Tuesday, December 28, 2021

Google error "Your Play balance can only be used in Canada"

 Similar to this post, except switching to mobile data did not fix it.

 Seems many people are having similar problems.

In my specific case, my son was trying to buy Robux.  I was able to redeem his gift card using mobile data (as in this post) but when attempting the purchase all I got was "Your Play balance can only be used in Canada".

To be fair, I am using a VPN - with a Canadian ISP address.  Turning it off did not help even though I live in Canada and my DNS was correctly reported as being in Canada.  Every setting I could find said the account was in Canada.

I was able to fix the issue by switching to mobile data, downloading the Roblox app on my own phone, signing in to Roblox as him, then paying for the Robux using my own payment information.

I could theoretically have done the same thing via a web browser, except that Roblox did not allow purchase of the # of Robux he wanted via web.  (On mobile, you can buy 80 Robux; on web, there is no option for 80.) 

This solution obviously does not fix the issue of him having a "stranded" Google Play balance that he cannot redeem.  I contacted Google about this issue and they had no answer.

I get that Google is trying to crack down on scammers but this is utterly absurd.  I certainly won't be purchasing any more Google Play cards, and I highly recommend that no one else does either.


Friday, November 19, 2021

Stuck because iTunes won't let you update your billing information

I suppose this is now a Windows-only issue.

 

Problem:  iTunes stops at "Verification is required.  Please click Billing Information", or similar, but when you click the "Billing information" button, nothing happens.  Prevents you from buying anything.


Solution: There is something wrong with your Apple ID that needs to be updated.

The "Billing information" button is supposed to take you to a web page where you can sign in using your Apple ID, where you can view, edit and update your billing information.  However, the iTunes button doesn't always work, making it seem like iTunes itself is broken.

Most likely there is some security or billing update that needs to be completed, like 2-step verification or updating your credit card information.  Neither of these can be done in iTunes itself; they both need to be done at the Apple ID web site.

So, the solution is to:

•  Go to appleid.apple.com and sign in.

•  Check / update your security settings and billing information.

•  Restart iTunes.

You should be able to purchase things using iTunes as normal.


Thursday, October 7, 2021

Etsy "Sorry, there was an error" on credit card processing

If you get this error, you can try signing back in again.

In my case, I had started to sign in, but it went wrong and I closed the sign-in pop-up dialog.  Nothing seemed wrong.

However, this left the Etsy site confused, even though I had opted to checkout as Guest.  Re-signing in using a Google account fixed the error.

Sunday, October 3, 2021

Getting error "This gift card can only be redeemed in Canada" - even though you are in Canada

Note: this affects ONLY the situation where you have a gift card FROM YOUR COUNTRY and are actually IN YOUR COUNTRY.  If you want a means to redeem a foreign card, or redeem a domestic card while in another country, look elsewhere.

 

 This issue popped up recently on Google Play.

Location services were off, but turning them back on did not help.

What did work was turning off Wi-Fi and completing the transaction over mobile data.

This might work for other countries, I don't know.

Saturday, August 14, 2021

Don't buy a Ducati Hyperstrada

I rode an older Hypermotard in Europe, and really had fun.  Probably the most fun I'd ever had on a motorcycle, to be honest.  So I naturally wanted my own.

Unfortunately, I'm too short to be in the Hypermotard club.  I was on tippy-toe the whole week in Europe, and it was not a good feeling.  I needed something shorter.

Enter the late-model Hyperstrada.  Very much like the Hypermotard, but including panniers, all Hyperstradas after 2015 are 939s, and the "low" model.  After test-sitting one I knew it was OK for me.

Don't get me wrong: this bike is fun.  Damn fun, actually.  And far more capable than I am capable of riding it.

Unfortunately, it suffers from some glaring flaws that - somehow - nobody sufficiently points out.

One is the legendary grabby clutch, a seeming constant characteristic of modern Ducatis.  That goes away once warmed up, but is irritating.  Still, no bike it perfect, and it's annoying but tolerable.

Similar is the "false neutral" issue of the transmission, which can rear up at the least opportune times.  Never an issue on any of the other bikes I have ridden, this happens infrequently enough so that might be considered an infrequent quirk, although such a loss of control is (in my mind) far too dangerous to be so easily dismissed.

