Ítalo Vieira c1075111c2 Disable disks spindown
With some disks of model HFS960G32FEH-BA1 running in a controller
Fusion-MPT 12GSAS/PCIe Secure SAS38xx (HBA355i Front),
notably after a swact, the system goes to a degraded state with errors
in sm services and in systemd-udev daemon.

The reason is that these disks were being spun down, causing latency
when spinning them up again, along with generalized errors across the
system.

The solution is to disable the disks spindown using the TuneD starlingx
profile.
This solution provides a temporary fix and may reduce the MTBF in
spinning hard disk drives (HDDs). However, HDDs are rarely used in
servers nowadays.

Test Plan:
    PASS: 10x swact, without Rook Ceph
    PASS: 5x swact, with Rook Ceph, 4 OSDs (HFS960G32FEH-BA) in each
          host, writing data intensively
    PASS: 10x swact, with Rook Ceph, 11 OSDs (HFS960G32FEH-BA) in each
          host, writing data intensively

Closes-Bug: 2098400
Change-Id: I1a5f52dd5f30976adaf4ea09347f97c9179c260f
Signed-off-by: Ítalo Vieira <italo.gomesvieira@windriver.com>
2025-02-13 17:29:03 -03:00

42 lines
897 B
Plaintext

#
# tuned configuration for StarlingX
#
# Local changes to this profile are not guaranteed to
# be preserved across upgrades. If you need to change
# these settings then either use another existing
# profile that has what you need or create your own.
# Also be mindful that puppet may change sysctl variables
# during host unlock.
[main]
summary=StarlingX tuned profile, mostly favors performance over power
[cpu]
governor=performance
energy_perf_bias=normal
[audio]
timeout=10
[video]
radeon_powersave=auto
[disk]
# Comma separated list of devices, all devices if commented out.
# Eg.: sda,sdb
# devices=
# List of devices using udev regex. All devices if commented out.
# Eg.: (ID_PATH=pci-0000:00:17.0-ata-1.0)|(ID_PATH=pci-0000:00:17.0-ata-2.0)
#devices_udev_regex=
# Disable Disk Monitor
dynamic=0
# Disable APM
apm=255
# Disable spindown
spindown=0
[scsi_host]
alpm=max_performance