[Discuss] Gluster startup, small-files performance

[ozbek at gmx.com (F. O. Ozbek)]

On 05/14/2014 10:07 AM, Richard Pieri wrote:
> F. O. Ozbek wrote:
>> We have tested ceph, glusterfs and moosefs and decided to use moosefs.
>
> Be careful with MooseFS. Last I knew it ignores fsync and O_SYNC. I call
> that a deal breaker for anything other than scratch storage. See
> previous commentary about SSDs that ignore fsync/O_SYNC and the
> associated data loss issues.

We have tested moosefs extensively. The commercial version has
redundant metadata servers and redundant chunk servers.
Ignoring fsync is not a problem. We will use in production
for real data. (not scratch.)

Fevzi