[OpenIndiana-discuss] How is this possible?

[Chris]

On 2021-04-16 11:38, Reginald Beardsley via openindiana-discuss wrote:
> I have a pair of almost identical Z400s. (In total I have 4.) One has 4x 2 
> GB
> DIMMs (#3)  and the other has 2x 8 GB DIMMs (#4) .  Both have Quadro FX 1800 
> cards
> and trayless SATA bays.  Both are connected to the monitor through an 8 port 
> KVM
> switch.
> 
> Hipster 2020.10 was installed on system #3.  The monitor was recognized and
> configured properly as 1600x1200 using the 340.108 driver.   If I move the 
> disk to
> #4 the nVIDIA driver didn't recognize the monitor or the resolution, set 
> 1024x768
> and would not allow me to set it to 1600x1200.
> 
> I  moved the KVM switch port for #4  to #3 and rebooted 2020.10 on #3.  It 
> behaves
> as expected, recognizes the Samsung Monitor and the correct resolution.
> 
> After I installed the 2x 8 GB DIMMs S10 u8 kernel panicked on a null pointer
> dereference, so I moved the 2020.10 disk to system #4 and ran it for several 
> days
> to see if fmd would log ECC errors on the DIMMs.  fmdump -eV reports the log 
> file
> is empty after 3 days.  I don't know what whether fmd in u8 logs ECC errors 
> as
> I've never seen any, though I have seen occasional ECC errors on system #2 
> which
> is running 2017.10
> 
> This morning  I put the S10 u8 disks back in the system, booted and started 
> scrubs
> at which point it immediately kernel panicked.  I rebooted in single user 
> mode.
> Both the root pool 3 way mirror and the RAIDZ1 export pool scrubs completed 
> with
> no errors.
> 
> The 3 disks forming the 2 pools were in a 6 DIMM slot Z400 (#1).  That MB 
> appears
> to have a failing SATA controller as it logged so many sector R/W errors it
> rebooted.  I replaced the suspect disk with a new drive of the same make, 
> model
> and vintage.  The errors continued so I switched the  u8 disks to #4.  A
> subsequent test of the drive that was reported bad showed no errors after a 
> 4 hr
> scan using the BIOS test.
> 
> After verifying that 2020.10 worked properly in #3, I moved the disk back to 
> #4.
> This time it came up at 960x540!  I changed out the video with the card from 
> #1
> and again it came up in 960x540. I then moved the video card from #3 to #4.
> Again, 960x540.  I reset the BIOS to factory defaults, changed the things 
> (e.g
> AHCI) I knew had to be changed.  Again.  Same behavior.
> 
> If I boot the 2020.10 Live Image on #3 it's 1600x1200.  On #4 it's 1024x768. 
>  I
> tried the video card in the other PCIe slot.  Same result.
> 
> I'm completely baffled.    I've never seen anything like this.
Cables? IOW if it is working on one but not others. I'd hazard a guess that 
the
system isn't receiving the EDID from the monitor through the KVM. Some KVMs
can be a problem that way by not providing enough of the power to the 
connections.
Other times (often) (cheap) cables are missing the "sync" wire. In any case,
I'd first start by swapping the cable from the working machine to the port
on the "different" machine and see if that doesn't solve it -- it's the 
fastest
and easiest first step. OTOH If the machine that OI was setup on provides 
video
on a different port than the machine your swapping to. That too may be a 
factor.
In any case; w/o any log output. This is all I feel safe offering. ;-)

HTH

--Chris

> 
> Reg
> 
> _______________________________________________
> openindiana-discuss mailing list
> openindiana-discuss at openindiana.org
> https://openindiana.org/mailman/listinfo/openindiana-discuss

--