: sensor.das_one_energy_today
      cycle: yearly
      

Note there's no utility meter value for yesterday because the plug natively tracks that. It's important to pay attention to indentation in the YAML, also all similar entities must be under one heading, e.g. all Utility Meters MUST be under utility_meter: and like wise all sensors must be under sensor: Go ahead and create the same in your configurations.yaml for all the cycles you want changing the name and sensor value as appropriate....remember to keep it simple! It will take a few hours for HA to start populating the utility meters with data but we can go on to the next stage.

Energy Consumption Card

The above dashboard card uses the Multiple entity row which we installed earlier. We can use the utility meters to give the value for Ultron & DAS One for each of the cycles you can see above. As you'll notice we also have a row totalling the energy usage and it's projected cost, where do we get these values from you may ask? Well that's where Template Sensors come in!

Template Sensors

This part concerns entities known as template sensors which derive their values from other data such as our Utility Meters. I'm using them to sum the value of the utility meters for both devices (plugs), all four time periods/cycles & the cost. We need to create another section in the configurations.yaml called sensors: and give our new sensors a name, friendly name (used for the GUI), a unit of measurement, icon to represent it and finally the sum itself. The four sensors will be named...

• server_usage_yesterday_power
• server_usage_daily_power
• server_usage_weekly_power
• server_usage_monthly_power

sensor:
      - platform: template
        sensors:
          server_usage_yesterday_power:
                friendly_name: "Server Usage Yesterday"
                unit_of_measurement: 'kWh'
                icon_template: mdi:speedometer
                value_template: "{{ (states('sensor.das_one_energy_yesterday') | float + 
                                                        states('sensor.ultron_energy_yesterday') | float) | round(2) }}"
      
      - platform: template
        sensors:
          server_usage_daily_power:
                friendly_name: "Server Usage Daily"
                unit_of_measurement: 'kWh'
                icon_template: mdi:speedometer
                value_template: "{{ (states('sensor.das_one_energy_today') | float + 
                                                        states('sensor.ultron_energy_today') | float) | round(2) }}"

      - platform: template
        sensors:
          server_usage_weekly_power:
                friendly_name: "Server Usage Weekley"
                unit_of_measurement: 'kWh'
                icon_template: mdi:speedometer
                value_template: "{{ (states('sensor.das_one_energy_weekly') | float + 
                                                        states('sensor.ultron_energy_weekly') | float) | round(2) }}"

      - platform: template
        sensors:
          server_usage_monthly_power:
                friendly_name: "Server Usage Monthly"
                unit_of_measurement: 'kWh'
                icon_template: mdi:speedometer
                value_template: "{{ (states('sensor.das_one_energy_monthly') | float + 
                                                        states('sensor.ultron_energy_monthly') | float) | round(2) }}"

Okay so this gives four new sensors we can use in our totals row

So how do we get the cost for the energy? Simple, more template sensors! We're going to create another sensor called price_per_kwh which has the value value_template: "0.1875" indicating the price in pence (Pound Sterling). This makes things easier in the future if the cost changes as we can change the value here which will filter though the other calculations.

• price_per_kWh

## Tariff

      - platform: template
        sensors:
          price_per_kwh:
                friendly_name: "Price per kWh"
                icon_template: mdi:currency-gbp
                unit_of_measurement: '£'
                value_template: "0.1875"

Now we need to create another four sensors just like we did above for the kWh totals that we can use to multiply total kWh values with the price_per_kWh variable which will then give us the cost for each cycle...

• cost_server_devices_yesterday_power
• cost_server_devices_daily_power
• cost_server_devices_weekly_power
• cost_server_devices_monthly_power
  

## Cost Sensors
      - platform: template
        sensors:
          cost_server_devices_yesterday_power:
                friendly_name: "Server Cost Yesterday"
                icon_template: mdi:currency-gbp
                unit_of_measurement: '£'
                value_template: "{{ (states('sensor.server_usage_yesterday_power') | float *
                                                        states('sensor.price_per_kwh') | float) | round(2)  }}"


      - platform: template
        sensors:
          cost_server_devices_daily_power:
                friendly_name: "Server Cost Daily"
                icon_template: mdi:currency-gbp
                unit_of_measurement: '£'
                value_template: "{{ (states('sensor.server_usage_daily_power') | float *
                                                        states('sensor.price_per_kwh') | float) | round(2)  }}"



      - platform: template
        sensors:
          cost_server_devices_weekly_power:
                friendly_name: "Server Cost Weekly"
                icon_template: mdi:currency-gbp
                unit_of_measurement: '£'
                value_template: "{{ (states('sensor.server_usage_weekly_power') | float *
                                                        states('sensor.price_per_kwh') | float) | round(2)  }}"
                                                        

      - platform: template
        sensors:
          cost_server_devices_monthly_power:
                friendly_name: "Sever Cost Monthly"
                icon_template: mdi:currency-gbp
                unit_of_measurement: '£'
                value_template: "{{ (states('sensor.server_usage_monthly_power') | float *
                                                        states('sensor.price_per_kwh') | float) | round(2)  }}"

Worth noting that HA always places unit_of_measurement as a suffix which isn't ideal when dealing with monetary values, you can delete this part if you wish or remove the symbol.

Now we have all the outputs we need to make energy consumption card as seen earlier on. HA will need to be restarted, but before we do the Yaml needs to be checked for errors. Luckily HA comes with with a built in checker which will alert you if something doesn't look right. You need to fix this otherwise it may not boot! Head into Developer Tools - YAML - Check Configuration. Hopefully all is well and you should get a config valid message, if not go back and check everything!

Go ahead and restart HA and wait for it to boot. Head into the Overview tab and edit the dashboard, little icon on the top left. Now we want to set up the energy consumption card and click Add Card - Entities - Show Code Editor, this can't be done in the GUI.