Of far more day-to-day concern is the horrible throttle response on the low end.  No longer a nice linear ramp, the throttle behaves far more like an on/off switch, resulting in a terrible low-speed riding experience.  Fixing this requires an ECU remap to correct the frankly abysmal low-end tuning of the 939 engine, at considerable cost, just to make the bike properly rideable.  

Still, this issue is fixable - albeit at great cost - and might be chalked up to a combination of Ducati "personality" or - more often - blamed on the European emissions regulations.  Why Ducati engineers simply can't make the bike properly rideable within the confines of the emissions specs, however, indicates to me a lack of expertise that I quite frankly find hard to accept.  Such is, after all, arguably the primary job of their engineering staff.

However, there is one "personality" issue that I have and am finding completely unforgivable:

The bike stalls.

Now, I'm not talking about stalling on startup, or in some kind of unique weather condition, or after tens of thousands of miles.  No, this bike - when NEW - will simply stall, randomly, when you pull in the clutch to come to a stop.

Don't believe me, check here, here, here, here, here, here, and here.

This is utterly and completely unacceptable.  There is NO CIRCUMSTANCE - whatsoever - where it is acceptable for a vehicle engine to simply randomly stall under normal driving conditions.  Not now, not in this day and age, nor in any other.  It simply SHOULD NOT HAPPEN. 

This is more than a simple embarrassment on how a "premium" Italian motorcycle is patently inferior to German or Japanese motorcycles.  The safety hazard here is clear and present, and surely has put more than one owner into seriously dangerous situations.  And yet it persists through all Hyperstradas of all model years and engines, as well as some Hypermotards as well.

How on God's green Earth does Ducati get away with this with hardly a peep from anyone?  The safety hazard associated with this is so obvious, yet nobody seems to raise anything more than an eyebrow.  No such issues are ever mentioned in "professional" reviews of the motorcycle, and the owner base seems to simply grin and bear it.

Most astonishingly, there appears to be no fix.  Nobody knows why this motorcycle, or it's 821cc sibling, simply randomly stalls at low speeds when the clutch is pulled in.  No one has developed a way to bump the idle up a few hundred RPM, change the mapping, nothing.  It's simply written off with a shrug as "a Ducati thing".

Some people do blame the charcoal canister,  some the ECU (needing an update), and so on.  But there is certainly no definitive fix.

Now, it seems the newer Hypermotard 950s don't have this issue - although they do / did have some issues with some kind of anti-corrosion gunk clogging up the throttle bodies.

So it seems that Ducati simply can't make their own 821/939 engines behave, and follows on with not knowing how to properly assemble the 950 engine.  So much more Italian engineering expertise.

I really wanted to love this bike, but this issue pisses me off.  With all of the improvements made in computerized engine management, a 30,000km 2001 CBR600 works better than a 3,000 km 2016 Ducati Hyperstrada.

Yes, I realize I'm probably unlucky.  Yes, I realize other owners haven't had issues.  Doesn't change my opinion much.

So, I would advise anyone looking to buy one of these: Don't.

 

[Update]:  Having ridden a 2024 Monster for 3,000 km, I can say the Monster does not have any of the issues of the Hyper, and is tons of fun if you have the roads available.  (IMHO the bike is wasted on anything other than a European Alps tour, or similar, but I guess it could be OK around the city if that's what you want.)  Roll-on and -off of the throttle in 2nd gear is perfect, no grabby clutch, no false neutral, no stalling.  (Also no saddlebags or panniers.)

The 2024 Multistrada RS, however, reportedly suffers from a very abrupt throttle response when transitioning from fully off (coasting / decelerating) to even slightly on-throttle in 2nd gear, which makes for an exceedingly jerky / abrupt and awkward transition.  This makes it (i) similar to the Hyperstrada, and (ii) unnecessarily difficult to ride, especially on those same mountain pass roads. 

The owner is convinced that the Multistrada issue is due to the two rear cylinders of the new Desmo v4 shutting off during deceleration, which is a feature of that engine that may be able to be deactivated by Ducati.  I hope he's right.

I do still have my Hyperstrada and will probably do for some time, but only as a city runabout and occasional highway day-tripper.  I did take it on a longer tour and it didn't exhibit it's issues to an extent that it affected by trip.

I had my doubts about the Monster but I think it handles - at least in 2024 form - much better than the Hyper, while the Hyper is more practical and a bit more comfortable on those straighter, longer, boringer highway jaunts.  (Not that my non-daredevil ass can get anywhere near the limit out of either bike, but still.)  I'm sure the Multistrada is a better all-around tourer than the Monster.

