SSD and TRIM

[BlueDragon]

Hi all,

I would like to implement TRIM for performance and lifespan purposes of my SSD. I found this interesting article about how to do it in Linux. What is the proper way to have a script run once a week to fstrim in Mageia 5.1? (NB: I'm very newbie in Linux command line.) Note: I have no encryption or LVM, just btrfs.

Thanks for your help.

[gohlip]

just btrfs.

This also reminds me. Do you get an error message when booting up? Something about "sparse file not found"?
If this is the error message you get, it is 'spurious' error; meaning that's okay and typing any key will continue to boot up.
There are ways to get rid of this error message (which I forgot and I don't use btrfs anymore, except one for my data partition).

[edit] - back to your original point on ssd and trim -
I think it's okay the way Mageia installs to ssd, but I'll let others answer this.

[alf]

can simply be done by enabling the systemd fstrim timer(as described in the linked article)

as root in a console

Code: Select all

systemctl enable fstrim.timer

this will trim your ssd once a week.

[jiml8]

I have two SSDs in my system, sdc and sdd.

Sdc1 contains /, sdc5 contains miscellaneous things including a bunch of virtual machines, and sdd5 contains /home and another bunch of virtual machines. Sdd1 is swap, and I don't trim it.

I have installed several files on my system for the SSDs. This includes a service that changes the I/O scheduler for the SSDs to account for their random access nature, a reduction in system swappiness (I do this because I have lots of RAM and don't want to swap unless it is really necessary), and of course the program that does the trim.

I probably only gain 1 or 2 percent in performance by doing these things, but why not?

To handle setting the I/O scheduler and set the swappiness, I have created a service which is located in /etc/systemd/system and named configSSD.service:

Code: Select all

[Unit]
Description = Configure system for SSD drives

[Service]
ExecStart = /usr/local/bin/configSSD

[Install]
WantedBy = basic.target


This service calls the function /usr/local/bin/configSSD:

Code: Select all

#!/bin/bash
#  configure for SSDs
echo deadline > /sys/block/sdc/queue/scheduler
echo deadline > /sys/block/sdd/queue/scheduler
echo 10 > /proc/sys/vm/swappiness


In /etc/cron/daily, I have placed this script named fstrim (which I did not write; I found it and it works perfectly for me so I use it):

Code: Select all

#!/bin/sh
#
# To find which FS support trim, we check that DISC-MAX (discard max bytes)
# is great than zero. Check discard_max_bytes documentation at
# https://www.kernel.org/doc/Documentation/block/queue-sysfs.txt
#
LOG=/var/log/trim.log
echo "*** $(date -R) ***" >> $LOG
for fs in $(lsblk -o MOUNTPOINT,DISC-MAX,FSTYPE | grep -E '^/.* [1-9]+.* ' | awk '{print $1}'); do
        fstrim -v "$fs" >> $LOG
done


[BlueDragon]

can simply be done by enabling the systemd fstrim timer(as described in the linked article)

as root in a console

Code: Select all

systemctl enable fstrim.timer

this will trim your ssd once a week.

Just did it (simplest way for me now) and the system replied with "created symlink..."

Is there a way to verify that it will be running? Any log or something?

What about swapiness reduction (presently no swap but I will correct that)? Any simple way to tell the kernel to "reasonably" reduce the swapping?

Thanks!

[alf]

If you didn't do a reboot after enabling the timer you can start the timer with

Code: Select all

systemctl start fstrim.timer

(also as root).
To check the timer is running do

Code: Select all

systemctl status fstrim.timer

(need no root rights).

[BlueDragon]

Thanks a lot alf for these advises. Easy and exactly what was needed.

Using the above status command, first the system replied "Loaded: loaded" but "Active: inactive (dead)" After the start command it changed to "Active: active (running) since..." So I guess now it should be executed weekly, though of course it would be interesting to know how to find out when exactly.

For those interested to read more about it, there is some good information here.

[BlueDragon]

About swappiness I found here this very useful code:

To temporarily set the swappiness in Linux, write the desired value (e.g. 10) to /proc/sys/vm/swappiness using the following command, running as root user:

Code: Select all

# Set the swappiness value as root
echo 10 > /proc/sys/vm/swappiness

# Alternatively, run this
sysctl -w vm.swappiness=10

# Verify the change
cat /proc/sys/vm/swappiness
10

# Alternatively, verify the change
sysctl vm.swappiness
vm.swappiness = 10

Permanent changes are made in /etc/sysctl.conf via the following configuration line (inserted, if not present):

Code: Select all

vm.swappiness=10

[doktor5000]

Using the above status command, first the system replied "Loaded: loaded" but "Active: inactive (dead)" After the start command it changed to "Active: active (running) since..." So I guess now it should be executed weekly, though of course it would be interesting to know how to find out when exactly.[/url].

If you execute this command in a konsole, then you see when the trigger is to be be triggered:

Code: Select all

# systemctl status fstrim.timer
● fstrim.timer - Discard unused blocks once a week
   Loaded: loaded (/usr/lib/systemd/system/fstrim.timer; enabled; vendor preset: disabled)
   Active: active (waiting) since Thu 2019-12-12 16:46:36 CET; 4s ago
  Trigger: Mon 2019-12-16 00:00:00 CET; 3 days left
     Docs: man:fstrim

[johnpenguin]

If you didn't do a reboot after enabling the timer you can start the timer with

Code: Select all

systemctl start fstrim.timer

(also as root).

Does this work with encrypted LVM setups? Then it would be far easier than the solution at https://forums.mageia.org/en/viewtopic.php?f=7&t=9732&p=65452#p65451.

Also is it possible to run this on demand? (For example you may wish to run it once per couple of months)

[doktor5000]

Does this work with encrypted LVM setups?

As far as I'm aware, it should, although you need to edit /etc/crypttab manually and add allow-discards option for each encrypted device.
See https://wiki.archlinux.org/index.php/Dm ... rives_(SSD)

And yes, you can also run fstrim manually.