Enter this into the code section swapping the entity names out for yours and hit save! Voila!

type: entities
title: Energy Consumption Server Gear
entities:
  - entity: sensor.ultron_energy_monthly
    type: custom:multiple-entity-row
    name: Ultron
    state_header: Monthly
    icon: mdi:gauge
    entities:
      - entity: sensor.ultron_energy_yesterday
        name: Yesterday
      - entity: sensor.ultron_energy_daily
        name: Today
      - entity: sensor.ultron_energy_weekly
        name: Weekly
  - entity: sensor.das_one_energy_monthly
    type: custom:multiple-entity-row
    name: Das One + Aux
    state_header: Monthly
    icon: mdi:gauge
    entities:
      - entity: sensor.das_one_energy_yesterday
        name: Yesterday
      - entity: sensor.das_one_energy_daily
        name: Today
      - entity: sensor.das_one_energy_weekly
        name: Weekly
  - entity: sensor.server_usage_monthly_power
    type: custom:multiple-entity-row
    name: Total
    state_header: Monthly
    icon: mdi:gauge
    entities:
      - entity: sensor.server_usage_yesterday_power
        name: Yesterday
      - entity: sensor.server_usage_daily_power
        name: Today
      - entity: sensor.server_usage_weekly_power
        name: Weekly
  - entity: sensor.cost_server_devices_monthly_power
    type: custom:multiple-entity-row
    name: Cost
    state_header: Monthly
    icon: mdi:currency-gbp
    entities:
      - entity: sensor.cost_server_devices_yesterday_power
        name: Yesterday
      - entity: sensor.cost_server_devices_daily_power
        name: Today
      - entity: sensor.cost_server_devices_weekly_power
        name: Weekly

Wattage Pie-Chart

This is a really easy one, no coding required in the Yaml although it does require you to have installed ApexCharts Card from HACS. It's purpose is to provide with an idea of what's drawing power at that current time. It can be added from in the same manner we did for the energy consumption card through editing the dashboard. Select Custom: ApexCharts Card and enter the following in the code editor adjusting sensor names out for your own...

type: custom:apexcharts-card
chart_type: pie
header:
  show: true
  title: kWh Today
  show_states: true
  colorize_states: true
series:
  - entity: sensor.monitor5_energy_today
    name: Monitor5
  - entity: sensor.monitor6_energy_today_2
    name: Monitor6
  - entity: sensor.tv1_plug_energy_today_2
    name: TV1
  - entity: sensor.tv2_plug_energy_today
    name: TV2
  - entity: sensor.tv3_plug_energy_today
    name: TV3
  - entity: sensor.optimus_energy_today
    name: Optimus
  - entity: sensor.primary_multiplug_energy_kwh
    name: Primary Multiplug
  - entity: sensor.secondary_multiplug_energy_kwh
    name: Secondary Multiplug

Watts Hourly Bar Graph

I use this for quick comparison between what my equipment is drawing a both locations, it's not strictly needed as I also have Pie Chart's that cover both values. This will require the creation of another two template sensors which will total all the wattage readings (ENERGY Power) for that group.

• server_usage_current_wattage
• monitored_devices_current_wattage

      - platform: template
        sensors:
          server_usage_current_wattage:
                friendly_name: "Server Usage Current Wattage"
                unit_of_measurement: 'W'
                icon_template: mdi:speedometer
                value_template: "{{ (states('sensor.das_one_energy_power') | float + 
                                                        states('sensor.ultron_energy_power') | float) | round(2) }}"

      - platform: template
        sensors:
          monitored_devices_current_wattage:
                friendly_name: "Monitored Devices Current Wattage"
                unit_of_measurement: 'W'
                icon_template: mdi:speedometer
                value_template: "{{ (states('sensor.monitor5_energy_power') | float + 
                                                        states('sensor.monitor6_energy_power_2') | float +
                                                        states('sensor.optimus_energy_power') | float + 
                                                        states('sensor.tv1_plug_energy_power') | float + 
                                                        states('sensor.tv2_plug_energy_power') | float +
                                                        states('sensor.tv3_plug_energy_power') | float +
                                                        states('sensor.primary_multiplug_power') | float +
                                                        states('sensor.secondary_multiplug_power') | float) | round(2) }}"

Check your Yaml again and reboot. You'll need the Mini graph card for this one. Add the same way as we did for the others and place this code into it...

type: custom:mini-graph-card
icon: mdi:flash
entities:
  - entity: sensor.monitored_devices_current_wattage
    name: Battlestation
    color: green
  - entity: sensor.server_usage_current_wattage
    name: Server Gear
name: Watts Hourly
points_per_hour: 1
hours_to_show: 24
aggregate_func: avg
show:
  graph: bar
  extrema: true

This is my energy dashboard below along with Battle-Station control board. It's a combination of everything we have implemented in this part, there are a few things that aren't covered such as the "Total Cost Including Standing Charge" card and the "Not Included in Cost" card, this will be covered in another part (as their currently in flux) along with how to turn the wH readings the Meross plugs give to kWh.

Here's my full .yaml for reference, personal stuff redacted of course.

This post will detail the hardware I've chosen for energy monitoring, remote control and basic cost analysis of all my tech gear running as, given the MESS the arse clowns we call "leaders" are making of the world, electricity is becoming expensive. Firstly a little bit of history to explain the choices I made regarding hardware.

LightWave RF

Back in 2012, I was an early adopter of the LightWave range of remote control devices. At this time I was really only interested in remote control and spent a good three figured sum on In-Wall Outlets and Dimmer switches. It wasn't a great success as the Hub/Network Bridge required was buggy and required access to the Cloud. Additionally, some of the outlets would jam in the off position requiring a complete power cycle of the ring main that particular outlet was on. The Android app was buggy, commands were slow and very often it would turn off neighbouring LightWave devices! Reliability improved over the years but some outlets still jam, they have been left in situ as I can't be arsed replacing them at this time.

