Atime and btrfs: a bad combination?

Posted Jun 1, 2012 6:05 UTC (Fri) by nestal (subscriber, #66970)
Parent article: Atime and btrfs: a bad combination?

Can the distros just disable atime by default, and enable it once the user install apps like mutt? The distros should know whether a user want atime.

If you "really" want atime, 2.2GB of space is like a small price to pay :P

Atime and btrfs: a bad combination?

Posted Jun 1, 2012 7:20 UTC (Fri) by ptman (subscriber, #57271) [Link] (2 responses)

That was just for one invocation of grep. Think about losing that much each time you execute grep (or per day that you do that, if using relatime).

Atime and btrfs: a bad combination?

Posted Jun 1, 2012 9:12 UTC (Fri) by Otus (subscriber, #67685) [Link]

> That was just for one invocation of grep. Think about losing that much
> each time you execute grep (or per day that you do that, if using
> relatime).

That could only happen if you did daily snapshots as well, right?

Otherwise the COW source data should be freed as unreferenced.

Atime and btrfs: a bad combination?

Posted Jun 1, 2012 10:30 UTC (Fri) by cwillu (guest, #67268) [Link]

It's once per snapshot, not once per grep.

Atime and btrfs: a bad combination?

Posted Jun 7, 2012 14:16 UTC (Thu) by slashdot (guest, #22014) [Link]

You can also just fix Mutt to use an extended attribute.

Using atime is broken anyway since the fact that a file was accessed doesn't mean that the user read the mail (e.g. it could be the user grepping the mailbox).