Tuesday, August 10, 2021

Everything you didn't want to know about installing new TPMS sensors on Lexus / Toyota vehicles

Note: this post is for Lexus, specifically the 4th Gen Lexus RX350 (2016+).  I am ASSUMING the process is similar for other Lexus and Toyota.

 

TPMS on the Lexus is a pain.

 On a GM vehicle, you do the following:

  • Install the new wheels with their new TPMS sensors on the car
  • Put the car into "relearn" mode; and
  • Poke each TPMS sensor with a $25 activation tool.

Done.

Not so easy on the Lexus/Toyota.  The basic steps are:

  1. Read the ID codes for the existing TPMS sensors (for future use)
  2. Obtain the ID codes for the new TPMS sensors
  3. Program / load these ID codes into an OBDII-capable TPMS programming tool; and
  4. Connect the tool to the car to upload the new sensor IDs to the car ECU.

The main issue with this is that many of the less expensive tools on the market will do only 1 or 2 of these steps.  You obviously need a tool that does all of them, and that means a more costly tool.

Unfortunately, vendors use confusing terminology such as "trigger", "activation", "re-learn", "reset", "electronic reset tool", and other misleading garbage, and often deliberately obscure the fact that some of their tools can't do some of these functions.   It's really all about what they don't say.

For example:

A.  The ATEQ Quickset will do 1, 3 and 4, but cannot do 2, because it cannot talk to TPMS sensors directly.  So if you don't have the ID codes for the new TPMS sensors, this tool will not work.

Reasons why you might not know the new TPMS ID codes:

  • You (or your shop) failed to write down the ID codes stamped on the TPMS sensors before they were mounted.
  • The sensors are aftermarket / generic, and did not have ID codes written on them.
  • The TPMS seller did not provide you with the ID codes.

If you do know - or can get - the IDs for the new TPMS sensors, the Quickset is probably a great tool, and cheap.  But if you don't already know the IDs, it's no help.

B.  Carista will do 1, maybe 3, and maybe 4.  It will not do 2 because it cannot talk directly to the TPMS sensors.

C.  The ATEQ VT31, VT36 and VT37 will do 1 and 2, but will not do 3 or 4, because they have no OBDII connection.  They talk only to the TPMS sensors, not to the car.

D.  The Autel TS401 and TS408 also lack an OBD connection, meaning they cannot do steps 3 or 4 either.

This leaves some people with a need for a tool that will talk to both the TPMS sensors directly and to the car via OBDII, and these are more costly.

At time of writing, the least expensive tools I can find are the ATEQ Quickset X and the Autel TS501.  These tools can talk directly to TPMS sensors and also have the necessary OBDII cable connection to allow complete identification and programming of new TPMS sensors for Lexus vehicles.  The Quickset X is the least expensive at about $300.

Notes:

•  I've not tried any of these tools myself (yet), so I can't confirm they work.

•  The TS501 includes features for testing keyfobs, programming virgin / blank TPMS sensors, etc.  The Quickset X (apparently) does not, but may be cheaper.

•  Some Lexus vehicles have a "TPMS reset" and/or a "TPMS initialization" procedure built in to the car.  These do not reprogram or relearn the TPMS sensor IDs, and so are no help when installing new TPMS sensors.  They should NOT be used, at all, when installing new TPMS sensors, or else the car ECU might "lock up" and prevent new TPMS IDs from being programmed at all.

•  Some Lexus cars will store two sets of TPMS IDs.  As far as I can tell, the only tool that can program the second set is the Techstream (?) tool used by dealerships.  All of the aftermarket tools will only program the first bank of IDs.

•  If you did buy a Carista, Quickset or similar "program only" tool in error, you might consider:

  • Downloading your existing TPMS IDs into the Quickset;
  • Having the dealership install and program the new TPMS sensors for you; and
  • Downloading the new TPMS IDs into the Quickset afterwards.

You will then have the TPMS IDs for both sets of wheels and can swap them at will yourself (at least until your existing TPMS sensors die in 7-10 years and you have to do this over again).

Alternatively, a dealership or tire shop might agree to just read the TPMS sensor IDs for you, for a fee.

•  Toyota dealers may be able to read / program Lexus TPMS systems, and may charge less than a Lexus dealer.