Meross

In 2019 I purchased a range of Meross MSS310 smart WiFi plugs to enable remote on/off of the many screens at the battle-station. I wanted WiFi as the wireless protocol as I don't want yet another bespoke HUB just to use them with. Three of the screens were Samsung TVs so one had to get creative with the remote or they would all respond, the auxiliary screens were manual only.  They did feature energy monitoring but only a very basic sensor package, an additional correction would be needed in HA as the plugs measured Watt Hours (wH) not Kilo Watt Hours (kWh). I was planning on buying more but it seems that model was discontinued in it's UK form. Another drawback was the width, they would not allow neighbouring plugs to be inserted on a multi-gang. These have been kept for low powered lamps and such.

Tasmota Smart Plugs

The tool of choice was some UK smart WiFi plugs pre-flashed with Tasmota available here, again I wanted WiFi as the comms protocol given I now have excellent coverage for the entire property. Tasmota is an open source firmware for ESP chipset devices offering a wide variety of control options using MQTT, Web UI, HTTP or serial plus OTA firmware upgrades. Designed to fit into the standard outlet profile, there's no trouble placing a few side by side in a multi-gang. Why not In-wall versions? Two reasons.....can't be bothered installing them and easy removal/upgrade if required. Integration is seamless with HA and provides a good number of sensors for HA to work with such as seen in the image below.

I bought 8 of these devices for the Battle-Station and Server Gear divided up between equipment differently depending on the device. I'll explain this more so you can understand why some plugs were placed BEFORE the UPS and some placed AFTER.

All the equipment is on 4 x APC 750i Smart UPS's divided up into two sets, 2 x for the battle-station and 2 x for the server gear. As the server gear is running 24/7 I want to be recording the FULL power draw which includes the UPS overhead, on/off control while present, isn't a priority as this gear must always be running. The battle-station is not a 24/7 operation and so on/off control along with energy monitoring per device is required. In this situation the plugs are divided up per device AFTER the UPS. Obviously the full overhead is not recorded however we can estimate it usage by introducing a static figure in HA for usage/costs once I determine the overhead of the UPS. The breakdown is as follows...

Server Gear

• Primary UPS - Ultron-Server, on it's own UPS due to it's power draw
• Secondary UPS - DAS One plus Auxiliaries which include 48P Gb Switch, 8P PoE Gb Switch (cameras, Raspberry Pi's, Wifi APs) and the HDHomerun TV tuners

Battle-Station

• Primary UPS - Optimus-PC, on it's own UPS due to it's power draw
• Secondary UPS - TVs 1,2 & 3, Monitors 5 & 6

There are other devices on the battle-station secondary UPS which are not as yet on on smart plugs, mainly because they're hardly used or draw very little. These include a set of Logitech z5500 speakers, Wacom Cintiq 32" Graphics Tablet, Nintendo Wii U, 8P Gb switch & Brother Printer. IEC C14 plug adapters, which are a few pounds each off amazon, were required to use UK plugs with the UPS.

Home Assistant Configuration

Okay so first of all we need to configure HA to talk to your new plugs. We'll need an MQTT broker followed by the Tasmota integration, so head into HA settings...

• Add-Ons
• Add-on Store
• Search for Mosquitto Broker (MQTT Broker)
• Install Mosquitto and start it up making sure the watchdog option is enabled

Now we need to set up a user credentials for the devices to use...

• Settings
• People
• Users (Top Right)
• Add User
• Fill out details as appropriate making sure its set for local network only and NOT an administrator
• Reboot HA

Once HA has restarted we need to setup the broker with the new credentials created...

• Settings
• Devices & Services
• You should now see the MQTT broker integration
• Hit Configure
• Enter the Host IP of your HA
• Enter username/password you just setup in the last step
• Hit Next
• Leave everything as is and select submit

Now we have to install the Tasmota Integration...

• Settings
• Devices & Services
• Add Integration
• Search for Tasmota
• Install Tasmota
• Reboot HA

Tasmota Config

Time to setup the Tasmota plugs themselves. Plug it into the mains and scan for the Tasmota device via WiFi and connect, this will bring up the Tasmota admin page where it will ask for your WiFi SSID, you may want to do this on a phone/ tabletp if a PC is not nearby. Enter the WiFi details and connect, the device will then reboot. I have mine connected via my NoT VLAN SSID which is completely blocked from going online, I highly suggest you do the same! As there's no WAN access allowed, the NTP function is blocked along with OTA updates. In my case Ultron functions as the NTP server so I can instruct the plug to use it by using the command NTPServer1 ULTRONS_IP in the console. Firmware updates are performed manually although running a private OTA server is possible.

Once it's reset it will have a new IP which you'll find in your router management page. The device can now be managed by this IP address which I recommend you make static to save on any headaches down the road. Put the new IP into your browser and go to the admin page.

We will now name the device and enter the MQTT details...

• Select Configuration
• Configure Other
• Enter the device name and friendly name, you can also set an admin password for the admin page at this stage if you want
• Save

The device may restart at this stage...

• Back into Configuration
• Configure MQTT
• Enter Home Assistants IP
• Enter the MQTT username/password setup previously

Now we need to set the device to show up in HA...

• Back into Configuration
• Console
• Enter SetOption19 0
• It should return a result of {"SetOption19":"OFF"}

That's the device now fully configured for use in HA, go back into the integrations page on HA and you should see it showing up under the Tasmota tab. If you click on the device entry you should now see all the sensors the devices features.

In the next post we'll look at using these sensors to create an energy dashboard such as the one below...cool huh?

This is the first part of many in a series of Home Assistant posts detailing my setup using a combination of various hardware and instructions on how to duplicate my setup. Home Assistant is a central control system for home automation with a focus on privacy for IoT devices from different manufacturers through a modular plugin system and includes native integration components for wireless communication protocols such as Zigbee, and Z-Wave. It also supports automation either through the built-in GUI, YAML or via Node-Red. It can be run on ARM or x86 platforms on Linux, Mac, Windows, Pi, Android, Docker and others with Android, iOS, http remote control being a key aspect. You are not tied down to any one ecosystem and can be 100% off the cloud! Some examples of supported devices...

