Testing Out The Btrfs Mount Options On Linux 3.2

[Kamikaze]

@Kamikaze and rrohbeck:

I'm not trying to mock your personal experience, far from it.
But please keep in mind that when it comes to file system corruption, "WORKS-FOR-ME" reports carry *far* less weight than corruption reports - even if only 1/1000 suffers from a catastrophic report. (And by looking at the Fedora bugzilla, btrfs has yet to reach 1/1000 level)

- Gilboa

Yep, fair enough. I wouldn't advocate it for any kind of business use at this stage and frankly I'm surprised that oracle are (possibly?) considering it production ready with the reports out there of unrecoverable errors.

I'm hoping this will change in a year or two though; once the btrfs fsck tools are released and stable, and a major distro is comfortable enabling the FS by default (and the bug reports have been addressed).

BTW - regarding the oracle front, anyone know if they are still considering allowing btrfs to be set as the root partition on their next release considering the btrfs tools still aren't out of their experimental "do not use" state?

[kebabbert]

@Kamikaze and rrohbeck:

I'm not trying to mock your personal experience, far from it.
But please keep in mind that when it comes to file system corruption, "WORKS-FOR-ME" reports carry *far* less weight than corruption reports - even if only 1/1000 suffers from a catastrophic report. (And by looking at the Fedora bugzilla, btrfs has yet to reach 1/1000 level)

- Gilboa

Valid argument.

Also remember that running in production in a critical business is another thing than running BTRFS at home. If you bet money, you want to go as safe as possible and minimize the risk.

[squan]

OK, so let's talk reliability. As it stands, BTRFS is much more prone to corruption and gradual degradation of speed and even available free spance than EXT4. Just look at all the reports of people saying BTRFS becomes unusable after a a few days of running. The fact is that for the vast majority of desktop users, BTRFS still has no advantage over EXT4.

Performance degradation is exactly what I'm experiencing:: Random apps often stall for many seconds while one or more cores are in I/O wait and no paging takes plae.
I even updated the kernel from 3.1 to 3.3 and performance degradation happened again within some days.
Could you give links to bug reports on this (I have difficulties to identify the correct one at bugzilla.kernel.org)?

[rrohbeck]

Performance degradation is exactly what I'm experiencing:: Random apps often stall for many seconds while one or more cores are in I/O wait and no paging takes plae.
I even updated the kernel from 3.1 to 3.3 and performance degradation happened again within some days.
Could you give links to bug reports on this (I have difficulties to identify the correct one at bugzilla.kernel.org)?

This seems to happen when the maximum number of threads it wants to spawn are busy, and it spawn a lot of threads.
Since I run with thread_pool=32 in the mount options I don't see it any more.