2.2.4 and how I enabled TRIM

[MetalGeek]

I have a 120gb Samsung 840 EVO SSD and I was able to enable TRIM using ThePOO's instructions without incident.  I'm running pfsense 2.2.6 and I booted off my pfsense 2.2.6 USB installer.  So far no issues.

I will say that my experience with this Samsung drive and it's use in 2 other systems hasn't been stellar which is why I've never purchased another Samsung and won't in the future.  For the record, my other SSDs are:

• 120GB PNY (Sandforce based)

• 240GB Toshiba

• 240GB OCZ ARC100

The PNY is "slow" by today's standards but just works (been in 3 systems) and the Toshiba and post Toshiba buyout OCZ are great.

[2chemlud]

Just to avoid confusion: I'm referring to my experience with 2 Samsung EVO 850 SSDs, MetalGeek has a EVO 840 ;-)

[elementalwindx]

@2chemlud:

Just to avoid confusion: I'm referring to my experience with 2 Samsung EVO 850 SSDs, MetalGeek has a EVO 840 ;-)

Not exactly sure why anyone in here is caring about TRIM. I've been running a pfsense box for at least a few years now on a samsung ssd. I think it's a 60 or 80gb 830 pro. I've had 0 issues with it. I don't even know if trim is on or not but I've made 0 effort to find out or change it.

[2chemlud]

@element

See above my post on the esoteric side of the TRIM discussion. But as a matter of fact I have here 2 Samsung EVO 850 120GB SSDs, for both I could not activate TRIM without corrupting pfSense, while it worked fine the same way on a 30GB Kingston NOW…

[keatre]

I just performed my first install on a new machine running pfsesnse 2.2.6 and TRIM was indeed not enabled.  Followed OP's instructions, and TRIM is now enabled for my Intel 535.

Confirming OP's instructions are still needed and work for SSD users on 2.2.6

[2chemlud]

I must confess I'm a little disturbed now, as I recognized that the problem with TRIM is apparently not related to the Samsung EVO 850, but to the i386 version of pfsense 2.2.6

I have a Kingston 30 GB SSD, which was doing fine with x64 pfsense full 2.2.6, but when installing i386 pfsense on boot it hangs on TRIM activation…


[Guest]

but to the i386 version of pfsense 2.2.6

Earlier or later this version (32Bit) will be not available, so anybody should take the chance to settle over
to the x86_64 version, perhaps we will see that changing is coming earlier as we all could expect or imagine it.

[2chemlud]

To see if the problem has to do with the activation of TRIM on my 64bit BSD, I followed exactly the steps given in the first post of this thread, same result, same hanging on booting.

This whole TRIM thing is a pile of sh*t on i386 BSD…

[kapara]

I wonder if it is a batch of bad Samsung ssd's

I recently built out a new box on supermicro MB.  Enabled trim without error.  I did have to use 64bit, and installed off of usb 2.0 port and disable AHCI.  Legacy IDE only.  No issues so far!

Samsung 850 Pro

[2chemlud]

" I did have to use 64bit,"

64bit worked here too, but not 32 bit–-

[sts]

@ThePOO:

–----  I was pretty frustrated getting TRIM enabled in 2.2.4

------  This is what worked for me to get the trim status set to enabled for my SSD.

[…]

Thanks for this guide, I just successfully enabled TRIM on the 16GB mSATA SSD module of the apu2c4!

[mikovirgoez]

Thank you very much, this very help me.

[xman111]

me too,  did it last night on 64bit pfsense 2.3, and an Intel SSD.

[Pippin]

Thank you to @ThePOO
It worked for me.

[mauroman33]

Thank you for sharing this guide, ThePOO

[nicholfd]

@jimp:

Devices purchased from the pfSense store are loaded with TRIM enabled since we know the hardware and disks are capable (or at least in some cases do not have a negative reaction to it)

This wasn't the case with a new set of HA rack mount SG-4860's just purchased (thread https://forum.pfsense.org/index.php?topic=113803.0).  Has this changed?  Should the HA set of SG-4860's have TRIM enabled?

Thanks,
Frank

[paulsnoop]

I just used this guide to enable TRIM on my new Samsung 850 evo msata drive. Rebooted and everything looks as it should.
Should I be concerned about corruption or would it be immediately obvious if there was a problem, I'm using pfSense x64.
Thanks.

[EricE]

@2chemlud:

This whole TRIM thing is a pile of sh*t on i386 BSD…

TRIM is an ugly hack, period.  Mainly a symptom of trying to graft old storage technology that works one way on top of flash memory that works an entirely different way.  Hopefully someone will come up with a replacement for SATA and develop a native flash memory stack end to end.

Until then I just buy SSDs that are over-provisioned and do their own garbage collection irrespective of TRIM.  Most Sandforce based SSDs work this way - it just takes a bit more digging to find 'em.

[somniture]

@EricE:

TRIM is an ugly hack, period.  Mainly a symptom of trying to graft old storage technology that works one way on top of flash memory that works an entirely different way.  Hopefully someone will come up with a replacement for SATA and develop a native flash memory stack end to end.

It's called NVMe.

[Waqar.UK]

@EricE:

@2chemlud:

This whole TRIM thing is a pile of sh*t on i386 BSD…

TRIM is an ugly hack, period.  Mainly a symptom of trying to graft old storage technology that works one way on top of flash memory that works an entirely different way.  Hopefully someone will come up with a replacement for SATA and develop a native flash memory stack end to end.

Until then I just buy SSDs that are over-provisioned and do their own garbage collection irrespective of TRIM.  Most Sandforce based SSDs work this way - it just takes a bit more digging to find 'em.

With regard to  over-provisioned SSD a good article (at least from Kingston) is
https://www.kingston.com/en/ssd/overprovisioning
Seagate:
http://www.seagate.com/gb/en/tech-insights/ssd-over-provisioning-benefits-master-ti/