• Smart Plugs
• Smart In-wall Outlets
• Lights
• In-wall light controls
• Flood Sensors
• PIR Sensors
• Boiler Control
• Thermometers
• Humidity Sensors
• Thermostats
• Fire/CO2 Alarms
• Door Sensors
• Cameras

There's wide variety of hardware options for running HA such as dedicated SFF PCs, Raspberry Pi's, Macs/ Mac Mini's, Full PCs, Virtual Machines & NAS devices. Recently they released their own dedicated hardware named "Blue" which is a small compact device solely for running HA. HA needs to be running 24/7 so your hardware choice is critical as you don't want to be running a high power draw machine. HA is very light weight and doesn't need much to run on.

Basically it comes down to this, if you have mid range NAS/ PC or Server running 24/7 then run it on there, if not use, a low powered device such as the HA Blue or Raspberry Pi. Another good option are second hand SFF/ Thin clients from HP (EliteDesk) or Dell (Optiplex) which came be found for very responsible prices on eBay, often cheaper than Raspberry Pi's, they are also very low in power consumption. If I wasn't running a server then these would be my choice as you get the full power of the x86 platform, full SSD and better expansion options. If you do go for these find one of the T models as these are even lower in power consumption. You'll find that while they draw twice the power, at around ten watts Vs the Pi's four watts, the performance is many times that of the Pi. Raspberry Pi's a good machines but they are slower had can develop issues with the SD card being corrupted as HA likes to write data, also good luck finding one these days.

As I already have a 24/7 server in the form of Ultron, I'll be installing Home Assistant OS in a VM in place of the container version as it's much more powerful and is the recommended installation. To install a VM normally you'd use an ISO image however we'll be downloading the pre-assembled qcow2 vDisk image.

1. Go here and download the latest qcow2 vDisk image
2. Go to your Domains folder on Unraid and create a new folder for the VM, name it as you wish
3. Place the downloaded qcow2 image in there
4. Setup a new VM in Unraid by choosing the generic Linux profile
5. Name your VM
6. Select the cores you want to give it
7. Set the RAM amount (4Gb in my case, it won't need anymore than that)
8. Set the Machine variable to Q34-6.2 or whatever the latest is
9. BIOS should be OVMF
10. Set the Primary vDisk location to the folder containing the qcow2 image
11. Set vDisk bus to SATA
12. Set the Network Bridge to whatever you're using for VM's
13. Start your new VM

Okay now that it's running click on the VM and use the VNC remote option. Once HA is ready you'll be presented with the the HA CLI, just navigate to the http address or IP given to begin configuring HA. Just follow the on screen instructions to tailor the installation as required. I highly recommend you assign a static IP for your VM or this will create issues down the line!

So now we have a vanilla installation to work with, there's some addons I recommend installing at this stage for future use. These can all be found in Settings - Add-Ons - Add-On Store.

• Studio Code Server (VSCode) - HA is YAML based so we need an easy way of editing the various config files. This undoubtedly the best method and comes with syntax highlighting as not everything can be done in the GUI.
• Samba Share - You'll need this on the off chance you screw up your configurations.yaml
• Terminal & SSH - Same reason as above plus CLI interface
• HACS - Community store, you'll need this for the to install the dashboards covered in the next article.  Installation is different, see here

In the next post I'll cover the setup of the energy monitoring devices

For many years I've been drooling of the Ubiquity AP range of hardware controlled WiFi access points. They do have a lovely ecosystem of cameras, AP's, Switches, Doorbells, Firewalls etc. The problem I had with Ubiquity was the increasing cloud based nature and possible backdoors built into their product range, they've also had some monumental hacks of late which they're trying to keep quite. They are also very costly for what they do.

For along time my Draytek 2860Vn was my main router/firewall and also hardware controller for my two Draytek APs. Now that it's been replaced with an OPNsense VM firewall, it was time to start on the WiFi upgrade project. Firstly I need to explain the construction of this property a bit. The house was built in circa 1850 and as such, being of Victorian construction, the walls are made of sandstone blocks a good few feet thick. The house was built even before gas lighting was a thing and never designed for cables at all. Wifi in the early years was always a struggle especially back in 2003 when I put in the first Linksys WRT54G router. The solution to this problem was more AP's but in those days Mesh (which I still don't trust) didn't exist and so my only solution was to run an ethernet backhaul from the primary router sited downstairs to an AP in the attic. AP's in the attic didn't have the same problem with sandstone walls as the AP's downstairs as the ceilings are just plaster, no problem for on suitably mounted attic AP to penetrate and cover all the upstairs rooms.

Downstairs was another matter entirely, again due to the thick walls almost every room would need it's own AP and ethernet backhaul! The primary router was in the kitchen is at one corner of the house and so would really only cover that area and maybe the next room with drastically reduced bandwidth. Another AP would be required in the living room which was at the diametric opposite from the kitchen. With this in mind in 2008 I decided to to run an ethernet backhaul cable from the primary router to the attic and then also down to the living room. Luckily where the router was sited, a small trunk built in from the 1960s when the house got its first boiler. This was to house all the hot water pipes running into the attic header water tank. These were all disconnected in the mid 90s so the old pipes were available for running cables down or just pass them down the trunk space. Getting a cable into the living room was slightly easier as it still had an old TV ariel cable installed in the exact location I wanted to site the AP, the solution was to attach the CAT6 attic end and pull it down. At the time of running these cables I also replaced the ariel coax with a new cable and also a telephone line cable from the kitchen into the attic, ADSL was still in use at the time and I wanted future flexibility which and ADSL modem sited in the attic, in the end it never got used. As for garden access...forget it! Outdoor AP's in those days went for an easy £800 for a standard device and twice that for a PoE version, not including the PoE switch! It also meant running more exterior cables which are unsightly and a major undertaking on sandstone walls, they don't take cable tacs and the grouting just fell away.

This arrangement was okay for a while however at the time the additional APs were repurposed cheap routers, not the best for administration plus there was no roaming or zero hand off at all. In 2010 I decided each room was getting dual CAT6 lines as now most rooms had media centres (some SFF PC's and soft-modded Xbox's) streaming AVI's via SMB from a few NAS devices, this kind of bandwidth use would kill the WiFi. I also installed a cable for future outdoor AP that sat unused until recently.

