Low-Power SSD for amtm, Diversion, Entware and Tailscale on the USB 2.0 Port (USB 3.0 configured as USB 2.0)

[jksmurf]

My recent (succesful) Tailscale install on my RT-AX86U (plaudits go to @RandomUser777) prompted me to follow up on what I have been wanting to do for a while and replace my USB Thumb Drive (ext4) with an SSD, for reliability only; I do not care about speed or size.

Cognisant of the repeatedly reported issues with interference to the 2.4GHz Wifi (IoT devices) when using USB 3.0 and the solution (apart from the oft-quoted sheilding with Alumimium foil) of limiting a USB 3.0 port to USB 2.0, left me with trying to find an SSD with very a low power demand to meet USB 2.0 5V 0.5A spec. Yes it "can" do more than 500mA apparently, but perhaps not reliably. I did not want to use a powered hub, too much clutter,

In any case, as I only use it for Diversion, Entware, amtm and now this Tailscale install, I found a nice SanDisk 16GB SSD (A Sandisk U110 SSD M.2 SATA 2242 Solid State Drive ref. SDSA6MM-016G-1001s Model SSD0E97864) which is rated at 3.3v and just 0.25A. The M.2 NGFF SATA III enclosure I use (not NVMe) has a Voltage Regulator from 5V down to 3.3V. The SSD0E97864 was used in the Lenovo ThinkPad Yoga 14 series, the Yoga S1 20C0 and X250, some of these as small (fast) cache disks to accompany normal hard disks.

There are very few SSDs (apart from eMMC-memory based Disks, which are not strictly SSDs), that have such a low current rating, the others I found were 0.6A and up, see a few pics below. At 16GB it is more than enough for my needs, there are even some 8GB ones out there.

So I formatted the new SSD as EXT4 using the amtm "fd" utility, copied all the files across from the USB to the SSD across using Colin Taylor's excellent advice here (although you must also reboot after step 9) and now I have a fully running SSD (which essentially 'cloned' my Diversion and Tailscale setup, albeit I did not use any specific cloning commands or 3rd-party software). The SSD is (hopefully) more reliable than (and will last longer than) my trusy USB stick, which is now a backup!

Anyway, thought I would add this as a note for anyone thinking of the same.

k.

EDIT: If you read the thread it appears as if the USB 2.0 ”mode“ on the USB 3.0 port may only limit throughput; and that the USB 3.0 voltage and current is retained. TBC. This being the case attempting to find a low-power solution to match USB 2.0 is somewhat moot.

 

[L&LD]

Just note that small capacity SSDs (less than 128GB/256GB) are usually very low quality and would probably be prone to failure closer to a USB stick than the expected SSD experience would normally be.

At the low price point of a 256GB drive, that or larger is what I would recommend.

More nand, means more reliable. (and usually indicates that the controller is also of higher quality too).

 

[jksmurf]

Just note that small capacity SSDs (less than 128GB/256GB) are usually very low quality and would probably be prone to failure closer to a USB stick than the expected SSD experience would normally be.

Without appearing to challenge you just for the sake of it and while I am aware SSDs have error checking capabilties which move "bad sectors" (or whatever they are called) to good parts of the disk (which tends to suggest larger is better) can you please provide some background to that statement? i.e. is this actually tested and proven to be the case? The particular SSD I have chosen was for cache OS in a branded (Thinkpad S3 Yoga 14) system. Do these devices SSD's fail? Yes. Do they fail more than newer (larger) SDs? No idea. Will it outlast my USB stick? Hopefully .

At the low price point of a 256GB drive, that or larger is what I would recommend.

I do not disagree but this was never about price; it was, if you read the piece carefully, about finding an SSD that would run off the 0.5A that is all USB 2.0 can deliver (without extra power). Had a larger disk been able to do that (as well as this one) I might have considered it; but they couldn't.

More nand, means more reliable. (and usually indicates that the controller is also of higher quality too).

OK.

k.

 

[John Fitzgerald]

@jksmurf , How large of a swap file are you using? I ask because I started using the 5GB size for the reason that there is more surface area to write and take care of bad/used areas. (not because more space is required for the actual swap)
@L&LD uses a 10GB swap, and wondering if this also was a planned thought process as well, despite the larger number of add-on scripts.

 

[elorimer]

If the USB3.0 port is operating in 2.0 mode, is its power also limited to the 2.0 spec? I understand the speeds would be.

 

[jksmurf]

@jksmurf , How large of a swap file are you using? I ask because I started using the 5GB size for the reason that there is more surface area to write and take care of bad/used areas. (not because more space is required for the actual swap)
@L&LD uses a 10GB swap, and wondering if this also was a planned thought process as well, despite the larger number of add-on scripts.

