Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Chrysalis]

I have a second unit now.

The first unit, has no hardware p-state support, and I also discovered wont reliably wake from sleep state (S3).  If I use hibernate, it also wont wake up, but then after a power cycle it will boot the hibernated state.
The second unit has hardware p-states and sleep works fine.

I found this link from intel alongside dozens of sleep complaints.

Although this is for different hardware, I suspect its a bios issue on the first unit.

https://www.intel.com/content/www/us/en/support/articles/000057914/intel-nuc.html

First will be used with pfSense and will use software p-states for power management, as its always on, sleep doesnt matter.  Second will be Windows use to replace my laptop.

[Ronski]

I'm having an issue with OpenVPN

My brother got an error along the lines of "Connection failed - you are using insecure hash algorithm in ca signature. Please regenerate CA with other hash algorithm"

I wasn't experiencing this error, but it turned out I had Legacy turned on in the advanced settings in my client.

So after some Googling I found I needed to change Auth digest algorithm setting to SHA256, as it was SHA1, I changed this, renewed all the certificates, exported the client config and imported on my phone, but it then never connects.

Change back to SHA1 without altering anything else and it connects.

Any ideas what I'm missing please?

PS. All certificates show they are Signature Digest: RSA-SHA256

[Chrysalis]

I dont mind trying to help, but it probably involves enabling more verbose logs, and once we go down that road it might be better dealt with in DM or something.

[Chrysalis]

Been working on my first unit, and discovered it has i/o instability, but I am unable to trigger the problem on demand which is really frustrating, likely either RAM or storage, both of which I took as inclusive (my previous NUC's I ordered barebones and used known brands for storage and RAM).

The unit has a internal mPCIE which is pin compatible with mSATA however installing a mSATA drive confirms its not connected to the onboard SATA controller.  It has a normal SATA port for standard SATA drives and 2 NVME, one for 2230 and another for 2280.

I got no LPDDR5 or NVME lying around, memory tests come up with nothing, extended SMART tests clean on the storage and has clean SMART stats, file system checks however fix corruption on the drive.  The BIOS is the most thread bare BIOS I have ever seen in a NUC, its like a laptop BIOS.

So will chance it and buy a 2230 NVME, as that will work in either NVME slot, so if the NVME slot is duff, it can get round that, or if its the existing NVME it will solve the issue.

Luckily the other NUC I brought bare (after realising the included components were only saving £30).

Also the speaker is soldered directly on to board with no jumper to disable it, there is jumpers to control always on behaviour, pin 1 isnt labelled, so those I will tinker with later as I am curious if will fix the shutdown/sleep issues.

[Alex Atkin UK]

Ironically all my boxes bought inclusive work fine, its the one box I bought bare bones that has problems.

It will just randomly lockup, from several times a day to running solid on Folding@Home for months (makes an nice hand warmer), its bizarre.  I'm reasonably sure the RAM is fine (pretty sure I ran memtest for a day to check), although its hard to tell given these closed boxes are going to make the RAM run hot.

I even re applied the paste which dramatically brought the temps down, but it still happens.

[Chrysalis]

Thought the unit was dead, but got a really quick support reply, flipped battery out and back and its alive again.

[EC300]

Ironically all my boxes bought inclusive work fine, its the one box I bought bare bones that has problems.

It will just randomly lockup, from several times a day to running solid on Folding@Home for months (makes an nice hand warmer), its bizarre.  I'm reasonably sure the RAM is fine (pretty sure I ran memtest for a day to check), although its hard to tell given these closed boxes are going to make the RAM run hot.

I even re applied the paste which dramatically brought the temps down, but it still happens.

Could be static damage during installing which typically results in strange random errors.  Did you wear a grounded wrist strap?

[tickmike]

re.....Could be static damage during installing which typically results in strange random errors.  Did you wear a grounded wrist strap?

All my working years in electronics I never used a wrist strap or did anyone else. 

[Ronski]

Neither have I, I simply touch something that is earthed, or if it has a built in power supply I plug the PC in but switch it off at the wall and the power supply, then the case is nicely earthed.

[Alex Atkin UK]

Yeah its pretty unlikely.  My understanding is grounding is ideal, but any large metal surface will dissipate the charge too.  I'm not a nut case like so many YouTubers, grabbing the RAM sticks by the connection pads (which is bad as skin oils can mess up the connection too).  Its the heatsink or the edges of the PCB.

In drier weather I will get a shock off none-grounded surfaces quite often, but being Yorkshire humidity is almost always 50-60% so this is rare.

[Chrysalis]

I brought these from amazon.

https://www.amazon.co.uk/dp/B0BLGPKZTG

Then one of the thin strips I put down covering two annoying LED's.

I like the soft blue on the power button, but the separate green one I felt was too bring and redundant, and the strip is also covering the red i/o LED below it as well.

[Chrysalis]

I increased the performance on the speed shift settings (lowered from 70 to 60) so that the clocks ramp up fully during high throughput situations.  This then made an increase of temps although only by 2C, so I started experimenting and ended up noticing if the slider is set to a very low amount the temperatures are actually lower, I can only assume this is because its getting more idle time due to processing things faster, so now am running it on 30, although it stayed low even on a value of 10.  The drop isnt small, its running about 12C lower.

[Alex Atkin UK]

According to the dashboard, mine runs hotter at 30 as it regularly boosts to 2.9Ghz and only seems to drop to 2.1Ghz.
At 50 it boosts to 2Ghz and drops to 1.8Ghz.  These are both with my network idle, which is to say not very idle but not under load either.
Pushing Gigabit it boosts to 2.9Ghz under both settings which suggests its doing its job well.  Setting 0 locks it to 2.9Ghz with temps 3-6C higher than 50, depending on load, which is more or less the same temps as 30.

Anything over 50 behaviour changes, it always idles around 1.8Ghz but at 80 under load it drops to 1.5Ghz, 100 it drops to 1.3Ghz.  Presumably as an active CPU pulls more watts than an idle one, at the same clock speed, its throttling the TDP.

The new SoCs in this line should be interesting, if they are too based on the e-cores of the latest CPUs.

Intel are claiming up to 68% boost in IPC.  I wonder if we will start to see 4 ports with 10Gbit?