The new Drayek routers introduced in 2013 had the ability to function as a hardware controller for Draytek AP's, these were okay but the zero handoff/roaming never worked well especially when I introduced VLANs. I went through two Draytek APs in the attic, each one would be fine for a while then started suffering from strange firmware issues, not responding, needing re-flashed etc. This was despite being PoE powered from a small Netgear PoE switch (for the 2012 PoE camera project) so if there was any surges the switch should have got zapped first, there was never any issue with the switch...strange. All the attic equipment is now protected by UPS. This brings me to present day where I had enough of the WiFi issues and also wanted to finally put in a garden/outdoor AP. Going back to Ubiquity, I was looking for a cheaper alternative and without the cloud nonsense they seem to be obsessed with. After asking about on the IBRACORP discord room the new TP-LINK Omada range was recommended. After some research Omada was absolutely perfect for my needs and at almost 1/3rd the expense of Ubiquity. The AP's would function on PoE 802 af where as Ubiquity would mainly need 802 at, no PoE switch upgrade needed...another expense.

Given technology had moved on, I bought two EAP225 Outdoor AP's, one for downstairs mounted on the pole that holds my two upper screens at the battle station and the other going in the attic. This was to evaluate the devices before buying a ceiling mounted AP for the attic which could only go there. The controller was then installed onto the unRAID server in a container, it was simply a case of plug and play, the container would send all the VLAN and SSID settings to the APs in seconds.  Zero handoff/roaming was absolutely seamless even with media players or VOIP. Bandwidth was also excellent on tests including coverage, now I only need one AP downstairs, one in the attic and one outdoor AP. All the media players were replaced with Amazon FireSticks years ago streaming from Emby (GPU Transcodes as the library is mainly x265). The Omada AP range comes in pole mounts, ceiling & wall plate versions, currently there is no desktop version.

Now the APs were proven after a few weeks of intense use I purchased a ceiling mounted EAP225 version to mount in the attic roof peak and moved the previous one (an outdoor AP) to the garden. The garden is about 35 metres in length with a small summer house at the bottom so this is where the real test came in. The AP's did not fail to disappoint! Even at the end of the garden I had a strong signal on all WiFi devices (mainly android phones) and excellent bandwidth at 600-700 Mbps..more than enough for streaming  1080p content. The android app is excellent also, you can control all aspects of your config very easily plus access information on all connected clients/neighbouring APs/statistics. Firmware update is also a single click to upgrade them all! The Omada container is on a NoT VLAN so it has no WAN access at all, no back doors!

In conclusion these have been an excellent choice and almost cost neutral as I sold my old Draytek equipment on ebay. The outdoor APs were £90 from Amazon and the Ceiling AP £65, controller is free if self hosted. Why not WiFI 6 you may ask? I don't as yet do 4k and it just isn't worth the expense at this stage, these will happily do 4k transcodes if need be, anything heavy runs off the CAT6 backhaul. Will it be an option in the future? Of course it will but when I have the need. Omada also supports other devices in the range such as switches and firewalls, the next stage will be to replace my 48 Port Netgear and 8 Port Netgear switch with Omada range devices. They also have a range of Omada WiFi 6 AP's coming out but I suspect these will require a 802 at switch which will be part of the external camera upgrade project.

**All Omada AP's come with a PoE injector so a PoE switch isn't required.

I'm always being given DVDs by a family member to rip. This stuff is generally special interest or British media that's not widely available online and so ripping is the only choice and for backup purposes. Currently on my daily driver PC I have 2 x DVD drives and so while ripping a few disks at a time is no big issue, I wanted a more automated process where the user simply puts the disk in the drive then the machine does it's thing, ejecting the disk ready for the next one. It also means I don't have to do it anymore and as I work offshore, I don't have a pile waiting on me when I return home.

I went in search of such software and came across the ARM project. ARM runs on Linux and is designed to do exactly what I need AND also rip Audio CDs and Blu-ray's. It can be run in a standalone system or on a Virtual Machine with the drives passed through. ARM is designed to process the media itself using Handbrake CLI however my implementation of ARM is slightly different. If you plan on ripping Blu-ray's you'll need the paid version of MakeMKV but not for DVD or Audio. ARM also comes with built in WebUI for changing various default options in the .yaml config file for each disk. There is a time delay between disk ingestion for the user to change these defaults before ripping begins.

ARM Features (From Github)

• Detects insertion of disc using udev
• Auto downloads keys_hashed.txt and KEYDB.cfg using robobrowser and tinydownloader
• Determines disc type...
• If video (Blu-Ray or DVD)
• Retrieve title from disc or Windows Media MetaServices API to name the folder "movie title (year)" so that Plex or Emby can pick it up
• Determine if video is Movie or TV using OMDb API
• Rip using MakeMKV or HandBrake (can rip all features or main feature)
• Eject disc and queue up Handbrake transcoding when done
• Transcoding jobs are asynchronusly batched from ripping
• Send notification when done via IFTTT or Pushbullet
• If audio (CD) - rip using abcde
• If data (Blu-Ray, DVD, or CD) - make an ISO backup
• Headless, designed to be run from a server
• Can rip from multiple-optical drives in parallel
• HTML UI to interact with ripping jobs, view logs, etc

I wanted a standalone machine that I could fit out with multiple drives then upload the ripped disks to Ultron (my primary server with Dual CPUs). Ultron resides in the attic plus all the internal SATA ports are in use with Cache drives. I also wanted a semi portable machine that I could give to family members where it would rip the DVDs to internal hard disk then upload it Ultron over the WAN by either FTP/SMB using Tailscale as the VPN tunnel. In the event their connection is too slow the I can set the machine to rip to its own Hard Disk then move the media to Ultron via my LAN when I get the machine back.