When I first installed Diversion some time ago on an RT-AC68U or RT-AC86U I elected to use a 1MB swap as I read that it was probably sufficient.

More recently though I was never asked for it at the time of my Diversion install, nor on amtm. Reading this thread, despite the many opposing opinions, suggests if it is needed by the Addon it is installed and that my current main Router, an RT-AX86U probably does not need it, for the few Addons I am running. To answer your question, I do not have any swap file running at all. At least the SW option in amtm is not running, nor does there appear to be any option for one in Diversion (any more).

i get your point though, but it begs the question, if the 5GB or 10GB allocated never actually grows, then is it actually spreading out to different areas? I have no idea how they work, whether they reserve 5GB in one spot and then next time they move to another spot? If that is the case there would appear to be merit in assigning it solely for that purpose.

 

[jksmurf]

If the USB3.0 port is operating in 2.0 mode, is its power also limited to the 2.0 spec? I understand the speeds would be.

I really don’t know, but you definitely get the prize for the best question of the day.

It is a question I should really have asked first, rather than blithely assuming “USB 2.0 mode“ reverted the port to ALL the parameters of USB 2.0, including power, not just the transmission speed. ASUS FAQ suggests the speed.

I am not an EE so I really have no idea what would affect the 2.4GHz channel (in terms of interference), transmission speed or power(or both) or indeed whether the higher speeds can only be achieved with a higher power output.

To test this I should now go out and buy one of those little inline USB LED devices that tell you Voltage and Current in real time, but maybe some enlightened person here has the answer already, because they know (and can prove it) or have actually measured it with the setting on or off. If the power remains at the USB 3.0 900mA spec, there‘s a little more headroom to play with, but not a lot (see cards attached above). It should move SSDs into bigger capacities though, probably not a bad thing.

 

[bibikalka]

I have a cheap 240GB SATA SSD with a Phison PS3111 controller hanging on a simple USB2 cable adapter for a couple of years now without any issues, so I searched for its specs. Here is the official spec sheet:

Power Consumption
Table 4-2 Power Consumption of PS3111 uSSD
Capacity Flash Structure Flash Type Read Write Partial Slumber Idle
32GB 32GB x 1 TSB BiCS 3 860 825 18 12 300
64GB 32GB x 2 TSB BiCS 3 1,130 1,030 17 11 310
128GB 32GB x 4 TSB BiCS 3 1,135 1,330 17 12 310
256GB 32GB x 8 TSB BiCS 3 1,215 1,495 18 12 305
Unit: mW

NOTES:
1. The average value of power consumption is achieved based on 100% conversion efficiency.
2. The total measured power voltage includes 1.8V, and 3.3V

So if I assume 3.3V, and divide 1,495 by 3.3, I get 453 mA for max write, and about 94 mA for idle. There are probably even lower power SSDs out there, but this one is low enough as it is!

 

[John Fitzgerald]

When I first installed Diversion some time ago on an RT-AC68U or RT-AC86U I elected to use a 1MB swap as I read that it was probably sufficient.

More recently though I was never asked for it at the time of my Diversion install, nor on amtm. Reading this thread, despite the many opposing opinions, suggests if it is needed by the Addon it is installed and that my current main Router, an RT-AX86U probably does not need it, for the few Addons I am running. To answer your question, I do not have any swap file running at all. At least the SW option in amtm is not running, nor does there appear to be any option for one in Diversion (any more).

i get your point though, but it begs the question, if the 5GB or 10GB allocated never actually grows, then is it actually spreading out to different areas? I have no idea how they work, whether they reserve 5GB in one spot and then next time they move to another spot? If that is the case there would appear to be merit in assigning it solely for that purpose.

So how many Partitions do you have on the drive? If the max Swap size is 10GB, formatted to ext4 with Journaling enabled, that would leave about 4.8GB for any remainder second partition.
How did you setup the drive? You cloned, but what did you clone? How much of the drive is in use and how much unallocated space is there?

 

[jksmurf]

So how many Partitions do you have on the drive? If the max Swap size is 10GB, formatted to ext4 with Journaling enabled, that would leave less than 4GB for any remainder second partition.
How did you setup the drive?

Currently, just the 1 partition, EXT4, Journaling enabled, all per amtm script recommendations for formatting disks.

i do not currently run any swap file; it was not asked for, nor recommended during any addon installs.

 

[John Fitzgerald]

Currently, just the 1 partition, EXT4, Journaling enabled, all per amtm script recommendations for formatting disks.

i do not currently run any swap file; it was not asked for, nor recommended during any addon installs.