•  Many Lexus cars will determine the position of the TPMS sensors automatically, meaning there is no predefined order for programming them. 


Thursday, July 1, 2021

Successfully transplanting wheels from one car to another (2011 Ford Flex wheels to 4th Gen RX350)

Skip to the middle to get to the things that you don't find everywhere else.

Yes, there are a zillion posts about this.  None of them were actually complete, IMHO, so here we go.

 For this example I will use:

•  Donor:  2011 Ford Flex

•  Recepient:  2016-2021 (4th Gen) Lexus RX350

Why?  The used Ford wheels were 1/4 the cost of new Lexus wheels.


Bolt pattern:  Should be obvious.  Both cars use 5x114.3.

 

Bore / centerbore:  Ford is 63.4mm,  Lexus is 60.1mm.  As the Lexus shaft is smaller than the Ford wheel hole size, I need hubcentric rings / centering rings that convert 63.4mm to 60.1mm.

Note that these rings are entirely intended to center the wheel during the mounting process, and do not bear any load once the wheels are torqued down.  So they can be plastic or a lightweight metal such as aluminum, no problem.


Diameter:  Both cars use 20" wheels.  Important since too-small wheels will not clear the brake calipers, especially in the front.  If you are unsure, the only way to check is to test-fit the wheels.

 

Width:  Important since wider wheels may hit suspension components at the back side and/or fenders on the front side.  

A test-fit is the best option.  However, you can also use Tire Rack or other resources to determine the allowable width of a wheel on a car.

Here, both cars use 20x8 wheels, so no problem.  The Lexus is also rated to accept 18x8.0 to 20x8.5 wheels, so anything in that range should also work.

 

Offset: The Ford wheels are 39mm.  The Lexus stock wheels are 30mm, and the car can handle 27mm to 42mm.

Still, the Ford wheels will be 9mm further in.  There is a slight risk of hitting the front or rear suspension components.

To confirm, I ran a hockey puck (25mm high) all around the back of both the front and rear wheels.  There is at least 25mm of space there, meaning moving the wheels 9mm further in is no issue.

Note running wheels with an offset radically different than stock could lead to long-term suspension issues, even if the wheels "fit".


The above you find everywhere.

 

What you DON'T find everywhere:


Lug nuts:  The donor wheels / car may take different lug nuts than your recipient car.

To ensure they fit properly, you need nuts that do the following:

•  Seat:  The nuts must have the proper seat type for the donor wheels.  

The Lexus RX uses flat seat nuts, the Ford uses 60° conical nuts.  So, in this case, nuts must be 60° conical, and I cannot use the Lexus OEM nuts on the Ford wheels.

Note: in marketing-speak, "cone" means "conical".

•  Thread:  The nuts must have the proper thread for the recipient car.

The Ford uses 1/2"-20 threads, the Lexus uses M12-1.5 threads.  The nuts must be M12-1.5.

•  Length:  The nuts must be long enough for adequate engagement on the wheel studs without bottoming out. 

Regrettably, the only way to be absolutely sure is to measure the donor wheel when fitted to the recipient car.  

Bottoming out is rarely an issue provided you aren't running extended-length wheel studs.  But you must have a depth of engagement at least equal to the size of the wheel stud (i.e. M12 stud = 12mm minimum engagement).

[Note: some people say 1.5x.  As far as I can tell, they're usually talking about open lug nuts on racing wheels, which is a totally different application.  Such setups must pass racing safety rules, which are different than OEM specs.]

Note that cone nuts will fit into a matching cone recess in the wheel, giving you another 2-3mm of engagement.  If necessary, you can count turns to determine the actual engagement.

If you can't test / must guess, you can compare McGard or Gorilla offerings for both the donor and recipient cars.  Chances are that both cars use lug nuts that are between 1.45" and 1.5" long.  If the lengths are about equal, chances are the new nuts will have plenty of engagement.

There is also the fact that car makers can't be sure what wheels will be fitted, wheel makers don't know what cars they'll be fitted to, and lug nut makers don't know either.  So everyone plays nice to ensure that most wheel/car combos have more than sufficient engagement - meaning you're more than likely safe regardless of the wheel you stick on the car.

The Flex wheels allow 12mm of engagement plus another 2-3mm for the cone, for 14-15mm overall.  That's close to the fitment of the OEM wheels and beats the minimum by a significant margin.

