Question SSD Boot Mysteries

[edmann41]

I just wanted to post this info and see what others have observed.

I have a Raspberry Pi 4 B with 8GB. I configured it to boot off a 1TB M.2 NVME SSD about a year and half ago. I ran it headless and used Windows RDP to access it. The SSD was connected via USB and is hosted by a Suptronics X873 board. Everything worked great and the Pi always booted.

Fast forward to about two weeks ago. I upgraded to a 2TB SSD (1TB getting full fast), installing Raspbian instead of Ubuntu. I use this as a Plex Media Server and from all that I read, Plex is faster on Raspbian than Ubuntu. Still a headless Pi but I had to switch to VNC to access it. All seemed great until this past weekend.,,

I rebooted the Pi after a config change. All of a sudden the bootloader went nuts. Sometimes I would let the Pi sit and eventually the bootloader would find the SSD and up came the OS. Often it would just search ad nauseam and never load the OS. Occasionally I would reboot and the bootloader would find the SSD right away and load the OS. Suffice it to say that this whole process was a painful crapshoot and I could make no sense of it.

I went with assumption that the bootloader was somehow garbled , so I updated it. No luck - same situation. I even thought that perhaps the Suptronics firmware update I applied last year may have gotten whacked, so I reapplied that as well. No change.

Then a little, lost voice deep within my brain, the one from my ancient "professional" IT days (now I'm just a hobbyist), welled up unexpectedly. "Reseat the SSD" it said. "BAH, nonsense, I thought."

Yeah, you know where this is going. I removed the little set screw and reseated the SSD. Now it works fine (again). I have rebooted the friggin' thing about 15 times now. Every time it comes back up, no problem.

Can anyone shed light or corroborate on this bizarre behavior? I am beside myself. Baffled. I'm happy to have my Plex server back but have this uneasy feeling that witchcraft and ghosts are afoot. I don't like the feeling one bit...

Thoughts or personal experiences, anyone?

 

[OldSurferDude]

Good to know you listened to your little IT ghost. I just fixed a dishwasher that had the same symptoms.

Quite often, just enough corrosion builds up on the contacts to create a high resistance short or open. Reseating it removed the corrosion. That's why reseating works. I would suspect that your SSD drive was sitting on a shelf for quite some time.

I would suggest you shutdown, power off, and clean the contacts of the SSD well with an electronic cleaner. Be sure to clean off the cleaner If you don't have cleaner, clean the contacts with a new[ish] eraser (aka rubber).

We will be seeing more and more of this in the future. Used to be the fingers of a flat cable and other areas where contact is made was plated with gold. The bean counters have brow-beat the engineers to under-design contacts so that they last just longer than the warrantee period.

OSD

 

[edmann41]

Hello OldSurferDude,

Many thanks for your feedback and additional education on how contacts are being degraded to save money. Your explanation makes perfect sense and I do remember the gum eraser trick from days gone by - I'm going to do this right now!

Wow, a dishwasher, huh? You are a Renaissance man!

 

[edmann41]

Hello OldSurferDude,

Many thanks for your feedback and additional education on how contacts are being degraded to save money. Your explanation makes perfect sense and I do remember the gum eraser trick from days gone by - I'm going to do this right now!

Wow, a dishwasher, huh? You are a Renaissance man!

Hello,

As a quick follow-up on this thread, about 5 days ago this NVME failed. I could no longer read it nor could I format it on any OS! So, it looks like it was a lemon out of the gate.

Your guidance was still solid and I have seen proof of that on one device in particular we use at my job that has contacts both on the device and the charging cradle. We are forever having to clean both to get a good connection!

 

[OldSurferDude]

Could be a cold solder joint. Reflow the solder on all the chips in the NVME. Take care not to create solder bridges.

(another bean counter trick, shorten the time the parts are in the solder oven, also lower the temperature of the solder temperature profile.)

OSD

 

[edmann41]

Could be a cold solder joint. Reflow the solder on all the chips in the NVME. Take care not to create solder bridges.

(another bean counter trick, shorten the time the parts are in the solder oven, also lower the temperature of the solder temperature profile.)

OSD

Thanks OSD!

Luckily Amazon has agreed to take it back and give full credit. Not without a lot of wrangling, of course. Glad I don't have to try what you suggest!

 

[OldSurferDude]

good solution Too bad it was defective in the first place.