I edited my above post with additional questions. I assume the clone shrunk the space??? @ColinTaylor

 

[jksmurf]

I have a cheap 240GB SATA SSD with a Phison PS3111 controller hanging on a simple USB2 cable adapter for a couple of years now without any issues, so I searched for its specs. Here is the official spec sheet:

Power Consumption
Table 4-2 Power Consumption of PS3111 uSSD
Capacity Flash Structure Flash Type Read Write Partial Slumber Idle
32GB 32GB x 1 TSB BiCS 3 860 825 18 12 300
64GB 32GB x 2 TSB BiCS 3 1,130 1,030 17 11 310
128GB 32GB x 4 TSB BiCS 3 1,135 1,330 17 12 310
256GB 32GB x 8 TSB BiCS 3 1,215 1,495 18 12 305
Unit: mW

NOTES:
1. The average value of power consumption is achieved based on 100% conversion efficiency.
2. The total measured power voltage includes 1.8V, and 3.3V

So if I assume 3.3V, and divide 1,495 by 3.3, I get 453 mA for max write, and about 94 mA for idle. There are probably even lower power SSDs out there, but this one is low enough as it is!

Geez that’s great! I never found one that low in that size. That’s still within USB 2.0 spec.

I probably did not look at mSATA form factor that closely as M.2 is smaller, sexier and more readily available here .

I had a quick look but cannot see any mSATA ones below 700mA (which would work for USB 3 power specs of 900mA) but still a wee bit high for USB 2.0 at 500m, so @elorimer’s question is pertinent.

My understanding is that the Voltage has to be right but that the device will draw only what it needs; so with USB 2.0 specs being 5V x 500mA (2.5W available), when the enclosure steps down the Voltage to 3.3V, can we assume (barring some losses in the regulator) that the available current to the device can actually be 758mA?

This being the case a 700mA device might be OK. I will need someone to confirm this logic though.

 

[jksmurf]

I edited my above post with additional questions. I assume the clone shrunk the space???

I did not clone it John, I followed Colin’s instructions (see original post) and copied the files to a new disk, renamed and mounted the new disk and I was away.

How did you setup the drive? You cloned, but what did you clone? How much of the drive is in use and how much unallocated space is there?

Only about 1% of the disk is in use if I read this correctly, it wouldn’t be an issue for me to add a swap file, but if it was not asked for, is not going to be used, then I am not sure why I need it?

 

[John Fitzgerald]

I did not clone it John, I followed Colin’s instructions (see original post) and copied the files to a new disk, renamed and mounted the new disk and I was away.


Only about 1% of the disk is in use if I read this correctly, it wouldn’t be an issue for me to add a swap file, but if it was nit asked for, is not going to be used, then I am not sure why I need it.

Thanks, I wasn't asking you to create a Swap file. I thought that you were cloning the previous drive that had a 1GB Swap, so I was curious if in the way you achieved your result you somehow side stepped the Partition size and if the whole drive was in use IE no unallocated space.

 

[jksmurf]

Thanks, I wasn't asking you to create a Swap file. I thought that you were cloning the previous drive that had a 1GB Swap, so I was curious if in the way you achieved your result you somehow side stepped the Partition size and if the whole drive was in use IE no unallocated space.

My apologies for the confusion John, it’s well past midnight here and I think I inadvertently confused you by discussing what I “used” to have (Diversion with a Swap file) vs what I have ended up now, so I should probably wrap it up.

TBH I can’t remember when the swap file actually disappeared, I am not sure I removed it myself, but I certainly do not seem to have one now.

k.

 

[ColinTaylor]

I assume the clone shrunk the space??? @ColinTaylor

No. See the post that he linked to. He just copied the data to the new drive.

 

[John Fitzgerald]

as I only use it for Diversion, Entware, amtm and now this Tailscale install

@ColinTaylor , Can I assume what was copied was not bound by swap parameters for the scripts and maybe only the Tailscale info was copied?

Yes, I am confused.

 

[jksmurf]

@ColinTaylor , Can I assume what was copied was not bound by swap parameters for the scripts and maybe only the Tailscale info was copied.

Yes, I am confused.

FYI, the logs and details in the Diversion Tab of the Webui were all retained.

 

[ColinTaylor]

@ColinTaylor , Can I assume what was copied was not bound by swap parameters for the scripts and maybe only the Tailscale info was copied.

Sorry, I don't understand what you're getting at. He said he simply copied all the files from his old drive to his new drive.

 

[John Fitzgerald]

Sorry, I don't understand what you're getting at. He said he simply copied all the files from his old drive to his new drive.

Thanks, I am under the impression that info from AMTM, Diversion(logs) and Entware reside on the router side and would not have been located on the USB drive if no swap was used.
What was copied? and why would this method of moving info be preferred over starting fresh (not including the Tailscale info, which I get) ?

I'm only interested in this as a way of learning alternative ways to accomplish goals.(multiple methods)