Moving on to the hardware part of the system. This is a zero cost project as I already had a number of old gaming machines all with either single or dual optical drives Anno 2011 sitting around. The system of choice was a Phenom based machine (specs below) as it already had one IDE DVD drive fitted plus a onboard GPU, the system will operate headless just fine though. I removed a drive from each of the other spare machines and installed them into this one, this gave me four drives.

• AMD Phenom II X6 1055T 2.8 Ghz CPU
• GA-MA78LMT-US2H motherboard
• 16 Gb RAM
• Corsair CMPSU-650TXV2UK 650 watt PSU
• ST500DM002 500 Gb Seagate HD
• 2 x Samsung SH-S223C DVD Drives
• 2 x ASUS DVD drives
• Cooler Master HAF 912 Mid tower

The OS of choice was Debian mainly because ARM has a fully automatic install process for Debian aside from a few variables the user needs to enter. Once installed it's a matter of editing the ARM config to suit your config needs. You'll be prompted in the CLI for any input the script needs, also if you have more than one drive you'll need to mount these drives in fstab. You'll have to figure out which device names correlate to the physical drives as you may find drive 2 is not second from top. An easy of doing this is tell it to eject drive 2 then swap the SATA cables over to reflect the physical layout and so on. The GitHub page has all the instructions for doing this here.

Currently, in it's primary mode, the machine is really just a extension of Ultron. I have ARM set to rip the disk to ISO saved to mounted network folder via LAN hosted on Ultron where it already has Handbrake monitor a watched folder for processing ISO's to MKV in x264, x264 is better for SD content. Handbrake then sends the processed files to a folder monitored by Emby so the files are available for viewing quickly before I categorise them properly. It's powerful enough to convert SD video but Ultron is much, much quicker and already has the profiles setup up. The Phenom machine is also very inefficient power wise, it's only powered on for the time it needs to do it's ripping, once it ejects the last disk...that's it...off it goes! Ripping to ISO only takes a few mins per disk, if I'm ripping Audio CDs then ARM processes to FLAC as it rips sending to Ultron on the fly.

Here's the entry in fstab for guidance. Note the section for mounting the network folder if you so choose. Please note that if you're going to rip to a mounted folder then the folder must be created by the user ARM. Both directories "/home/arm/ARM-DVDRips" & "/home/arm/ARM-AudioRips" were made using the ARM user account, don't use the root account to make these folders otherwise ARM can't write to them. You can then mount the folder at start up via fstab.

##<file system> <mount point>   <type>  <options>       <dump>  <pass>
##/ was on /dev/sda1 during installation
UUID=1cc6e953-ef5f-431a-8330-d5c316f2675c /               ext4    errors=remount-ro 0       1
## swap was on /dev/sda5 during installation
UUID=d79354af-e035-4b10-88bf-0f5023c17b4a none            swap    sw              0       0
/dev/sr0        /media/cdrom0   udf,iso9660 user,noauto     0       0
/dev/sr1        /media/cdrom1   udf,iso9660 user,noauto     0       0
/dev/sr2        /media/cdrom2   udf,iso9660 user,noauto     0       0
/dev/sr3        /media/cdrom3   udf,iso9660 user,noauto     0       0

//YOURSERVERIP/Videos/DVD-ISO/ARM-DVDRips /home/arm/ARM-DVDRips cifs username=YOURSERVERUSERNAME,password=YOURSERVERPASSWORD,rw,uid=1000,gid=500
    
//YOURSERVERIP/Videos/DVD-ISO/ARM-AudioRips /home/arm/ARM-AudioRips cifs username=YOURSERVERUSERNAME,password=YOURSERVERPASSWORD,rw,uid=1000,gid=500

/dev/sr0  /mnt/dev/sr0  udf,iso9660  users,noauto,exec,utf8  0  0
/dev/sr1  /mnt/dev/sr1  udf,iso9660  users,noauto,exec,utf8  0  0
/dev/sr2  /mnt/dev/sr2  udf,iso9660  users,noauto,exec,utf8  0  0
/dev/sr3  /mnt/dev/sr3  udf,iso9660  users,noauto,exec,utf8  0  0

In its secondary operating mode i.e. remotely, I installed Tailscale VPN. This will provide you a static VPN IP you can SSH into the machine securely over the WAN. Depending on the remote WAN bandwidth the machine can be set to either rip to the internal HD or to the network mounted folder as described before. My entire LAN subnet is routed via Tailscale so no change of IPs required in fstab for the mounted folder. To change the ripping directory I have two config files a .yaml and a .bak version, I just change the file extension for which one I want it to use as they are predefined with the directories required, then reboot. I may even write a small script to do this for me, laziness is key to an simple operation lol!

In order to login via SSH remotely you'll have to change this (assuming you have installed SSH to begin with)...

sudo nano /etc/ssh/sshd_config

FROM:
PermitRootLogin without-password
TO:
PermitRootLogin yes

Save and Exit.

It was finally time to retire my trusty Draytek 2860N plus router after a good 7 years. While the functionality of the unit was excellent it simply wasn’t able to handle the 350Mb connection I have and with 1Gb available in my area, I needed a future proof machine. There are newer Draytek models available of course but as these are SMB oriented units they come with a hefty price tag. The unit also functioned as a hardware controller for my Draytek access points, these are also being replaced. See the TP-Link Omada article. Also, on the market, are prosumer devices capable of those speeds, fine if your only have a few PC’s in your house but for Homelaber such as myself, these don’t cut it either, far too lowbrow and limiting. More advanced machines such as the UDM Pro do offer a high level of functionality but yet again at a high price tag and locked into Ubiquity's ecosystem, I really don’t like being locked down plus they had a huge security breach recently

pfSense enterprise grade firewall is the way to go and very popular in the hobby or it’s open source fork, OPNsense. Both of these offer a huge amount of options including Multi-WAN connections, enterprise grade firewall, additional packages etc. Well that’s the software part decided, what about the hardware? Well there are a number of options available, including small boxes designed solely for running Pf/OPN, good units but again expensive for what they do, you could use any old PC but the power usage would be ridiculous for the function, or you can run it in a Virtual Machine. As I like to tinker, and because I need to to be reliable, I settled on both Virtual and Physical machine. The idea is to run OPNsense on Ultron as it’s already on 24/7 and have the SFF PC as a backup for when Ultron needs to be shut down.