Do note that even 1 additional thread after 12 threads adds 8%, two add 16% and 3 is 25% over minimum.  Those are non-negligible margins of safety, especially considering that the minimum engagement will include a significant safety factor.

•  Hex drive:  OEM nuts will probably be 13/16" or 3/4".  If you want to use the OEM lug nut wrench, you want the new nuts to match.  If you don't care, you can pick either, or go with a special drive type such as a spline drive or locking type.

•  Style:  You may have a choice between "cone" and "bulge cone".  

Both types seem to offer equal functionality - except for the hex socket size - but some wheels might have smaller indents for lug nuts than others, in which thinner "cone" nuts may fit better.  Otherwise the differences are purely cosmetic. 

Here, old Lexus 13/16" "bulge cone" lug nuts from a 2005 RX330 fit with over 12mm of engagement, with no issues fitting into the Ford lug nut recesses.  Given this, 13/16" McGard or Gorilla cone or bulge cone nuts would also obviously fit; 3/4" nuts would also probably fit.

 

Load rating: Not normally a problem, but if you're moving wheels from a lighter vehicle to a heavier one, it may be.

The Flex GVWR is 6150 lbs; the RX350 is only 5666 lbs.  So the Ford wheels should be OK. 

More specifically, the Flex wheels are rated for 2000-2200 lbs per wheel (depending on who you talk to), for 4000-4400 lbs per axle or 8000-8800 GVWR against:

•  The Lexus GVWR of 5666 lbs;

•  Assuming 60/40 front-back weight distribution (which is very conservative), that's 3400 lbs on the front Lexus axle; and

•  Assuming 65/35 distribution, that's 1700 lbs per corner, maximum.

So the Flex wheels are fine.

Also know that GVWR is the absolute maximum for a fully loaded vehicle, and wheels have a significant margin of safety built in.  Few people will ever load their cars to maximum, and the margin of safety will range from around 1.5x to over 2.0x.

Still, due to modern conveniences, many modern trucks and SUVs are considerably heavier than their older counterparts.


TPMS:  All newer cars have tire pressure monitoring sensors (TPMS).  The Ford sensors obviously do not work on a Lexus.

Of course, aftermarket wheels will also need new TPMS sensors.  Just don't forget them.

Generally, sensor makers will try to make one model of TPMS that works in most wheels and with most cars.  So - generally speaking - they're all physically compatible with all wheels.

In my case, I verified that ITM makes a Uni-Sensor Duo for both the Flex and the RX350.  That means the sensor with the Lexus interface is physically identical to the sensor used in the Ford wheels, so they must fit.


Center caps:  Purely cosmetic, so if you don't mind running Ford caps on a Lexus, no prob.

If you want replacements, you have to determine the center cap size of the donor wheels.  Sometimes the internet knows, sometimes it doesn't.  Sometimes you just have to measure them.


If you get all of the above, your donor wheels should be a good match to your recipient car.

Friday, May 21, 2021

Cost to upgrade underground electrical to 200A service

I was badly misled by online information on this one.  I don't know if that's because a lot of people have overhead service or not, but whatever.

Per Enmax, in the Calgary (non-downtown) area, the cost to update underground service to 200A is about as follows:

•  Transformer on your property:  $4,000 to $8,000

•  Transformer is 1 lot away:  $8,000 to $15,000

•  Transformer is 2 lots away:  $20,000 to $40,000

•  More than 2 lots away, or 2 lots away and across the road: $50,000 to $80,000

Most of this is the cost of underground horizontal drilling and Hydrovac, which obviously increase (steeply) by distance.  Given this I expect the costs would be similar throughout Western Canada.

About the only "good" news is that because they drill, there is no surface disturbance to roads or neighboring property.  But that's not any kind of consolation given the massive cost.

This is for a single detached home in the 'burbs.  Heavily urbanized or rural areas will probably be different.

Note this is just to install a new underground conduit.  Actually pulling a 200A (#1 aluminum) line, hooking it up, permits, new electrical panel or subpanel, etc. are all additional.

As a homeowner, they were not able to tell me the actual size of the existing line - not because they didn't want to, but they (seemingly) just didn't have the information.  But an electrician offered to find out, and it seems that when such requests are made from electricians, Enmax will go out and look at the transformer to see what size is currently installed at no cost.  I don't know how long that takes.

