Project

General

Profile

Support #7120

High CPU usage on the server (dom0)

Added by Philippe May about 3 years ago. Updated almost 3 years ago.

Status:
In Progress
Priority:
Normal
Assignee:
Start date:
23/11/2018
Due date:
% Done:

0%

Close

Description

System (bare metal, domUs are kinda OK) is slow, etc: top shows ~100% of kworker on the dom0.

History

#1 Updated by Philippe May about 3 years ago

Installed perf and linux-perf-4.18.

Run perf top: acpi issue?

Seen in https://bbs.archlinux.org/viewtopic.php?id=223799, run:

grep . -r /sys/firmware/acpi/interrupts

/sys/firmware/acpi/interrupts/gpe6F has a very high number (of interrupts).

So:

echo "disable" >/sys/firmware/acpi/interrupts/gpe6F

Great, the damn ACPI doesn't flood the CPU with interrupts and load back to normal.

What is gpe6F? Bug in the BIOS firmware (it's a Dell PowerEdge T30/07T4MC, BIOS 1.0.12 12/06/2017)? No idea, but there's an update: https://www.dell.com/support/home/us/en/04/product-support/product/poweredge-t30/drivers (1.0.15).

So:

  • Try firmware update and check if the problem reappears
  • If yes, make the hack resilient over reboots

#2 Updated by Philippe May almost 3 years ago

Seen again last Saturday, 23/2/2018, rebooted the dom0.

Something strange: the running kernel (4.18.0-0.bpo.1-amd64) is marked from unknown source in apt and not updated. Need to install 4.19 from backports instead.

#3 Updated by Philippe May almost 3 years ago

Kernel upgraded to 4.19.0-0.bpo.2-amd64 (including for the domUs, which use direct kernel boot, ie. provided by the dom0)

#4 Updated by Philippe May almost 3 years ago

  • Status changed from New to In Progress

Also available in: Atom PDF