[Solved] Read Error on all disks

[lorddrinkal0t]

Dear all,

 

I had a big scare today, my array turned RED and reported ReadErrors an all disks.

Can some point me in the right direction for my investigation?

 

Could it be that my new GPU (3060TI) pulled too much power and reset all the drives at the same time, without crashing the server all together?

I was playing around with madVR around that time....

 

 

tower-diagnostics-20210926-2325.zip

[JorgeB]

Sep 26 22:05:21 Tower kernel: mpt2sas_cm0: SAS host is non-operational !!!!

 

Problem with the HBA, but before this there were already multiple disk issues, could be controller, cables, power issue, etc, start by checking the controller is well seated and sufficiently cooled, you can also try another PCIe slot if available.

[lorddrinkal0t]

Hi,

 

thanks for your feedback.

 

You mean the errors related to the spindown events?

 

No other PCIe Slot available.

After a reboot the Array came back and the files are accessible again, currently a parity check is running.

I will open the case when the check is finished, I think the powersupply is really old anyway.

 

But the timing with the GPU Load and shortly after all the files were not accessible anymore makes me think the power supply is at fault here.

 

If I'm not mistaken I should calculate with a maximum consumption of about 500-550 Watts (Ryzen2700X, RTX3060TI, 8 x Seagate 3.5", 2 x SSD, 2 x NVME, 10GbE, Fans and so on)

So a 750W Powersupply would be suffient and with some luck run in the optimal range regarding efficiency.

 

 

 

 

[JorgeB]

43 minutes ago, lorddrinkal0t said:

You mean the errors related to the spindown events?

Yes, didn't notice they were caused by spin ups.

[lorddrinkal0t]

Hi,

after changing the powersupply, no more problems.

So problem solved!