Regardless, if I can't stick with my existing service, any financial rationale for buying EVs just went out the window.  Even with fuel savings, there is no way an upgrade could possibly pay for itself in my lifetime.  

Still, as this article points out, all is not necessarily lost even if  you are stuck with 100A service.  240V 20A is usually enough to replenish daily driving needs overnight.


Thursday, March 4, 2021

Moving away from Teamviewer on older Windows systems

Problem:  You have an old XP (!) or Windows 7 box.  Teamviewer no longer supports your OS, and/or is endlessly bugging you with "Commercial Use Detected!" nag-boxes.

Solution:  Try AnyDesk.

Why?

•  It's free.

•   It has a portable version - you can try it without installing a thing.

•  It works on old operating systems (including XP (!))

•  It's pretty nice.

Some features are pay-only, any the subscriptions ain't cheap.  But, so far, I've not found anything the free version lacks that I need.

This saved a very old PC of mine after Teamviewer stopped supporting old boxen.  I need this ancient PC to run a prehistoric printer, with neither being supported by newer software.  AnyDesk saved it when Teamviewer stopped connecting in v15.

Sunday, February 28, 2021

Kodi randomly leaves full-screen mode to smaller window at top-left-hand side of screen (especially on Intel NUC), leaving it locked up under Windows 10

Problem: Per title.  Affected was Kodi Leia, but can happen to any Kodi.  May also happen to other media players.

 

Explanation:  This is a manifestation of the EDID issue (also known as the "quarter screen issue") where:

•  Windows is losing the EDID for the display device (TV, projector) when the display is shut off or disconnected.

•  Windows then resizes the screen resolution to the "default" display device, which forces Kodi to resize as well.  This usually makes the Kodi display lower resolution (i.e. smaller).

•  When the display is turned back on / reconnected, Windows re-acquires the EDID and changes back to the original resolution.  

Kodi does not handle this gracefully.  

The result is that Kodi gets "stuck" in the lower resolution, which manifests itself as a smaller (i.e. lower-resolution) window at the top-left corner of the screen.  

Some people find that although Kodi is resized, it is still responsive, and so can be restored to full-screen mode via keystrokes.  In my case, it also locks up, forcing me to kill and restart it each time.

 

This may or may not affect a HTPC directly connected to a TV.  It may also seem to occur randomly, not every time.

The problem does appear to be exacerbated when:

•  The HTPC is an Intel NUC; and/or

•  There is a receiver between the HTPC and the display device; and/or

•  The receiver is relatively old.

Keeping the receiver on full-time might not help since older receivers might interfere with EDID acquisition even when active.

This does not seem to occur on Android since Android doesn't handle EDID the same way.   So if I had kept my Shield instead of getting a NUC, I probably wouldn't have this issue.(*)


Solution: Buy a "EDID emulator" device (like this one or this one).  I tried the second one and it seems to have fixed the issue.

You can also get a "Dr. HDMI".  This is the same thing but more expensive.

These devices lock the EDID to >one< resolution only.  So be sure the purchase the one that matches the intended resolution of your TV, projector or monitor.


(*) I exchanged the Shield for the NUC because the Shield didn't correctly support analog 7.1 audio.

Sunday, February 14, 2021

VPN kill switch for single machine or IP address range on pfsense

There are obviously LOTS of ways to do this in pfsense.  This is just the way that I found, which seems to work.

This rule should prevent an individual machine from "talking" using the default non-VPN WAN connection.  This will prevent it from sending any traffic if the VPN goes down.

 Go to Firewall/Rules/Floating

  • Action: Block
  • Quick: Checked
  • Interface: WAN
  • Direction: any
  • Address family: IPv4
  • Protocol: Any
  • Source: Single host or alias / [set local IP address of host here i.e. 192.168.1.100]
  • Destination: any
  • Description: enter any description here [i.e. "Block this IP from using non-VPN WAN"] 
  • Click "Save"

Obvious notes:

-  This only really works if the machine in question uses a static IP address.

-  You can define multiple rules for different devices.

-  You can set the rule to block a range of IP addresses by using "Network" instead of "Single host".

This approach does not use policy-based routing and does not decide which machines use the VPN and which don't.  It just forces all traffic from a machine through the VPN, else the traffic gets blocked.

