Sorry for the dumb question, but is 39.4 the percentage before it shuts down or something? It doesn’t change my analysis, just curious about that number.
This does sound like a hardware issue. The three most likely causes for this type of behavior are overheating, a failing PSU, and a failing NVMe. Incompatible RAM is another possibility (worth pointing out since there is an ongoing RAM crisis and some folks are getting creative in that department). What is your level of DIY expertise? Some changes (in order of difficulty not order of cost) if you want to spend more to try to fix it:
Purchase a new power supply, specifically rated for the G3.
Replace the RAM
Replace the NVMe
Replace the fan
Re-seat the heatsink with fresh thermal paste
Personally, I would start with the cheapest options first. For the NVMe, I would test with a small capacity cheap one before replacing it with a large new 2TB one. That and replacing the RAM are the most expensive items.
Of course there is no guarantee for these tweaks to fix the issue, so you’ll have to decide how far you are willing to take it before accepting a loss.
Thank you Paul for the response. My level is everyday is a school day so let’s go and enjoy school . I’ve upgraded the nvme and ram, tried the old hardwear and new in the G3 still same result. Launched with nomodset and intel-something from start9 cmd line. boot from cmd instantly powered down
The 39.4748161 started journey service. Is the point it shuts off trying to install Ubuntu.
New power supply ordered arriving tomorrow failing that I’m onto the next steps. Thanks again for the support and quick reply.
The replacement power supply is definitely the right next test. Since both StartOS and Ubuntu cause the entire unit to power off (and it behaves the same with both the original and replacement RAM/NVMe) the problem is unlikely to be the StartOS image or the storage.
I believe the NucBox G3 expects a regulated 12 V / 3 A supply (with the correct polarity and connector size of course). A weak or incorrect adapter can run the BIOS and boot menu but collapse when Linux enables the CPU, graphics, NVMe, networking, and normal power-management features. That would produce an instant power loss rather than a useful error message.
The Ubuntu line at 39 seconds probably is not the cause (most likely it is just the last message that was printed before power disappeared).
When the new supply arrives I would try the below steps. These are generic troubleshooting steps – applicability to the BIOS on the G3 specifically may vary (I can dig around in the BIOS on one of my G3’s here if you have trouble finding the items mentioned):
Load the BIOS defaults and disable Fast Boot and Secure Boot.
Boot with only the RAM module, the installer USB, keyboard, and monitor connected.
Check the CPU temperature and fan operation in the BIOS for several minutes.
Try Memtest86+ or another diagnostic USB.
Try booting an Ubuntu live environment with the NVMe temporarily removed.
Also note exactly what happens when it fails. If the power LED and fan both go off, this strongly suggests the power adapter, overheating protection, the DC jack, or motherboard power circuitry. If the LED remains on and only the display disappears, it may instead be an Intel graphics or HDMI initialization issue.
If a known-good 12 V / 3 A supply does not change anything, and the CPU temperature is normal, I would begin treating the second-hand unit itself as faulty (most likely the DC input or motherboard power-delivery circuitry) rather than continuing to experiment with Linux kernel parameters.
Hello. Thank you for the video. I’m keen to run a knots node and I now have a mini pc with 16 Gb RAM and a 4-core CPU and a 2 TB drive, which should be enough for a node. I’m not highly technical. I am at a complete loss as to what I should download from Github. I see bitcoin-knots-startos in the repositores, but I can’t find an iso file in the assets. Any advice would be greatly appreciated.
Start9 have re-arranged their repos on Github, and have combined start-cli, start-sdk, start-tunnel, and startos all in the same repo. The latest release page for StartOS 0.4.0 is HERE.
Based on your system, you probably want the x86_64/AMD64 iso file.
StartOS 0.4.0 looks a bit different than 0.3.5.1 which I demo’d in the video, so if you get lost, let me know and I can help you navigate it.
Hi y’all, Im interested in spinning up and running my first Bitcoin node. Im interested in running knots and Bip110. I bought a Lenovo think center m700, Crucial BX500 1TB 3D NAND SATA 2.5-Inch Internal SSD, up to 540MB/s to replace the factory ssd, and a Crucial 8GB, 260-pin SODIMM, DDR4 PC4-19200 ram stick. I recently saw that the additional ram stick may interfere with running my node. Can someone tell me if the ram stick that I bought will have this problem?
Coincidentally, I’ve tested a lot on the M700 and M900 models (I’ve had a video planned for a while on disabling IME and installing coreboot on these).
The main issue I’ve found with these two models is that they can be finicky with the specific model of RAM that you use (even with specs that match what is supported in the documentation). How this manifests is particularly annoying – everything is going perfectly fine, running smooth as butter, and you get some percentage into the IBD. Then, out of the blue, Knots gets into a restart loop, and checking the logs shows that the data is corrupt. I changed a bunch of parts and did different SSDs and power supplies before I figured out that the RAM was the problem.
I think what happens is that errors in memory eventually lead to the blockchain data becoming corrupted. When I switched to RAM I got from eBay that specificaly had Lenovo stickers on it, it fixed the problem. I’ve since confirmed with a number of other RAM sticks with Lenovo stickers, all working, and other RAM sticks that are hit-or-miss – sometimes work but sometimes have the problem.
Anyway, not trying to scare you with the RAM, just something to watch for. If you are in the US and run into this problem, I can probably swap you for some equivalent RAM that I’ve confirmed works on my M700.
Much appreciated Paul, I find myself feeling rushed because I want to take part in running knots and signaling my support for BIP110 before August. I will let you know what I find out ASAP.
As I was looking for Lenovo specific ram sticks, I could hardly find any that were economical. Then I thought I could buy another m700 for parts on eBay, and it seems there are plenty of them missing the OS, hardrive and AC adapter for under $60 USD. But seems to still have the 8gb ram stick inside. Is there any risk of getting a “tired” or busted ram stick from one of these?
It’s possible. Normally, I would say this is low-risk because RAM tends to be pretty durable. But with the ongoing RAM/SSD crisis, who knows what some folks might be doing.
Quick update, luckily on my part it was a power supply issue every is working as instructed after replacing the supply. Thanks again Paul for the tutorial and instant support when needed.
My node is up and running. The only thing I haven’t been able to figure out is how to generate an onion address for tor to verify visibility on the network. Tor is installed. Any advice most welcome.
Yes, go to the Bitcoin Knots service Dashboard. In the “Service Interfaces” section look for the row that has “p2p” listed in the “Type” column, and click on that. You should see a section called “Plugin: Tor”. Click on the “+ Add Onion Service” button and click “Submit”. A Tor address should appear in the list, and after some time peers should start connecting to it.
. I’ve upgraded the nvme and ram, tried the old hardwear and new in the G3 still same result. Launched with nomodset and intel-something from start9 cmd line. boot from cmd instantly powered down 
