On 03/27 09:23, Michael wrote:
> On Friday, 27 March 2020 08:46:33 GMT Peter Humphrey wrote:
> > On Friday, 27 March 2020 05:34:58 GMT [email protected] wrote:
> > > hopefully in the next daus my first SSD drive will arrive
> > > (corona makes everything more difficult...).
> > > 
> > > To prevent an "installed and works"-experience which ends
> > > a month later in a damaged or over-weared SSD with a drastically
> > > shortened lifetime, I want to ask here for own experiences:
> > > 
> > > - What is the best filesustem to be used with a SSD, which
> > > 
> > >   will used for /root when it comes to prolong life of that
> > >   SSD ?
> > 
> > Ext4. I did try F2FS several years ago, but it caused loss of data. It may
> > be better now, but I wouldn't risk it as there's no need: these days SSDs
> > are quite happy with ext4.
> 
> There were some fs supposedly optimised for SSDs, but I haven't used any of 
> them.  I have used ext4 and btrfs.  They both have worked as expected.  Leave 
> a little empty space when partitioning, for the drive's firmware to perform 
> its wear leveling magic.
> 
> 
> > > - What options are recommended for the according mount command?
> > 
> > Nothing special, just 'defaults,relatime' in my case, but put something like
> > this in root's crontab:
> > 
> >     15 1,13 * * * /sbin/fstrim -a
> > 
> > (I'm sure someone will correct me if that's no longer necessary.)
> 
> Mounting with 'discard' option is an alternative, but only if the drive is 
> written to rarely.  For desktop usage a / partition is better trimmed with a 
> simple cron job, or using a script to do it - like SSDcronTRIM, as mentioned 
> here:
> 
> https://wiki.gentoo.org/wiki/SSD#SSDcronTRIM
> 
> Generally speaking, reducing write operations on the SSD is commensurate with 
> a longer disk life, therefore many operations which perform frequent/
> continuous writes are usually configured to be offloaded to RAM, or a 
> spinning 
> disk.  However, many people leave swap on their SSD, apparently without any 
> noticeable adverse effect.
> 
> Failures are rare these days, but when they occur they are usually 
> catastrophic - so backups are a necessity if you value your data.  When/if 
> smartctl reports a failure it is best to remove all data off the drive 
> immediately, *before* you power it down.  It is likely all data will be gone 
> irretrievably after a reboot.
> 
> In earlier years some SSDs were released with half-baked firmware and 
> developed errors, lost data, etc.  So it was advised to upgrade the firmware 
> as soon as you bought the drive.  I don't know if this still applies today, 
> when the technology is more mature.


Hi all,

very good infos :) Thanks a lot!

Fstrim cares for performance and not for lifetime of the SSD, doesn't
it?

Cheers
Meino




Reply via email to