Okay so I needed some additional equipment to achieve this goal..

• 1 x Quad Port NIC (HP 593743-001 NC365T Intel i340-T4) for Ultron, ebay £14
• 1 x HP T610 Plus (fits a full PCI-E card) SFF PC, ebay £50
• 1 x Quad Port NIC (HP 593743-001 NC365T Intel i340-T4) for SFF, ebay £14

eBay is the place to go these days for ex-enterprise equipment, hardware is sold at a fraction of retail costs and as it’s enterprise grade, far more reliable and adaptable. The NICs are dirt cheap especially the OEM Dell/HP stuff plus they all use Intel chipsets. To get the VM firewall running all I need to do was install one Quad NIC into Ultron and pass through two of the NICS to the VM (making sure to VFIO it off). One NIC is reserved for future use and the other used by Unraid for containers. Ultron has two Intel NICs built into the motherboard already, one for VMs and the other for general access.

As for the SFF, install NIC, software using the downloaded ISO. The easiest way to do this is using Ventoy. Ventoy will turn a USB drive into a bootable stick with multiple ISO images of any description. I have a USB3 64Gb with Ventoy and all the common ISO images, truly excellent solution. I won’t go into much detail on the full installation process as Spaceinvader One already has an excellent series of videos on installing pfSense on both Virtual Machine and bare metal here.

Right so, as I stated earlier the HP SFF will boot when the VM is down or crashed. To achieve this I have a first gen Raspberry Pi that I use as a syslog server for Ultron. This was set to run a script a five minute CRON schedule, the script pings the VMs IP. If no response after three attempts, it then sends a WOL packet to the HP SFF MAC address booting it up. Now this is purely a backup scenario, not High Availability (which is build into pfSense & OPNsense). The issue here being is that any changes made in the VM config will not be available in the SFF. Currently that is a manual job of downloading the config from one and writing it to the other. It’s no big deal really as all my VLANs, static IPs, firewalls are there already.

Regarding High Availability, its entirely possible between virtual and bare metal machines but you need at least three WAN IPs from your ISP. Currently I have a domestic connection with Virgin Media with only one. This is where the third spare NIC on both Quad cards will come in useful as they may be need for High Availability.

I ended up using OPNsense due to pfSense having an issue with the ISP provided cable modem. OPNsense doesn’t seem to suffer this so that’s the firewall of choice for now. Future mods will include a 4G backup WAN connection. I had no end of issues setting this up initially as the Virgin supplied Cable Modem (Super Hub 3) suffers from a variety of issues. One such annoying problem is it has a “Max-CPE” setting meaning it will lock to the first MAC address provided by the router, in my case the OPNsense VM. It would not give the HP SFF a WAN IP due to the change of MAC detected unless you power cycled the modem. Virgin, in the infinite wisdom, locked out any SSH/Telnet access meaning a manual power cycle, no use at all for a backup. Another solution would be to put the modem on smart plug and integrate that into the Pi script. To solve this I spoofed the MAC from the VM WAN NIC to the SFF WAN NIC, now the cable modem thinks it the same device, profit!

Despite what everyone says, the SH3 (cable modem) WILL issue a WAN IP to any port on the device but as stated it locks to the first MAC it gets. A further addition will be to add a script in Unraid that tells the SFF to power off when the VM starts.

If your going to do this  yourself you'll need to make sure your bare metal machine has WOL enabled in the Bios. As for the Pi script, instructions are as follows....

SSH into your Pi and run the following commands

crontab -e

Enter the following line at the bottom of that file

*/5 * * * * bash /home/YOURHOMEFOLDER/router-check.sh

The CRON job will run the script called “router-check.sh” every 5 mins. Make sure you change the YOURHOMEFOLDER to the one that exists on your machine. Save this by Ctrl+O and Ctrl+X to exit.

sudo nano /home/YOURHOMEFOLDER/router-check.sh

Enter the following making sure you change the VM IP and MAC address of the backup machine to suit your setup.

#!/bin/bash

VM_IP=200.200.1.1
BACKUP_MAC=f8:0f:41:48:f1:e1

# pings IP, prints status to console, and send WoL packet to MAC address if ping fails
ping -c 3 $VM_IP > /dev/null 2>&1
if [ $? -ne 0 ]
then
   echo "IP address $VM_IP is offline, sending magic packet" < /dev/null
   sudo etherwake $BACKUP_MAC -D
else
   echo "IP address $VM_IP is online" < /dev/null
fi

Save this by Ctrl+O and Ctrl+X to exit.. Reboot the Pi.

grep -i router-check.sh /var/log/syslog

This will tell you if the job is running.

With thanks to MountainGod on Ibracorp for the script, cheers mate! You can also get me over at Ibracorp discord as a Team member.

Last year I came across an article on serverbuilds.net with regard to building a high capacity DAS for Unraid. As the previous article on network reappraisal, I was looking for a more efficient way if using my drives, this was the inspiration. Equipment wise, I already had the case, drive cages (some were purchased for Janus 2)and a Corsair PSU from an old gaming system, plus a whole load of drives.

Some additional items were required to get this up and running..

