MiniDLNA database to usb key (jffs2)

[Eric Wick]

I have used for testing a 2gb usb key for the minidlna database. With ext3 this works around 4 weeks, then the database get hosed by block problems. I found out that ext2/3 have no kind of wear levelling for the usb key. Now i am testing this with ext2.

Idea: Insert jffs2 into the next release, then we have a wear levelling filesystem for that.

Bye
Eric

[Joao Cardoso]

On Friday, June 8, 2012 8:21:45 PM UTC+1, Eric Wick wrote:

I have used for testing a 2gb usb key for the minidlna database. With ext3 this works around 4 weeks, then the database get hosed by block problems. I found out that ext2/3 have no kind of wear levelling for the usb key. Now i am testing this with ext2.

Idea: Insert jffs2 into the next release, then we have a wear levelling filesystem for that.

I'm afraid that using jffs2  wont solve the problem, as the USB pen will still use its own internal scheme.

That only make sense for internal flash chips without an internal wear leveling scheme.

Even SSD has its own internal mechanism, follow http://en.wikipedia.org/wiki/TRIM

 

Bye
Eric

[Eric Wick]

Okay, i read that some usb keys have an internal wear levelling. Will check this with ext2 now.

For minidlna the external database results to faster dlna data while playing to more devices.

Thanks
Eric

[Joao Cardoso]

On Sunday, June 10, 2012 7:47:17 AM UTC+1, Eric Wick wrote:

Okay, i read that some usb keys have an internal wear levelling. Will check this with ext2 now.

I'm afraid that you will have no luck on this subject.

 

For minidlna the external database results to faster dlna data while playing to more devices.

Yes, reading is faster, the problem is that writing will be slower each time you write, and will eventually degrade the internal flash chip.

USB pens are not designed for such intensive writes, as that is not its typical usage pattern; SSDs on the other hand are designed to replace HDDs, so they have better provisions for dealing with the limited number of write cycles (some manufactures say 1 million write cycles)

Perhaps your USB pen was old anyway...

 

[Eric Wick]

Did alt-f mount the usb stick with 'noatime'? If not where can i change this, at the moment the stick works pretty well.

Thanks
Eric

[Joao Cardoso]

On Tuesday, June 12, 2012 7:07:40 PM UTC+1, Eric Wick wrote:

Did alt-f mount the usb stick with 'noatime'?

The default filesystem settings are used.

You can see then if at the command line you use the command

   cat /proc/mounts

For ext3/4, 'relatime' is used

 

If not where can i change this, at the moment the stick works pretty well.

Under Disk->Filesystem, fill-in the 'Mount options' entry, then FS Operations->'Set mnt Options'

As the online help says:

Set Mnt options sets mount options using the Mount Options entry. The options will be used whenever the filesystem is mounted. 

 

As usual, for the changes to persist after a reboot, you have to 'save settings'

You should probably use the defaults and add at the end your own options.

Try setting 'defaults,noatime' first, then use  'cat /proc/mounts' to see if they are as you wish. If not, use the 4th field from the 'cat /proc/mounts' command ouput and add/replace your own, such as 'rw,noatime,barrier=0,data=ordered'