| 1 |
On Fri, 4 May 2012 03:37:05 +0200, Alan McKinnon wrote: |
| 2 |
|
| 3 |
> > In my completely uninformed guess... a) tmpfs automatically 'cleans |
| 4 |
> > up' every reboot, making sure old folders aren't sitting around stale |
| 5 |
> > even if something did go wrong, and/or b) it's guaranteed writable for |
| 6 |
> > the service that needs to make those mount points. I could probably |
| 7 |
> > come up with a 'c', but I'd likely have to actually do a bit of |
| 8 |
> > reading on the topic before rising looking even more foolishly un-read |
| 9 |
> > on the topic than I already do! :-P |
| 10 |
> > |
| 11 |
> |
| 12 |
> Here you go, one time c): |
| 13 |
> |
| 14 |
> /run can be guaranteed to exist immediately after / is mounted, which |
| 15 |
> fixes a whole slew of really horrible problems if it isn't. |
| 16 |
|
| 17 |
But it cannot be guaranteed that / is mounted rw at this time, so /run o |
| 18 |
tmpfs makes sense from that perspective. However, it is an illogical place |
| 19 |
to mount removable devices, whereas the function of /media is immediately |
| 20 |
obvious from its name. The link given indicates that systemd was already |
| 21 |
mounting /media as a tmpfs, is it really worth switching to an |
| 22 |
unintuitive location for the mountpoints just to save one tmpfs which |
| 23 |
uses so little resources? |
| 24 |
|
| 25 |
|
| 26 |
-- |
| 27 |
Neil Bothwick |
| 28 |
|
| 29 |
If Bill Gates had a dime for every time a Windows box crashed... |
| 30 |
...Oh, wait a minute, he already does. |
Archives