[Borgbackup] borg prune crashing: AttributeError: 'tuple' object has no attribute 'items'

[Bryan Fields]

On 11/10/20 6:53 AM, Bzzzz wrote:
> If it can help OP ; in my experience, electrical shocks can lead to
> RAM/CPU/MB trashing, especially lightnings when you dwell far from
> town, even with so called surge protectors - worse, trashed RAM, even
> ECC, isn't always uncovered by memtest86+ in this case.

Yea, that would be a surge.  Grounding is very important, you want to have
everything tied to a single point ground or utilize a halo type ground inside
your room.  I'm in St Petersburg, and we are well known for the lightning
storms here.  The other issue is all cables into and out of the room should be
surge protected and grounded the same as well.  This way during a surge, or
strike event your systems all rise and fall in potential together.

Things like bend radius of wire, and loops externally of cable come into play.
 a .01mH inductor cut with several million volts across it couples a nice
surge into things.

That said, I have a .2 Ohm @200 kHz ground here at home, measured with a
ground tester, and my server having issues is in a co-lo facility inside a
concrete building with a steel structure and redundant filtered 208v power off
a HVDC plant.

Some men are Catholics, some are Muslims; my religion is grounding :)

> We even had a customer whose MB+RAM+CPU was trashed all together but
> without any visible sign until it crashed badly after 10'~15' running -
> repair came effective from iterations in changing HW.

Interestingly enough, I went though a major issue with a SCSI controller in a
server here.  Only caught it with ZFS during checks where it was hitting the
serial bus heavily.  I could rebuild/build the array fine, but two disks kept
dying during scrubs.  I swapped the cages, the disks, the cables and then
finally the controllers.  Problem was the controller.  It was missing some
0402 sized caps on the serial lines on the underside of the board.  Unsure if
it was defective or broken during install, but the caps filtered the noise off
the bus, and during busy operations they allowed enough corruption of the
signal that ZFS would fail the disks.


-- 
Bryan Fields

727-409-1194 - Voice
http://bryanfields.net