• 1 x LSI 9201-16e SAS HBA card £50
• 4 x SFF-8088 to 4 x SATA 2 metre breakout cables £80
• 1 x 24 Pin PSU tester £10
• 2 x SATA to 4 SATA Splitter power cable £20
• 2 x Molex to 5 x 3 Pin connector for fans £20
• 2 x 140mm Corsair Red LED fans £25

Total additional cost approx. £200

The key to this project is the LSI 9201-16e SAS HBA card which enables connection of up to 16 SATA drives in an external enclosure using the 4 x SFF-8088 breakout cables, 4 x SATA drives per cable. The 24 Pin PSU tester connects to the 24 Pin power cable enabling power on without the motherboard simply by using the the ON/OFF switch on the rear of the unit. The PSU, which is from 2012, came with 2 x 4 SATA power cables and 2 x 2 SATA cables, the purchase of the SATA splitters would then allow for the full 16 drives planed. More modern PSU’s all come with 4 x 4 SATA power cables. Molex to 3 Pin splitters will provide power for the fans of which there are already 2 x 140mm fans in the front and 1 x 140mm on the rear. Additional fans in the form of the 140mm Corsair Red LEDs were added to the top of the case. The drives spin down when not in use so the PSU is only providing power for the fans.

The DAS must be left powered on at all times when Ultron is active otherwise the drives will drop out of the array destroying parity. Both machines are on a 500 watt APC Smart UPS each, Ultron draws much more power than DAS does and so in any power failure situation, Ultron will shut down long before DAS runs out of juice. Even at full load DAS draws about 50 watts vs Ultron's idle 180 watts.  It’s possible to link the PSU’s together so they both turn on from the server case switch but this involves sourcing 24 pin splitters and such, another consideration would be a flimsy connection between the two cases. I may do this mod in a future project but it's unnecessary for now.

Moving on to the subject of cooling. The forward stack remains cool with drives ranging between 20c - 25c under full load however the after stack averages a few degrees more from about 24c - 28c due to the decreased airflow. To combat this I came across two GPU fan ducts that came from some old Cooler Master cases designed to fit 120 mm fans. The plan will be to stack these vertically blowing on the after stack then build some shrouds using plasticard around to direct the airflow. I'll probably buy another cage for the aft stack to shunt the drives up one cage height. This will give increased airflow at the bottom of the stack. Still a WIP though as it might not be needed, I’ll see what happens when the summer comes.

10 drives were installed at first, all data was transferred off the last 10 bay Qnap via LAN using Midnight Commander. Syncthing wouldn't work as it bypassed the way Unraid splits up the data filling one drive only. The remaining 6 drives were added to the array once this was complete along with the cache pool drives.

I decided to scrub the Janus project for a few reasons.

• Power consumption, each Janus would have it’s own CPU/RAM/motherboard to power, not needed as Ultron’s already there running 24/7.
• Additional Unraid license costs, each Janus would need a Plus license at $89, Ultron is already on Plus, upgrading to Pro at a small cost would give me 28 array drives, 2 parity drives and 30 cache drives.
• Eliminates using multiple LAN connections per machine
• Reduced network traffic
• More drives per case
• Only once case required for now
• Less cluttered case, 10 drives per machine would fit but not a lot of room
• Better thermals, No CPU/RAM/motherboard
• Reduced storage requirements from the Qnap days thanks to recoding HD media into x265. This halved my library in size although it did take five months!

Okay so the original network layout was this...

Server Ultron - 1 x 480Gb SSD for (VM storage)

- 1 x 480Gb SSD Cache drive (containers etc)

- 3 x 1TB Seagate Greens (data array and downloads one drive as parity)

Primary NAS - 1 x 10 bay Qnap RAID 6 (10 x 6TB WD Reds drives holding media and data)

- 1 x 8 bay Qnap RAID 5 (5 x 6TB WD Reds holding films)

Secondary NAS - 1 x 10 bay Qnap RAID 6 (10 x 6TB WD Red drives, backups from primary)

- 1 x 8 bay Qnap RAID 5 (5 x 6TB WD Reds, backups from primary)

Secondary systems booted once a week for backing up, Primary's ran 24/7 with drives all spun up which is a condition of RAID, not very power efficient. Another drawback with this setup was losing storage capacity due to the way RAID works, down 1x drive in RAID 5 and 2 x in RAID 6, this totalled to down 6 x drives across the Qnaps, this was not an efficient use of the disks available. With this in mind I decided the best course of action would be to integrate ALL these drives into one monolithic build, in this case Ultron. The new setup looks like this.

Server Ultron - 1 x 480Gb SSD for (VM storage)

- 1 x 480Gb SSD Cache drive (containers etc)

- 2 x 6TB WD Reds for parity

- 16 x 6TB WD Reds as data array = 96TB total

- 2 x 6TB WD Reds as download cache pool = 12TB total

- 3 x 1TB Seagate Greens as general cache pool in RAID 0 = 3TB total

- 1 x 6TB WD Red media cache = 6TB total

This uses 21 of the 6TB drives with 4 spares. The other 5 remaining all had the beginning of issues with reallocated sectors and were stood down from service. Cache pool drives are running 24/7 with array data drives spun down all day. This is a significant power saving as they only spin up when something on them is required. Data older than two weeks is moved off the cache pools every week to the array. Data that needs 24/7 access lives in the general cache and backed up to the array once a week. High value data (pictures, documents) on the array is backed up by writing to other drives in the array as with Unraid, the disks can be pulled from the array an read as normal unlike in RAID. This gives me a live copy and two backups. Previously media was backed up, it’s not anymore due to being easily replaced these days plus I’ve still got two drives parity protection. Parity drives run 24/7.

I’ll explain the cache pools a bit more...

Cache-VM - Stores Virtual Machines

Cache-SSD - Docker/Containers

Cache-downloads - Downloads go here plus seeding - RAID 0 for combined space

Cache-media - New processed media is stored here. Files older than 14 days are moved automatically to the array, this happens once a week

Cache-general  - Used as a general cache, temp transcoding, LAN cache, VM backups and also some large VM’s. As this pool is in RAID 0 you get all the speed benefits of an SSD with large amounts of storage, perfect for all caching jobs plus running a few of the large VMs

Ultron’s case holds the SSD/Cache drives plus Parity drives. All other drives live in an external enclosure, that enclosure being DAS One. In the future I'll build a secondary DAS enabling full weekly backup of all data once again. Due the fact I’m already using 16 of the 28 supported, DAS Two will all be shucked 12Tb drives. Parity's will also be replaced by 12TB drives as a necessity.

See the next article on the DAS build.