Speed change between Netgear R7000 / DD-WRT and Core I7-7500U / pfsense routers using PIA AES-256

This was a quick test only, using a popular torrent client.  

The Netgear is a recent (faster) example running the latest DD-WRT.  The pfsense box is an eBay i7-7500U special with the latest pfsense.

Speeds are peak speeds through Private Internet Acccess (PIA) using AES-256 / SHA256 on their "Next Gen" network.  There were no changes to the rest of the setup between the two trials.

Results:

  • Netgear: 4 MiB/s 
  • i7-7500U: 38 MiB/s

That's almost a 10x improvement.

Note I couldn't get these speeds using all torrents, as some of the sources are limited. 

However, I know that the Netgear can't exceed 5 MiB/s no matter what, while the i7 definitely can do 38 MiB/s.  It's possible the i7 might be able to go even faster.

From this, it does appear obvious that the Netgear is quite limited with respect to VPN crypto. 

Obviously the i7 is vast overkill.  I got it since the price difference between the i3 and i5 was nominal and I really only feel like doing this once.  See here for tips on buying such a box.

Tuesday, February 9, 2021

My experience with Montana Mailbox / DYK Post

First parcel, no issues.  Took a little while to get the incoming ticket, but didn't think anything of it - I figured the delivery had been slow, and I didn't care.  Picked it up at DYK without any issues.

Second parcel was Amazon, so I had a delivery date.  It took Montana Mailbox 7 working days (12 calendar days) to issue the incoming ticket. 

I wanted the parcel earlier, so I sent an email 4 working days after delivery, to see if Montana could expedite it.  I never got a reply.  I was hoping to get at least a "Sorry, we can't help with that", but nada.

MM is undoubtedly big, but I can't see them having a 7-day incoming queue.  So they obviously prioritize commercial traffic over individual parcels, meaning one-off packages will take however long they take.

It then took another 9 days for DYK to just pick it up, plus an additional 2 days to get it to the correct office.  So about 15 working days, or 23 days total, until it arrived.

Again, DYK is not likely to have a 9-day queue, especially when they pick up every day.  It may be the package got held up in Customs, which can happen to any package.

So, while the Montana Mailbox / DYK Post teamup seems a good service, do NOT use it for anything time-sensitive.  Or, if you do, allow at least 10 working days after delivery for Montana Mailbox to issue your incoming ticket, and another 10 working days for DYK to have it ready to pick up.


Monday, January 11, 2021

Wi-Fi interference on MX Revolution mouse

Check the wireless channel settings on your 2.4 GHz router.

I had been messing with my network and had set one of my routers to 2.4 GHz channel 3, which resulted in significant interference (skipping, lagging) on my MX Revolution mouse.  Resetting the router to "Auto" fixed it improved it somewhat.

Additionally, putting both my wireless APs on the same channel seemed to improve things as well.  They naturally did this themselves on their "Auto" setting, so I'm guessing it's somehow OK.

This might apply to any 2.4 GHz peripheral, but especially older ones that don't use Bluetooth.  My MX is old and is non-Bluetooth.

Alternatively:

-  If your router is set to "Auto", try setting it to a fixed channel;

-  Try moving the mouse receiver closer to the mouse; and/or

-  Try moving the router further away from the mouse.

These did not help my situation but are always worth a try.


Tuesday, January 5, 2021

Is Link Aggregation / LAG / LACP actually faster?

Simple answer:  

•  From a server to multiple clients, YES.

•  From a server to a single client, NO.

Why:  LAG/LACP does not combine several gigabit connections into a single fat connection. 

This is because all traffic for a single IP address has to go through one physical connection.  It's just too complex to do it any other way. 

So, a 4-port LAG is not a party line, where everyone can hear everyone else.  It's one person with 4 phones talking to four separate people each holding 1 phone - none of which can hear each other.

This means that a 4-port NAS can shove out 4 Gbps in total - but only 1 Gbps to each individual client.

(Yes, there are rare exceptions - just enough to make it unclear if LAG to a single client is really faster or not.  But it's not.)

This doesn't make LAG useless.  It just means it's only useful where you have several clients connecting to the NAS simultaneously.  A single client can't take advantage of the LAG, but multiple clients can.

This also applies if you connect two NAS devices together via LAG, and transfer / migrate / replicate data between them.  Each one is a single client, and so can only receive a single lane of data (1 Gbps) - not the 4 Gbps you might expect.