On 11/2/22 21:23, gene heskett wrote:
On 11/2/22 21:01, David Christensen wrote:
On 11/2/22 12:07, gene heskett wrote:
On 11/2/22 08:19, gene heskett wrote:
All 5 of the samsung SDD's in THIS machine have now been
-test=long'd, all 5 report
with the -a option that the read test failed as seen below, but a -H
says they are healthy.
Whats next?
I have found that smartctl(8) version 7.2 has trouble decoding some of
the statistics for my Intel SSD 520 Series drives (Debian Stable),
while smartctl(8) version 7.3 does a better job
(FreeBSD-12.3-RELEASE-amd64). So, perhaps you are seeing bad
information from smartctl(8) version 7.2.
Debian backports does not seem to offer smartmontools 7.3.
Debian Testing and Unstable have smartmontools 7.3.
Alternatively, Samsung does make software tools for their storage
products. Unfortunately, I believe the tools are Windows
applications; so, they require a Windows installation.
Have you tried smartmontools 7.3 and/or Samsung tools?
AIUI you have a HBA/RAID card in that computer. I suspect that you
also have USB devices and/or other connected hardware. What happens
if you disconnect everything? If the problem goes away, do a search
and find the problem hardware.
Not a raid card, just another $20 6 port sata card, its software raid.
There is no drive activity
of any kind when it hangs. The only indication of life at all is a row
of color changing leds on
the front edge of the mobo are cycling the colors, possibly a little
slower that they are right now.
I assume you understand that expansion cards can have extension firmware
that runs during POST, which can delay or halt POST (?). Similarly, the
motherboard firmware goes through a USB enumeration process during POST
and connected USB devices must have compatible firmware; issues can
cause delays and/or halt POST (?).
Does the boot process change when you disconnect all of the drives?
When your remove the card? If so, use a process of elimination to find
the problem item(s).
I've considered that disconnect as my usb tree looks like a 50 yo
weeping willow.
Everything but the keyboard and mouse buttons can be disconnected by
about 6 usb cables.
Getting these 88 yo knees down on the floor to see about plugging them
back in is a
bit of a chore so they don't get disconnected very often.
That is why they invented knee pads and/or (great?) (grand?) children.
Does the boot process change when you disconnect all USB devices and
connect the keyboard directly to the motherboard? Different ports?
Different keyboards? If boot changes, use a process of elimination to
identify the problem keyboard(s), port(s), and/or USB devices(s).
This is the main computer of a 7 machine home network that by the time
they are done arguing
about who won next Teusday's election, might have 4 more bannana pi's to
connect to.
I'd like to have bought rpi4b's w/8G's but their scarcity has ran the
price above $300.
Much newer bananna pi's BPI-M5's with 4G are $90. And have faster usb as
all 4 ports are usb3.1.
Do like I did with the rpi4b running my biggest cnc lathe, and move all
the high traffic stuff
to a $30 SSD and get uptimes in years. I have an automatic 20kw standby
and the pi has a small
ups. That made the pi into a workstation, building its own real-time
kernel AND linuxcnc from a git pull if
the buildbot at linuxcnc.org gets a tummy ache.
All the more reason to figure out why the server is misbehaving and fix it.
David
